ABSTRACT
Background
Suboptimal vitamin D (VitD) status has been associated with poor bone health and other adverse health outcomes and is common among children. Various early-life factors are associated with child VitD, yet few studies have examined multiple factors simultaneously in a single study population.
Objectives
We aimed to characterize relations of early-life factors with plasma 25-hydroxyvitamin D [25(OH)D] concentrations in early and mid-childhood, and to explore potential differences in these associations between white and black children.
Methods
We identified associations of various early-life factors with 25(OH)D concentrations in early and mid-childhood among 961 children in Project Viva using linear regression models. All variables associated with 25(OH)D were included together in final multivariable models at each outcome time point: 1 in the overall sample and additional models for children whose mothers identified them as being white or black.
Results
Overall mean ± SD 25(OH)D concentrations were 86 ± 29 nmol/L in early childhood and 68 ± 21 nmol/L in mid-childhood. After accounting for other predictors, children who took VitD supplements (compared with those who did not) had 25(OH)D concentrations 5.6 nmol/L (95% CI: 2.0, 9.2 nmol/L) higher in early childhood and 8.2 nmol/L (95% CI: 4.8, 11.6 nmol/L) higher in mid-childhood. Other factors consistently associated with higher 25(OH)D were blood collection in summer or fall, white race, nonfall birth season, prenatal exposure to higher 25(OH)D, and higher dietary intake of VitD. Greater waist circumference was associated with lower 25(OH)D in early childhood (β: −3.8; 95% CI: −7.4, −0.2 per 1-SD increase) among black children only.
Conclusions
Our findings may help clinicians better target children at risk of lower 25(OH)D for screening and/or intervention and may inform research focused on associations of 25(OH)D with different exposures and outcomes or causal effects of early-life factors on later VitD status.
This trial was registered at clinicaltrials.gov as NCT02820402.
Keywords: predictors of vitamin D status, 25(OH)D, pediatric cohort, Project Viva, early-life factors
Introduction
Vitamin D (VitD) status is associated with various markers of concurrent health in both children (1–3) and adults (4) and may contribute to programming future bone health, adiposity, and cardiometabolic risk (5, 6). Whereas severe deficiency in children is known to cause growth problems such as rickets (7), less severe but suboptimal VitD status may be associated with bone mineral depletion and other adverse health outcomes (4, 8). Although the importance of adequate VitD status in childhood is well-established, our understanding of factors associated with VitD status in early life is incomplete.
VitD can be obtained through the diet or synthesized by the body after exposure to UVB light (9). VitD status is commonly ascertained via blood concentrations of 25-hydroxyvitamin D [25(OH)D], which reflect both VitD intake and synthesis (10). In higher-latitude areas where ambient sunlight is limited during parts of the year, such as the northeastern United States, residents may be at risk of VitD deficiency or insufficiency, and low VitD status is common in pediatric populations (9, 11). In addition, 25(OH)D synthesis in response to UVB light exposure may be limited in individuals with high skin pigmentation (10). Several factors predict lower VitD status among children, including sampling during seasons with less sunlight (12–15), darker skin color (16), female sex (3, 12, 16–19), older age (3, 13, 17, 19), exposure to tobacco smoke (19), higher adiposity (16, 19, 20), lower consumption of dietary sources of VitD (13), not using VitD supplements (13, 16), indicators of less sun exposure (13, 14, 20), and lower maternal education (16). However, associations are inconsistent across populations, and few studies have examined associations of these factors with 25(OH)D at particular time points during childhood. In addition, co-occurrence of these factors has made it difficult to identify independent predictors in order to generate hypotheses about biological mechanisms affecting VitD status.
The objective of this study was to examine associations of child, maternal, and household characteristics in early life with plasma concentrations of 25(OH)D in early (median: 3.2 y), and mid- (median: 7.7 y) childhood. We aimed to better characterize these relations to identify potential confounders of associations of biological factors and vitamin D status, as well as associations of vitamin D status and health outcomes, for future research. In addition, we were interested in identifying independent predictors of childhood 25(OH)D concentrations to inform the design of randomized controlled trials aimed at improving health outcomes related to VitD or observational studies emulating such target trials. Finally, we explored the extent to which the early-life variables associated with childhood 25(OH)D concentrations were similar or different between subgroups of children identified as either white or black, using mother-reported race as a proxy for skin pigmentation. We included 961 children from the well-characterized Project Viva cohort based in eastern Massachusetts.
Methods
Participants
Project Viva recruited pregnant women during 1999–2002 at their initial obstetric appointment at 1 of 8 offices of a large multispecialty group practice. Detailed recruitment and retention procedures have been described previously (21) and participant flow is outlined in Supplemental Figure 1. Of the 2128 liveborn singleton infants in the Project Viva cohort, we included 961 children in this analysis: 306 who had 25(OH)D measurements from early childhood only, 216 who had 25(OH)D measurements from mid-childhood only, and 439 who had measurements from both visits. This approach provided total sample sizes of 745 for analysis of 25(OH)D in early childhood and 655 for analysis of 25(OH)D in mid-childhood.
Compared with children not included (n = 1167), the 961 included were more likely to be fully breastfed at 6 mo (27% compared with 23%). Included and excluded children were similar in terms of sex, birth season, race/ethnicity, proportions of college-educated mothers, household income, maternal BMI, and maternal prenatal and cord plasma 25(OH)D concentrations (Supplemental Table 1). Institutional Review Boards of participating institutions approved the study protocols, and mothers provided written informed consent for their own and their child's participation.
Potential early-life factors associated with childhood 25(OH)D
Based on scientific knowledge and the available literature, we identified variables potentially associated with childhood 25(OH)D concentrations. Supplemental Table 2 provides details on variables that were recategorized for this analysis. We measured the mother's plasma 25(OH)D during the second trimester of pregnancy and 25(OH)D concentrations in umbilical cord plasma collected at delivery. Mothers reported their household income, highest education level, and prepregnancy weight at the initial prenatal visit, and we calculated prepregnancy BMI in kg/m2 using self-reported weight and height, which has been previously validated in this cohort (22). We obtained child sex from delivery records. Mothers reported their child's race/ethnicity at the early-childhood visit (Supplemental Table 2). References to “white” or “black” children are based on race categories assigned according to maternal report of the best way to describe her child's race or ethnicity. We used the mother's self-reported race/ethnicity if the child's was not available (n = 90). In early childhood, mothers reported their child's diet via an FFQ validated for use in preschool-age children (23; https://www.hms.harvard.edu/viva/Data-collection-forms/three-year-question-tyq7-sec-g.pdf). The FFQ assessed the child's diet during the past month, and we estimated usual intake of VitD and calcium from foods using the Harvard nutrient composition database, which includes food composition values from the USDA (24) and is supplemented by other sources. We adjusted individual nutrient estimates for total energy intake using the nutrient residual method (25, 26). In mid-childhood, mothers reported their child's diet during the past month using the PrimeScreen, a brief dietary screener validated in 160 adults against a full FFQ and plasma biomarkers (27). We cannot estimate intakes of individual nutrients using this screener, but we did obtain information on intakes of fish, milk, and other dairy products, which are major dietary sources of VitD. Mothers (or rarely another parent/guardian) reported their child's exposure to cigarette smoke, use of VitD supplements, and amount of active play in early childhood, and exposure to cigarette smoke, use of VitD supplements, time spent outdoors, and sun protection in mid-childhood via questionnaire (Supplemental Table 2).
Trained research assistants measured waist circumference, standing height to the nearest 0.1 cm using a stadiometer (Shorr Board®, Weight and Measure LLC), and subscapular (SS) and triceps (TR) skinfold thicknesses to the nearest 0.1 cm (Holtain calipers, Holtain Ltd) at both visits; weight to the nearest 0.1 kg (Seca model 881, Seca Corp) in early childhood; and weight and percentage body fat via bioimpedance (Tanita model TBF-300A, Tanita Corporation of America) in mid-childhood. We calculated BMI as kg/m2 and determined age- and sex-specific z scores for BMI using US national reference data (28) and calculated the sum of the 2 skinfolds (SS+TR) as a measure of overall adiposity (29).
All data collection instruments used in Project Viva are publicly available from: https://www.hms.harvard.edu/viva/.
Measurement of 25(OH)D
Phlebotomists trained in pediatric blood collection collected samples at the early and mid-childhood visits. Whole-blood samples were refrigerated for <24 h, centrifuged at 2000 rev/min at 4°C for 10 min, and divided into aliquots for storage in liquid nitrogen at −80°C (20). The assay used for early-childhood samples measured total 25(OH)D (25-hydroxyergocalciferol + 25-hydroxycholecalciferol) using an ELISA (Immunodiagnostic Systems Inc.), standardized using internal standards traceable to the isotope dilution LC–tandem MS 25 hydroxyvitamin D Reference Measurement Procedure (30). The assay is sensitive down to 5.0 nmol/L, and day-to-day variabilities at concentrations of 40.3, 72.0, and 132.0 nmol/L are 4.6%, 6.4%, and 8.7%, respectively. In mid-childhood, Project Viva measured total 25(OH)D by isotope dilution LC–tandem MS. The method was optimized in the laboratory based on published procedures (31). The limit of detection is 5 nmol/L for ergocalciferol (D2) and 7.5 nmol/L for cholecalciferol (D3). The between-run CV for a quality control serum containing a total 25(OH)D concentration of 57 nmol/L is 7.5%. We defined categories of VitD status according to cutoffs recommended by the Institute of Medicine: <30 nmol/L (risk of deficiency), 30–49 nmol/L (risk of inadequacy), and ≥50 nmol/L (sufficiency) (10). To facilitate comparison with other studies, we also calculated the proportions of participants categorized according to the Endocrine Society guidelines: <50 nmol/L (deficiency), 50 to <75 nmol/L (insufficiency), and ≥75 nmol/L (sufficiency) (32).
Statistical analysis
We were interested in identifying factors associated with 25(OH)D concentrations in early and mid-childhood, both in isolation and with other predictors included in the model. Because the impact of certain variables on 25(OH)D may depend on baseline concentrations and black children tend to have much lower 25(OH)D concentrations and higher risk of deficiency than white children (33, 34), we were interested in examining potential differences in associations between white and black children. Although we defined these categories as accurately as possible based on our available data, we acknowledge that there is considerable variability in individual characteristics, including skin pigmentation, within each subgroup.
Using linear regression models, we examined associations of each factor with early- and mid-childhood plasma 25(OH)D concentrations. We identified all factors associated with plasma 25(OH)D concentrations (P < 0.05) in either the full sample or ≥1 of the subgroups and included these in final multivariable models (overall and in white and black children separately) at each outcome time point. Children identified as belonging to another racial/ethnic category were included in the analyses of the overall sample, but we did not perform a separate analysis in this subgroup owing to heterogeneity and small sample sizes for each racial/ethnic group included in this category. We also included child sex in the final models because it has been identified as a predictor of VitD status in other studies (12, 16–19, 35), although it did not meet the criteria for inclusion in our sample. We did not include milk or calcium intake in the final models for early-childhood 25(OH)D owing to the likelihood of collinearity with dietary vitamin D. Milk intake was strongly associated with total dietary VitD intake, with mean VitD intakes of 102 IU/d among children with low milk intake, 164 IU/d among children with moderate intake, and 274 IU/d among children with high milk intake; and dietary intakes of calcium and VitD were strongly correlated (Spearman r = 0.9). Similarly, we included waist circumference as a measure of central adiposity but did not include other measures of adiposity (BMI z score or sum of skinfolds). Although multiple measures of adiposity (waist circumference, sum of skinfolds, and percentage body fat) were associated with 25(OH)D among black children in mid-childhood, we included only 1 of these in the final models owing to high collinearity and selected waist circumference for consistency with the early-childhood models. We also omitted milk intake because it is included in total dairy intake.
We used multiple imputation to account for missing data on variables potentially associated with 25(OH)D concentrations. We generated 50 imputed data sets using chained imputation (36) and combined estimates using Rubin's rules (37). We present results from the imputed analysis throughout the article unless otherwise indicated. All 2128 live births were used to generate the imputed data set, but our analyses included only the children who had 25(OH)D concentrations measured at the specified time point (n = 745 in early childhood and n = 655 in mid-childhood). Supplemental Table 3 provides the proportions of each variable missing (and thus imputed) for each analysis.
We performed all analyses using SAS Studio Release 3.7 (SAS Institute Inc.) and considered 2-tailed P values < 0.05 to be statistically significant. We report 95% CIs to indicate the precision of each estimate as well as statistical significance.
Results
Table 1 presents characteristics of the sample. Approximately 63% of the children were identified by their mothers as white, 18% as black, and 19% as other races/ethnicities. Mean maternal 25(OH)D concentrations and cord plasma 25(OH)D concentrations were both higher among white children than black children or those of other races/ethnicities, and lower among black children than children who were white or of other race/ethnicity. Most children were exposed to little or no cigarette smoke, although the proportions exposed were lower among white children than black children. In early childhood, nearly 60% of mothers reported that their child had taken a VitD-containing supplement in the past month, and 50% reported VitD supplement use in mid-childhood.
TABLE 1.
Characteristics of 961 Project Viva children, overall and stratified by race/ethnicity1
| All | White | Black | Other race/ethnicity | |
|---|---|---|---|---|
| Participants, n | 961 | 603 | 171 | 187 |
| Child characteristics in infancy (median: 6.3 mo) | ||||
| Race/ethnicity | ||||
| White | 603 (62.7) | — | — | — |
| Black | 171 (17.8) | — | — | — |
| Other race/ethnicity | 187 (19.5) | — | — | — |
| Sex | ||||
| Male | 490 (51.0) | 298 (49.4) | 89 (52.2) | 103 (55.0) |
| Female | 471 (49.0) | 305 (50.6) | 82 (47.8) | 84 (45.0) |
| Season of birth | ||||
| Fall | 216 (22.5) | 128 (21.2) | 44 (25.6) | 44 (23.6) |
| Summer | 244 (25.4) | 160 (26.5) | 44 (25.8) | 40 (21.4) |
| Winter | 250 (26.0) | 151 (25.1) | 43 (25.2) | 56 (29.9) |
| Spring | 251 (26.1) | 164 (27.2) | 40 (23.5) | 47 (25.1) |
| Cord plasma 25(OH)D, nmol/L | 45.3 ± 22.1 | 51.3 ± 19.8 | 32.1 ± 21.0 | 38.2 ± 20.8 |
| Breastfeeding status at 6 mo | ||||
| Fully or partially formula-fed | 698 (72.7) | 407 (67.5) | 142 (83.1) | 150 (79.9) |
| Fully breastfed | 263 (27.3) | 196 (32.5) | 29 (16.9) | 38 (20.1) |
| Child characteristics at early-childhood visit (median: 3.2 y) | ||||
| Season of blood sampling | ||||
| Fall | 268 (27.9) | 177 (29.3) | 44 (25.6) | 47 (25.3) |
| Winter | 172 (17.9) | 109 (18.1) | 34 (20.2) | 28 (15.1) |
| Spring | 272 (28.3) | 178 (29.5) | 43 (25.1) | 51 (27.4) |
| Summer | 250 (26.0) | 139 (23.1) | 50 (29.2) | 60 (32.3) |
| Age, mo | 39.5 ± 5.2 | 38.9 ± 3.6 | 41.6 ± 8.7 | 39.4 ± 4.8 |
| Cigarette smoke exposure | ||||
| None or <1 h/wk | 880 (91.6) | 567 (94.1) | 146 (85.6) | 167 (89.1) |
| ≥1 h/wk | 81 (8.4) | 36 (5.9) | 25 (14.4) | 20 (10.9) |
| BMI z score | 0.7 ± 1.0 | 0.7 ± 0.9 | 0.9 ± 1.2 | 0.7 ± 1.1 |
| Waist circumference, cm | 51.4 ± 3.8 | 51.5 ± 3.5 | 51.7 ± 4.9 | 50.8 ± 3.9 |
| Sum of skinfolds, mm | 16.6 ± 4.5 | 16.8 ± 4.1 | 16.9 ± 5.8 | 15.9 ± 4.5 |
| Dietary vitamin D intake,2 IU/d | ||||
| Tertile 1 (4–167) | 342 (35.6) | 184 (30.5) | 88 (51.5) | 70 (37.5) |
| Tertile 2 (168–257) | 310 (32.2) | 186 (30.9) | 48 (28.0) | 76 (40.4) |
| Tertile 3 (258–634) | 310 (32.2) | 233 (38.6) | 35 (20.5) | 41 (22.1) |
| Dietary calcium intake,2 mg/d | 869 ± 275 | 917 ± 255 | 776 ± 282 | 797 ± 262 |
| Milk intake | ||||
| Low (0–4 times/wk) | 163 (16.9) | 77 (12.8) | 50 (29.4) | 35 (18.8) |
| Moderate (nearly daily/daily) | 261 (27.2) | 153 (25.4) | 57 (33.3) | 51 (27.3) |
| High (≥2 times/d) | 537 (55.9) | 372 (61.8) | 64 (37.3) | 101 (53.9) |
| Vitamin D supplement use | ||||
| No | 410 (42.7) | 242 (40.1) | 77 (44.9) | 92 (49.1) |
| Yes | 551 (57.3) | 361 (59.9) | 94 (55.1) | 95 (50.9) |
| Active play, h/d | ||||
| <1 | 40 (4.2) | 16 (2.6) | 14 (8.3) | 10 (5.6) |
| 1–3 | 651 (67.7) | 428 (71.1) | 98 (57.4) | 125 (66.5) |
| >3 | 270 (28.1) | 159 (26.3) | 59 (34.3) | 52 (27.9) |
| Plasma 25(OH)D concentration, nmol/L | 86.2 ± 28.9 | 92.6 ± 27.8 | 73.1 ± 27.1 | 77.7 ± 25.5 |
| Child characteristics at mid-childhood visit (median: 7.7 y) | ||||
| Season of blood sampling | ||||
| Summer | 297 (30.9) | 162 (26.9) | 56 (32.7) | 79 (42.0) |
| Fall | 197 (20.5) | 122 (20.3) | 36 (21.1) | 38 (20.5) |
| Winter | 232 (24.1) | 168 (27.9) | 31 (18.2) | 33 (17.4) |
| Spring | 236 (24.5) | 150 (25.0) | 48 (28.0) | 38 (20.1) |
| Age, mo | 95.4 ± 10.8 | 94.4 ± 9.9 | 98.5 ± 12.2 | 96.1 ± 11.6 |
| Cigarette smoke exposure, h/wk | ||||
| None or <1 | 922 (95.9) | 591 (98.1) | 154 (90.0) | 177 (94.3) |
| ≥1 | 39 (4.1) | 12 (1.9) | 17 (10.0) | 11 (5.7) |
| BMI z score | 0.6 ± 1.2 | 0.5 ± 1.1 | 1.0 ± 1.4 | 0.6 ± 1.3 |
| Waist circumference, cm | 60.1 ± 8.9 | 59.3 ± 7.7 | 63.3 ± 11.9 | 60.1 ± 9.1 |
| Sum of skinfolds, mm | 20.0 ± 10.5 | 18.8 ± 9.0 | 23.9 ± 14.1 | 20.3 ± 10.3 |
| Body fat, % | 19.4 ± 7.6 | 18.4 ± 6.8 | 22.3 ± 9.4 | 19.6 ± 7.7 |
| Fish intake | ||||
| Never | 257 (26.8) | 183 (30.4) | 27 (15.7) | 47 (25.1) |
| <1 to 1 time/wk | 547 (56.9) | 350 (58.1) | 90 (52.7) | 107 (57.1) |
| ≥2 times/wk | 157 (16.3) | 69 (11.5) | 54 (31.7) | 33 (17.8) |
| Dairy intake, servings/d | ||||
| <1 | 163 (17.0) | 70 (11.6) | 59 (34.5) | 34 (18.3) |
| 1–2 | 450 (46.8) | 274 (45.5) | 82 (48.2) | 93 (49.8) |
| >2 | 348 (36.2) | 259 (42.9) | 30 (17.4) | 60 (31.9) |
| Vitamin D supplement use | ||||
| Never | 481 (50.0) | 294 (48.8) | 90 (52.5) | 97 (51.7) |
| 1–5 times/wk | 292 (30.4) | 183 (30.4) | 56 (32.5) | 53 (28.5) |
| ≥6 times/wk | 188 (19.5) | 125 (20.8) | 26 (15.0) | 37 (19.7) |
| Outdoor activity (May–October) | ||||
| Low | 8 (0.8) | 1 (0.2) | 5 (3.1) | 1 (0.7) |
| Moderate | 150 (15.6) | 75 (12.4) | 34 (19.8) | 41 (22.0) |
| High | 803 (83.5) | 526 (87.3) | 132 (77.1) | 145 (77.3) |
| Outdoor activity (November–April) | ||||
| Low | 282 (29.4) | 162 (26.9) | 63 (36.7) | 57 (30.6) |
| Moderate | 543 (56.5) | 346 (57.4) | 84 (49.1) | 113 (60.4) |
| High | 136 (14.2) | 95 (15.8) | 24 (14.2) | 17 (9.0) |
| Sunscreen use/covering skin | ||||
| Never/rarely/occasionally | 169 (17.6) | 44 (7.2) | 72 (42.0) | 54 (28.7) |
| Most of the time/always | 792 (82.4) | 559 (92.8) | 99 (58.0) | 134 (71.3) |
| Plasma 25(OH)D concentration, nmol/L | 67.9 ± 21.1 | 72.6 ± 21.8 | 57.4 ± 19.8 | 62.3 ± 19.8 |
| Maternal or household characteristics | ||||
| Annual household income at enrollment, $ | ||||
| ≥70,000 | 576 (59.9) | 446 (73.9) | 46 (26.6) | 85 (45.1) |
| <70,000 | 385 (40.1) | 157 (26.1) | 125 (73.4) | 103 (54.9) |
| Mother's educational attainment | ||||
| College degree or higher | 630 (65.6) | 478 (79.2) | 65 (37.9) | 88 (47.0) |
| Less than a college degree | 331 (34.4) | 125 (20.8) | 106 (62.1) | 99 (53.0) |
| Mother's prepregnancy BMI, kg/m2 | ||||
| <25 | 596 (62.0) | 405 (67.2) | 78 (45.6) | 113 (60.1) |
| ≥25 | 365 (38.0) | 198 (32.8) | 93 (54.4) | 75 (39.9) |
| Mother's second-trimester (median: 28.1 weeks of gestation) plasma 25(OH)D, nmol/L | 57.8 ± 22.7 | 62.8 ± 21.8 | 46.3 ± 23.5 | 52.2 ± 22.4 |
Values are n (column %) for categorical variables or mean ± SD for continuous variables unless otherwise indicated. 25(OH)D, 25-hydroxyvitamin D.
Adjusted for total energy intake.
Among the 745 children with plasma 25(OH)D measured in early childhood, there were substantial differences in 25(OH)D concentration and VitD status between white and black children (Table 1). Overall, only 5% of the children were at risk of VitD deficiency or inadequacy based on 25(OH)D concentrations <50 nmol/L, but the proportion was much higher among black than among white children (Supplemental Table 4). Among the 655 children with plasma 25(OH)D measured in mid-childhood, there were persistent differences between white and black children (Table 1). In mid-childhood, 18% of children overall were at risk of VitD deficiency or inadequacy [25(OH)D <50 nmol/L], and this proportion remained substantially higher among black children than white children (Supplemental Table 5). These findings continue trends beginning before birth, because higher proportions of black children than white children in our sample had mothers with plasma 25(OH)D concentrations <50 nmol/L in mid-pregnancy (Supplemental Table 6) and cord plasma 25(OH)D concentrations <50 nmol/L (Supplemental Table 7). This reflects the strong correlation between maternal and cord 25(OH)D concentrations, with lower concentrations in the cord plasma, observed in our cohort and other studies (38). The proportions of mothers and children in each category defined by the Endocrine Society are presented in Supplemental Tables 8-11.
Early-life factors associated with 25(OH)D concentrations in early childhood
Table 2 presents associations of each predictor with early-childhood 25(OH)D concentrations, with season of blood draw included in the base model because it is a strong predictor of 25(OH)D concentrations independently of individual characteristics (Model 1 results). When considering predictors individually, child race/ethnicity was strongly associated with early-childhood 25(OH)D. Fall birth (compared with other seasons), older age at visit, and exposure to cigarette smoke (compared with no exposure) were all associated with lower 25(OH)D concentrations. Children whose mothers did not have a college degree had lower 25(OH)D concentrations than children whose mothers were more educated, and children from lower-income households had lower concentrations than those from higher-income households, but these associations were weaker within racial strata. Higher maternal 25(OH)D during pregnancy and cord plasma 25(OH)D concentrations and higher intakes of dietary VitD, calcium, and milk were all associated with higher 25(OH)D concentrations. Associations with dietary factors appeared to be particularly strong among black children (P = 0.11 for race × dietary VitD interaction, P = 0.03 for race × dietary calcium interaction, and P = 0.11 for race × milk interaction in the subset of children categorized as either “white” or “black”). Measures of adiposity were not associated with 25(OH)D in the overall sample or among white children, but higher waist circumference was associated with lower 25(OH)D among black children (P = 0.10 for race × waist circumference interaction).
TABLE 2.
Associations of child, maternal, and household characteristics with plasma 25(OH)D concentrations in early childhood among 745 Project Viva children1
| All | White | Black | ||||
|---|---|---|---|---|---|---|
| Model 12 | Model 23 | Model 12 | Model 23 | Model 12 | Model 23 | |
| Child characteristics in infancy (median: 6.3 mo) | ||||||
| Race/ethnicity | ||||||
| White | Ref. | Ref. | — | — | — | — |
| Black | −19.6 (−24.5, −14.7) | −8.0 (−13.6, −2.5) | — | — | — | — |
| Other race/ethnicity | −15.8 (−20.3, −11.2) | −8.7 (−13.5, −4.0) | — | — | — | — |
| Sex | ||||||
| Male | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| Female | −1.4 (−5.1, 2.2) | −2.3 (−5.6, 1.0) | −1.8 (−6.2, 2.6) | −2.2 (−6.5, 2.1) | −2.5 (−11.0, 5.9) | −1.4 (−9.6, 6.8) |
| Season of birth | ||||||
| Fall | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| Summer | 9.5 (4.1, 14.8) | 8.3 (3.3, 13.3) | 7.8 (1.3, 14.2) | 8.8 (2.4, 15.2) | 3.9 (−7.9, 15.7) | 3.8 (−7.8, 15.5) |
| Winter | 6.7 (1.3, 12.1) | 8.6 (3.6, 13.5) | 4.3 (−2.5, 11.1) | 7.5 (0.8, 14.2) | 8.3 (−3.4, 20.0) | 11.5 (0.5, 22.6) |
| Spring | 8.6 (3.1, 14.2) | 8.6 (3.5, 13.7) | 6.9 (0.0, 13.7) | 9.2 (2.4, 15.9) | 3.9 (−8.6, 16.5) | 6.5 (−5.5, 18.5) |
| Cord plasma 25(OH)D, 10-nmol/L increase | 5.1 (3.9, 6.2) | 3.0 (1.2, 4.7) | 3.9 (2.4, 5.4) | 3.0 (1.0, 5.0) | 5.1 (2.1, 8.1) | 3.5 (−0.4, 7.3) |
| Breastfeeding status at 6 mo4 | ||||||
| Fully/partially formula-fed | Ref. | — | Ref. | — | Ref. | — |
| Fully breastfed | 2.3 (−1.8, 6.3) | — | −1.7 (−6.4, 3.0) | — | 2.0 (−8.7, 12.7) | — |
| Child characteristics at early childhood visit (median: 3.2 y) | ||||||
| Season of blood sampling | ||||||
| Fall | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| Winter | −15.3 (−20.8, −9.8) | −14.7 (−19.8, −9.6) | −14.0 (−20.6, −7.4) | −15.3 (−21.9, −8.7) | −22.5 (−34.9, −10.0) | −19.7 (−31.4, −8.0) |
| Spring | −14.3 (−19.2, −9.4) | −15.3 (−20.1, −10.5) | −12.8 (−18.5, −7.1) | −14.8 (−21.0, −8.6) | −19.8 (−31.9, −7.7) | −20.0 (−33.0, −8.3) |
| Summer | −2.8 (−7.8, 2.1) | −3.1 (−7.9, 1.6) | 1.3 (−4.7, 7.4) | −2.4 (−8.7, 3.8) | −10.4 (−21.5, 0.8) | −9.6 (−20.6, 1.4) |
| Age, mo | −0.7 (−1.1, −0.3) | −0.4 (−0.8, −0.1) | −0.6 (−1.2, 0.0) | −0.6 (−1.2, 0.0) | −0.5 (−1.1, 0.0) | −0.3 (−0.8, 0.3) |
| Cigarette smoke exposure | ||||||
| None or <1 h/wk | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| ≥1 h/wk | −11.8 (−19.2, −4.4) | −4.3 (−11.2, 2.6) | −9.1 (−19.8, 1.7) | −5.4 (−15.9, 5.1) | −4.2 (−17.5, 9.1) | −2.2 (−15.1, 10.6) |
| BMI z score4 | 0.2 (−1.6, 2.0) | — | 1.4 (−1.0, 3.8) | — | −3.3 (−6.8, 0.1) | — |
| Waist circumference, 1-SD change | 0.3 (−1.5, 2.2) | 0.7 (−1.0, 2.4) | 0.1 (−2.3, 2.5) | 0.7 (−1.6, 3.1) | −4.3 (−7.7, −0.9) | −3.8 (−7.4, −0.2) |
| Sum of skinfolds,4 1-SD change | 1.3 (−0.6, 3.2) | — | 1.7 (−0.8, 4.2) | — | −3.0 (−6.4, 0.4) | — |
| Dietary vitamin D intake | ||||||
| Tertile 1 | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| Tertile 2 | 6.7 (2.0, 11.4) | 5.3 (1.1, 9.6) | 3.3 (−2.8, 9.4) | 3.3 (−2.5, 9.0) | 12.3 (2.0, 22.6) | 9.3 (−1.3, 20.0) |
| Tertile 3 | 10.2 (5.6, 14.8) | 6.3 (2.0, 10.6) | 5.2 (−0.4, 10.8) | 4.9 (−0.5, 10.4) | 13.8 (3.0, 24.7) | 12.2 (0.8, 23.7) |
| Dietary calcium intake,4 10-unit difference | 0.2 (0.1, 0.2) | — | 0.1 (−0.0, 0.2) | — | 0.3 (0.1, 0.4) | — |
| Milk intake | ||||||
| Low (0–4 times/wk) | Ref. | — | Ref. | — | Ref. | — |
| Moderate (nearly daily/daily) | 2.6 (−3.4, 8.6) | — | −4.2 (−12.2, 3.8) | — | 10.2 (−1.0, 21.4) | — |
| High (≥2 times/d) | 9.9 (4.6, 15.2) | — | 1.3 (−5.8, 8.4) | — | 15.1 (4.9, 25.4) | — |
| Vitamin D supplement use | ||||||
| No | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| Yes | 6.7 (2.8, 10.5) | 5.6 (2.0, 9.2) | 6.1 (1.4, 10.9) | 6.2 (1.5, 10.8) | 8.0 (−0.7, 16.6) | 4.2 (−4.2, 12.6) |
| Active play,4 h/d | ||||||
| <1 | Ref. | — | Ref. | — | Ref. | — |
| 1–3 | 7.7 (−1.8, 17.2) | — | 8.0 (−7.0, 23.1) | — | 8.2 (−6.7, 23.2) | — |
| >3 | 5.9 (−4.0, 15.7) | — | 9.2 (−6.0, 24.5) | — | 0.1 (−15.7, 15.9) | — |
| Maternal or household characteristics | ||||||
| Annual household income at enrollment, $ | ||||||
| ≥70,000 | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| <70,000 | −9.7 (−13.5, −6.0) | −3.2 (−7.1, 0.7) | −4.0 (−9.1, 1.1) | −3.7 (−8.7, 1.3) | −6.3 (−16.0, 3.3) | −1.6 (−12.3, 9.1) |
| Mother's educational attainment | ||||||
| College degree or higher | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| Less than a college degree | −8.2 (−12.0, −4.3) | 2.3 (−1.8, 6.4) | −2.4 (−8.0, 3.1) | 1.2 (−4.4, 6.8) | −2.5 (−10.9, 6.0) | 2.5 (−6.6, 11.6) |
| Mother's prepregnancy BMI, kg/m2 | ||||||
| <25 | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| ≥25 | −6.4 (−10.1, −2.7) | −1.1 (−4.7, 2.5) | −4.9 (−9.6, −0.2) | −1.8 (−6.5, 2.9) | 1.9 (−6.6, 10.4) | 3.9 (−4.1, 11.8) |
| Mother's second-trimester plasma 25(OH)D, 10-nmol/L increase | 3.4 (2.5, 4.3) | 1.3 (0.1, 2.6) | 2.6 (1.4, 3.7) | 1.3 (−0.2, 4.8) | 3.1 (1.1, 5.1) | 1.2 (−1.5, 3.9) |
The full sample included 745 children: 493 white, 113 black, and 139 children categorized as “other race/ethnicity” who were not included in the stratified analyses. Estimates are βs (95% CIs) from multivariable linear regression models. 25(OH)D, 25-hydroxyvitamin D.
Model 1 includes season of blood sampling.
Model 2 includes season of blood sampling, child race/ethnicity (model for full sample only), sex, season of birth, cord plasma 25(OH)D concentration, age at visit, cigarette smoke exposure, waist circumference measured at visit, dietary vitamin D intake, and vitamin D supplement use, household income at enrollment, and mother's education level, prepregnancy BMI, and plasma 25(OH)D concentration in mid-pregnancy.
Not included in the final multivariable model (Model 2).
With all other factors in the model, child race/ethnicity remained strongly associated with 25(OH)D. Children born in the fall had lower 25(OH)D than children born in all other seasons, and older age at visit completion was associated with lower 25(OH)D. Maternal education and prepregnancy BMI, household income, and child smoke exposure were no longer significantly associated with child 25(OH)D concentrations after accounting for all other factors owing to wide CIs, but point estimates indicated that children from lower-income households and those exposed to cigarette smoke had lower 25(OH)D concentrations.
Both maternal 25(OH)D during pregnancy and cord plasma 25(OH)D were directly associated with early-childhood 25(OH)D concentrations in the multivariable model. In addition, there was a dose–response relation between tertile of VitD intake and 25(OH)D concentrations, and those who took VitD supplements in early childhood had higher 25(OH)D (than those with no VitD supplement use), with similar associations among white and black children. Finally, the inverse association of child waist circumference with 25(OH)D concentrations among black children persisted, with higher waist circumference corresponding to lower 25(OH)D. The P value for the race × waist circumference interaction was 0.20 in Model 2.
Early-life factors associated with 25(OH)D concentrations in mid-childhood
Table 3 presents factors associated with mid-childhood 25(OH)D concentrations. Child race/ethnicity was a strong predictor of mid-childhood 25(OH)D. Smoke exposure and having a mother without a college degree or with overweight or obesity were associated with lower 25(OH)D, whereas higher maternal 25(OH)D during pregnancy, higher cord plasma 25(OH)D, higher intake of milk and dairy products, and more frequent VitD supplement use were associated with higher 25(OH)D. Mid-childhood 25(OH)D concentrations were inversely associated with various measures of adiposity overall and among black children, but not among white children.
TABLE 3.
Associations of child, maternal, and household characteristics with plasma 25(OH)D concentrations in mid-childhood among 655 Project Viva children1
| All | White | Black | ||||
|---|---|---|---|---|---|---|
| Model 12 | Model 23 | Model 12 | Model 23 | Model 12 | Model 23 | |
| Characteristics in infancy (median: 6.3 mo) | ||||||
| Race/ethnicity | ||||||
| White | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| Black | −16.9 (−20.1, −13.6) | −6.4 (−10.4, −2.4) | — | — | — | — |
| Other | −13.3 (−16.7, −9.9) | −6.4 (−10.0, −2.8) | — | — | — | — |
| Child sex | ||||||
| Male | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| Female | −0.4 (−3.2, 2.5) | −0.5 (−3.0, 1.9) | −0.7 (−4.0, 2.6) | −0.6 (−3.8, 2.6) | −0.3 (−6.4, 5.8) | −0.2 (−6.3, 5.8) |
| Season of birth | ||||||
| Fall | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| Summer | 3.1 (−1.0, 7.1) | 2.8 (−0.8, 6.4) | 1.2 (−3.6, 6.0) | 1.9 (−2.7, 6.5) | −1.5 (−10.3, 7.3) | 0.9 (−7.8, 9.6) |
| Winter | 2.8 (−1.2, 6.9) | 4.5 (1.0, 8.0) | 2.2 (−2.7, 7.1) | 4.3 (−0.4, 9.0) | 0.7 (−8.2, 9.5) | 4.1 (−4.4, 12.6) |
| Spring | 1.1 (−3.0, 5.2) | 1.5 (−2.1, 5.1) | 0.9 (−4.0, 5.9) | 2.5 (−2.2, 7.2) | −3.0 (−12.0, 6.1) | 0.8 (−8.2, 9.8) |
| Cord plasma 25(OH)D, 10-nmol/L increase | 4.2 (3.4, 5.0) | 2.2 (1.0, 3.4) | 2.9 (1.8, 4.0) | 2.0 (0.6, 3.5) | 3.8 (1.7, 5.9) | 2.8 (0.1, 5.5) |
| Breastfeeding status at 6 mo | ||||||
| Fully/partially formula-fed | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| Fully breastfed | 3.8 (0.5, 7.1) | −1.0 (−4.0, 1.9) | 0.2 (−3.6, 3.9) | −1.1 (−4.7, 2.4) | −3.3 (−11.5, 5.0) | −5.5 (−13.7, 2.6) |
| Child characteristics at mid-childhood visit (median: 7.7 y) | ||||||
| Season of blood sampling | ||||||
| Summer | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| Fall | −5.8 (−9.8, −1.7) | −6.5 (−10.1, −3.0) | −7.6 (−12.5, −2.7) | −8.7 (−13.3, −4.0) | −13.3 (−21.9, −4.7) | −14.0 (−22.9, −5.1) |
| Winter | −12.3 (−16.1, −8.5) | −15.3 (−18.7, −11.8) | −17.2 (−21.6, −12.8) | −17.2 (−21.5, −13.0) | −14.6 (−23.7, −5.6) | −15.3 (−24.9, −5.8) |
| Spring | −12.6 (−16.4, −8.9) | −14.7 (−18.0, −11.4) | −15.1 (−19.6, −10.6) | −16.2 (−20.6, −11.8) | −15.1 (−22.7, −7.5) | −13.0 (−20.5, −5.5) |
| Age, mo | −0.3 (−0.4, −0.1) | −0.1 (−0.2, 0.0) | 0.0 (−0.2, 0.2) | 0.0 (−0.2, 0.2) | −0.4 (−0.6, −0.1) | −0.4 (−0.6, −0.1) |
| Cigarette smoke exposure, h/wk | ||||||
| None or <1 | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| ≥1 | −10.0 (−17.0, −3.0) | −2.9 (−9.2, 3.5) | −5.1 (−16.8, 6.5) | −2.0 (−13.4, 9.4) | −4.8 (−16.2, 6.6) | −6.1 (−18.4, 6.2) |
| BMI z score4 | −1.7 (−2.9, −0.5) | — | −0.4 (−2.0, 1.1) | — | −1.8 (−4.1, 0.6) | — |
| Waist circumference, 1-SD change | −2.8 (−4.2, −1.4) | −0.1 (−1.5, 1.3) | −1.0 (−3.0, 1.0) | 0.6 (−1.5, 2.8) | −3.0 (−5.2, −0.7) | −1.1 (−3.7, 1.6) |
| Sum of skinfolds,4 1-SD change | −3.3 (−4.7, −1.9) | — | −1.2 (−3.3, 0.9) | — | −4.0 (−6.2, −1.7) | — |
| Body fat percentage,4 1-SD change | −3.1 (−4.5, −1.7) | — | −0.6 (−2.5, 1.3) | — | −3.5 (−6.0, −1.0) | — |
| Fish intake | ||||||
| Never | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| <1 to 1 time/wk | −1.2 (−4.6, 2.2) | −0.5 (−3.5, 2.5) | −1.8 (−5.5, 2.0) | −1.8 (−5.5, 1.8) | 8.3 (−0.9, 17.5) | 11.6 (2.4, 20.8) |
| ≥2 times/wk | −6.5 (−11.0, −2.1) | −0.9 (−5.0, 3.2) | −4.4 (−10.4, 1.7) | −4.3 (−10.1, 1.5) | 4.2 (−5.5, 14.0) | 8.5 (−1.3, 18.3) |
| Milk intake,4 servings/d | ||||||
| <1 | Ref. | — | Ref. | — | Ref. | — |
| 1–2 | 2.9 (−1.2, 7.0) | — | 5.2 (0.1, 10.2) | — | 1.6 (−6.1, 9.3) | — |
| >2 | 5.0 (1.0, 8.9) | — | 2.9 (−1.8, 7.6) | — | −0.3 (−8.6, 8.0) | — |
| Dairy intake, servings/d | ||||||
| <1 | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| 1–2 | 6.2 (2.3, 10.0) | 1.4 (−2.1, 5.0) | 3.3 (−2.0, 8.6) | 0.8 (−4.4, 5.9) | −0.8 (−7.7, 6.1) | −1.8 (−8.6, 5.0) |
| >2 | 9.6 (5.5, 13.7) | 1.5 (−2.3, 5.4) | 5.2 (−0.2, 10.6) | 2.0 (−3.3, 7.3) | −0.5 (−10.3, 9.4) | −5.0 (−15.5, 5.5) |
| Vitamin D supplement use | ||||||
| Never | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| 1–5 times/wk | 4.4 (1.2, 7.6) | 4.3 (1.4, 7.1) | 2.2 (−1.6, 6.0) | 2.0 (−1.7, 5.7) | 7.9 (1.2, 14.6) | 6.6 (−0.5, 13.7) |
| ≥6 times/wk | 10.5 (6.7, 14.3) | 8.2 (4.8, 11.6) | 9.6 (5.3, 13.8) | 8.8 (4.7, 12.9) | 12.4 (3.0, 21.9) | 13.0 (3.9, 22.2) |
| Outdoor activity (May–October)4 | ||||||
| Low | Ref. | — | Ref. | — | Ref. | — |
| Moderate | 8.6 (−6.3, 23.4) | — | 9.6 (−22.5, 41.7) | — | 5.2 (−14.1, 24.6) | — |
| High | 13.7 (−0.8, 28.2) | — | 13.9 (−17.9, 45.6) | — | 4.6 (−13.8, 23.1) | — |
| Outdoor activity (November–April) | ||||||
| Low | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| Moderate | 4.0 (0.8, 7.2) | 1.5 (−1.3, 4.4) | 3.9 (0.0, 7.8) | 2.4 (−1.5, 6.2) | −0.5 (−7.3, 6.3) | 0.0 (−6.6, 6.6) |
| High | 7.2 (2.6, 11.9) | 4.1 (−0.0, 8.2) | 6.7 (1.3, 12.1) | 3.7 (−1.6, 8.9) | 4.8 (−5.0, 14.5) | 8.0 (−1.7, 17.7) |
| Sunscreen use/covering skin | ||||||
| Never/rarely/occasionally | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| Most of the time/always | 11.1 (7.6, 14.7) | 1.9 (−1.6, 5.5) | 9.9 (3.5, 16.3) | 6.1 (−0.3, 12.4) | −1.2 (−7.7, 5.3) | −4.9 (−11.3, 1.5) |
| Maternal or household characteristics | ||||||
| Annual household income at enrollment, $ | ||||||
| ≥70,000 | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| <70,000 | −10.2 (−13.1, −7.4) | −3.1 (−6.1, −0.1) | −4.9 (−8.7, −1.1) | −3.8 (−7.5, −0.0) | −5.8 (−13.4, 1.8) | −2.8 (−11.0, 5.4) |
| Mother's educational attainment | ||||||
| College degree or higher | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| Less than a college degree | −10.5 (−13.4, −7.7) | −1.0 (−4.1, 2.1) | −3.2 (−7.3, 0.9) | −0.4 (−4.5, 3.8) | −4.9 (−11.2, 1.4) | 0.7 (−6.2, 7.6) |
| Mother's prepregnancy BMI, kg/m2 | ||||||
| <25 | Ref. | Ref. | Ref. | Ref. | Ref. | Ref. |
| ≥25 | −6.6 (−9.5, −3.8) | −1.9 (−4.6, 0.7) | −4.3 (−7.8, −0.8) | −1.5 (−5.0, 2.0) | −3.7 (−10.0, 2.5) | 0.3 (−6.0, 6.5) |
| Mother's second-trimester plasma 25(OH)D, 10-nmol/L increase | 2.9 (2.2, 3.5) | 0.9 (0.1, 1.8) | 2.2 (1.4, 3.0) | 1.1 (0.1, 2.1) | 2.5 (1.0, 4.0) | 1.3 (−0.8, 3.3) |
The full sample included 655 children: 386 white, 141 black, and 128 children categorized as “other race/ethnicity” who were not included in the stratified analyses. Estimates are βs (95% CIs) from multivariable linear regression models. 25(OH)D, 25-hydroxyvitamin D.
Model 1 includes season of blood sampling.
Model 2 includes season of blood sampling, child race/ethnicity (model for full sample only), sex, season of birth, cord plasma 25(OH)D concentration, breastfeeding status at 6 mo, age at visit, cigarette smoke exposure, waist circumference, intake of fish and dairy products, vitamin D supplement use, outdoor activity in November–April, and use of sunscreen/skin covering, household income at enrollment, and mother's education level, prepregnancy BMI, and plasma 25(OH)D concentration in mid-pregnancy.
Not included in the final multivariable model (Model 2).
After accounting for all other factors, child race/ethnicity remained a strong predictor of 25(OH)D. Children born in the fall had lower 25(OH)D concentrations than those born in all other seasons, but the difference was only statistically significant for winter compared with fall births. Among black children, age at mid-childhood visit remained inversely associated with 25(OH)D. Children exposed to secondhand smoke (compared with not exposed), those whose mothers had overweight/obesity (compared with lower BMI), and black children who were fully breastfed at 6 mo (compared with fully or partially formula-fed) had lower 25(OH)D concentrations, but the CIs were wide and estimates not statistically significant. Higher maternal 25(OH)D and cord plasma 25(OH)D were both associated with higher mid-childhood 25(OH)D, with similar associations among white and black children. After accounting for other factors, waist circumference and maternal education were no longer associated with 25(OH)D, but children from lower-income households had lower 25(OH)D concentrations. After accounting for all other variables, higher intake of fish and seafood was associated with higher mid-childhood 25(OH)D among black children (P = 0.07 for race × fish/seafood intake interaction). There was a dose–response relation between VitD supplement use and 25(OH)D.
Discussion
This investigation into early-life factors associated with early and mid-childhood 25(OH)D concentrations identified several important factors as well as some potential differences in the associations between white and black children. As expected, season of blood draw emerged as one of the strongest predictors, with lower 25(OH)D at both time points among children sampled in winter and spring than those sampled in summer or fall. In early childhood, the largest seasonal differences were observed among black children, who had lower 25(OH)D during seasons with limited sunlight. Evidence suggests that UVB light exposure is more effective in increasing 25(OH)D when baseline concentrations are lower (39, 40). Mean early-childhood 25(OH)D concentrations were ∼20 nmol/L higher among black children sampled in fall than among those sampled in winter or spring, suggesting that they were able to synthesize considerable amounts of VitD during months of plentiful sunlight. Interestingly, 25(OH)D concentrations were higher among children who spent more hours outdoors during the winter months, when endogenous UVB-induced synthesis is not believed to occur in appreciable amounts in Boston-area residents (9).
Black children had lower 25(OH)D concentrations than white children after accounting for all other predictors, including SES, adiposity, dietary factors, and prenatal 25(OH)D concentrations. After accounting for season of blood draw, 25(OH)D concentrations were 20 nmol/L lower in early childhood and 17 nmol/L lower in mid-childhood among black children than among white children. This striking difference, and the substantial attenuation observed after accounting for other predictors, suggest both that race is associated with many other factors contributing to lower VitD status and that race categorization, likely reflecting skin pigment which can inhibit UVB light absorption that stimulates endogenous VitD synthesis (41, 42), is important independently of other factors. Other studies have demonstrated lower 25(OH)D and higher risk of VitD deficiency and insufficiency among black children/adolescents (33, 34, 43) than among white individuals. Novel aspects of our study include the simultaneous inclusion of many other predictors of 25(OH)D status that are also associated with race and the identification of differences in some of these predictors between white and black children.
Individuals with overweight/obesity are at higher risk of VitD deficiency (44) than those with lower BMI. However, in our cohort, adiposity in early childhood was inversely associated with VitD status only among black children, although white and black children had similar measures of overall and central adiposity in our cohort. Racial differences in body fat distribution may be more pronounced in samples including older children than in our sample of prepubertal children, and differences among racial strata may vary by age group and developmental stage. In our cohort, differences in adiposity between white and black children were beginning to emerge by mid-childhood.
Notably, we observed the strongest associations of dietary VitD with 25(OH)D among black children, suggesting that increasing VitD in the diet may be particularly effective in improving VitD status among black children. This finding is consistent with findings from NHANES 2011–2014: among non-Hispanic blacks aged ≥1 y, the population-wide prevalence at risk of VitD deficiency decreased linearly with increasing intakes of VitD but it only decreased at the highest intake category among non-Hispanic whites (33). Dietary sources of VitD are limited and include fortified foods, such as milk and other dairy products. In our sample, intake of milk (almost certainly VitD-fortified) was strongly associated with total dietary VitD intake in early childhood and mean total VitD intakes by category of milk intake were similar among black and white children. However, when total dietary VitD intake was not accounted for, milk intake was associated with 25(OH)D among black children only. This difference may reflect evidence suggesting that 25(OH)D responds more to dietary VitD when baseline concentrations are lower (10), as they were among black children in our cohort. Thus, increased intake of VitD-fortified milk might be particularly effective in improving VitD status among black children. Dietary sources of VitD may also be more important for black children owing to their limited ability to synthesize 25(OH)D in response to UVB light. We observed similar associations of VitD supplement use with 25(OH)D concentrations among black and white children, again mirroring findings from NHANES demonstrating a reduction in risk of VitD deficiency and inadequacy with use of VitD supplements among both non-Hispanic black and non-Hispanic white subgroups (33).
Those who were older at their study visit had lower 25(OH)D. This is consistent with several studies of young children (13, 17, 19), although age did not predict VitD status in others (14, 16). A recent analysis of NHANES data demonstrated that in the United States, the proportion of those at risk of VitD deficiency (defined as <30 nmol/L) among 6- to 11-y-olds was nearly 3 times that among 1- to 5-y-olds, and the risk of inadequacy was nearly double (33). An analysis of earlier NHANES data found that odds of VitD deficiency (defined as <15 ng/mL, equivalent to 37.5 nmol/L) among children aged 1–21 y increased linearly with age, after adjusting for sex, race/ethnicity, BMI, and supplement use (3). Age may be a proxy for body fat (17), yet similarly to the NHANES study, we found that increasing age predicted lower 25(OH)D concentrations even after accounting for many other factors including adiposity. Other potential explanations include increased skeletal muscle, which may improve 25(OH)D retention (45), and dietary and lifestyle changes (e.g., lower milk consumption, less supplement use, and less time spent outdoors) with increasing age. Several studies of different age groups have reported lower VitD status among girls than boys (12, 16–19, 35, 43). Similar to our findings, Dong et al. (43) found that black girls had the lowest 25(OH)D concentrations of all race/sex groups, but sex was not an independent predictor of 25(OH)D after accounting for other factors. Differences in body fat, skeletal muscle, and hormones, not yet apparent in prepubertal children, may explain sex differences in VitD status. However, a study of children aged <2 y in New Zealand demonstrated higher 25(OH)D concentrations among boys than girls (12).
Children born in the fall had lower early-childhood 25(OH)D than children born in any other season, even after accounting for season of blood draw, maternal and cord plasma 25(OH)D concentrations, and other predictors. This could be a chance finding in our cohort. However, this relation was also apparent, although weaker, in mid-childhood. In addition, there is some evidence for an association of birth season with later VitD status. Among 11,000 Italian adults, those born in the winter had lower serum 25(OH)D as adults than those born in spring or summer, independently of age, sex, and season of blood sampling (46, 47). Seasonal differences in daylight hours at the time of birth seemed to explain this association, possibly reflecting persistent alterations in VitD metabolism resulting from sunlight exposure at birth. However, a similar study of 10,000 adults in Norway found no relation between season or month of birth and 25(OH)D concentrations in adulthood (48). Interestingly, the authors note that widespread maternal and infant VitD supplementation in Norway might explain the discrepancy in findings between these studies, yet neither study accounted for maternal or infant VitD status in their analysis. We observed a clear inverse association of fall birth with early-childhood 25(OH)D concentrations after including maternal and cord plasma 25(OH)D concentrations in the model, suggesting that prenatal VitD status did not explain this relation. It is unclear why fall birth predicted lower 25(OH)D in our study whereas winter birth was associated with 25(OH)D concentrations similar to those observed for children born in spring or summer. Our data showed the expected pattern of cord plasma 25(OH)D concentrations being highest among infants born in the summer, followed by those born in fall, then spring, and the lowest among infants born in winter. It is possible that season of conception rather than season of birth is the key factor because infants born in the fall would have been conceived during the winter.
Although maternal education and household income both predicted 25(OH)D concentrations in early and mid-childhood accounting only for season of blood draw, these associations were generally weaker within racial strata and after including all other predictors of 25(OH)D in the model. The exception was the association of household income with mid-childhood 25(OH)D: among both black and white children, those from lower-income households had lower 25(OH)D even after including all other predictors in the model. Possible explanations include limited ability to purchase VitD supplements or VitD-fortified foods and lack of access to safe outdoor play spaces among lower-income participants.
VitD status in our cohort reflects trends in the general US population (34): recent findings from NHANES demonstrated that the overall prevalence of risk of VitD inadequacy [25(OH)D <50 nmol/L] was 6.6% (95% CI: 5.2%, 8.3%) among children aged 1–5 y and 12.3% (95% CI: 10.0%, 15.0%) among children aged 6–11 y. In our cohort, the proportions with 25(OH)D concentrations <50 nmol/L were 5% in early childhood and 18% in mid-childhood.
Our study used a large sample from a well-characterized cohort with data on many potential predictors of 25(OH)D, which we examined simultaneously as well as separately. Many interrelated factors are associated with VitD status, and it is important to identify those related to 25(OH)D concentrations independently of their associations with other factors to identify targets for studies that will support conclusions about causal relations. We had data and 25(OH)D measurements from 2 distinct time points, allowing us to examine whether predictors of 25(OH)D remain stable throughout childhood. Our results suggest that factors including race/ethnicity (likely as a proxy for skin pigmentation), VitD status at birth, and VitD intake from food and supplements are important determinants of 25(OH)D concentrations in childhood, whereas high maternal BMI and low maternal education level are not independently related to 25(OH)D but may be used to flag children at risk of VitD deficiency. We examined associations within subgroups of participants stratified by race, itself a major predictor of child VitD status, to identify consistent and variable factors associated with 25(OH)D between black and white children.
Our study had some limitations. First, 19% of our sample was categorized as having a maternal-reported race/ethnicity other than black or white. Because this category was heterogeneous and included children with varying degrees of skin pigmentation, a highly relevant factor in considering VitD status, it would not be informative to examine stratum-specific associations among the children categorized as “other race/ethnicity”; therefore, this group was excluded from the stratified analyses. In addition, we acknowledge that the racial/ethnic categorizations are broad and that each group includes children with variable individual characteristics, including a range of degrees of skin pigmentation which we were unable to measure in this study. We acknowledge that our study was not powered to detect interactions of early-life factors by race/ethnicity, and although we identified some potential differences in predictors of 25(OH)D concentrations between white and black children, the stratified analyses were exploratory and our results should be verified with additional studies specifically designed to assess these interactions.
In addition, none of the children in our study population were identified as being at risk of deficiency in early childhood and few were at risk of inadequacy, and there was also a low prevalence of vitamin D deficiency in mid-childhood. Therefore, our ability to examine predictors of clinical vitamin D deficiency according to the cutoffs recommended by the IOM was limited, yet identification of predictors of lower vitamin D on a continuous scale is arguably more useful because there is no clear consensus on the appropriate cutoffs for categorizing VitD status, including deficiency (49).
Our VitD intake assessment is likely subject to substantial measurement error. Our FFQ did not distinguish intake of different types of fish or the VitD content of commonly fortified foods, such as milk. However, we assume that most children were drinking milk fortified with VitD. We adjusted total dietary VitD in early childhood for total energy intake and ranked participants into tertiles of dietary VitD and categories of milk, dairy, and fish intake. Any residual measurement error would likely be random and result in attenuated effect estimates. In addition, we were not able to ascertain the exact dose or frequency of VitD supplement use. Our assumption that children who had taken a VitD supplement during the month before questionnaire administration were using these supplements regularly is a limitation to consider when interpreting our results. In addition, the VitD content of supplements varies, and evidence suggests that there is a nonlinear, dose-dependent response of serum 25(OH)D to VitD intake (10). However, the threshold dose for a differential response has been identified as 1000 IU/d, and the vast majority of children in our cohort were taking children's multivitamins with a typical dose of 400–600 IU/d and had a mean dietary VitD intake in early childhood of ∼200 IU/d.
Although we examined associations of 25(OH)D concentrations with factors measured at the same time, these variables are indicators of behaviors exhibited over the long term [i.e., before measurement of 25(OH)D], and there is no reason to think that they would be affected by the child's VitD status. A possible exception is child adiposity, because the direction of the relation between VitD status and adiposity is unclear (50, 51). In the future, we plan to use our longitudinal data to examine 25(OH)D concentrations at various time points in early life in relation to later adiposity.
We collected only 1 blood sample at each time point, and the children were sampled throughout the year. However, estimates of the impact of season of blood collection were similar with and without accounting for other factors such as child age, sex, household income, and race/ethnicity, suggesting that children sampled in different seasons were not systematically different relative to other characteristics. Finally, we used different assays to measure 25(OH)D concentrations in early and mid-childhood. However, we do not expect that the type of assay would affect the ranking of participants at each time point. Our data may be used to identify factors that are associated with 25(OH)D concentrations at each time point, including those that are consistent and variable, but the magnitude of these associations should not be compared between time points.
In conclusion, in our cohort of children based in the northeastern United States, we identified several key early-life factors related to 25(OH)D concentrations in early and mid-childhood. We anticipate that our results will be useful in identifying children potentially at risk of VitD deficiency as well as informing potential recommendations on how to improve VitD status in childhood. Our results also highlight some potential differences in early-life predictors of 25(OH)D between white and black children and confirm others’ observations that race, likely as a proxy for skin pigmentation, is a strong predictor of 25(OH)D concentrations even after accounting for many other factors. Finally, our findings may inform future research attempting to elucidate causal relations between various exposures and child VitD status using appropriate methods that simulate randomized controlled experiments.
Supplementary Material
ACKNOWLEDGEMENTS
The authors’ responsibilities were as follows—KMS, CAC, and EO: designed the research; KMS: analyzed the data, wrote the paper, and had primary responsibility for the final content; SLR-S: prepared the data set and assisted with statistical programming; CAC and EO: provided study oversight; HF: assisted with the data cleaning and literature review; and all authors: read and approved the final manuscript.
Notes
Supported by Eunice Kennedy Shriver National Institute of Child Health and Human Development grant R01 HD034568 (to EO) and NIH Office of the Director grant UH3 OD023286 (to EO).
Author disclosures: the authors report no conflicts of interest.
Supplemental Tables 1–11 and Supplemental Figure 1 are available from the “Supplementary data” link in the online posting of the article and from the same link in the online table of contents at https://academic.oup.com/jn/.
Abbreviations used: SS, subscapular; TR, triceps; VitD, vitamin D; 25(OH)D, 25-hydroxyvitamin D.
Contributor Information
Karen M Switkowski, Division of Chronic Disease Research Across the Lifecourse, Department of Population Medicine, Harvard Medical School and Harvard Pilgrim Health Care Institute, Boston, MA, USA.
Carlos A Camargo, Jr, Department of Emergency Medicine, Massachusetts General Hospital, Harvard Medical School, Boston, MA, USA.
Sheryl L Rifas-Shiman, Division of Chronic Disease Research Across the Lifecourse, Department of Population Medicine, Harvard Medical School and Harvard Pilgrim Health Care Institute, Boston, MA, USA.
Hannah Fuller, Division of Chronic Disease Research Across the Lifecourse, Department of Population Medicine, Harvard Medical School and Harvard Pilgrim Health Care Institute, Boston, MA, USA.
Emily Oken, Division of Chronic Disease Research Across the Lifecourse, Department of Population Medicine, Harvard Medical School and Harvard Pilgrim Health Care Institute, Boston, MA, USA; Department of Nutrition, Harvard TH Chan School of Public Health, Boston, MA, USA.
References
- 1.Petersen RA, Dalskov S-M, Sørensen LB, Hjorth MF, Andersen R, Tetens I, Krarup H, Ritz C, Astrup A, Michaelsen KFet al. Vitamin D status is associated with cardiometabolic markers in 8–11-year-old children, independently of body fat and physical activity. Br J Nutr. 2015;114:1647–55. [DOI] [PubMed] [Google Scholar]
- 2.Ke L, Mason RS, Baur LA, Cowell CT, Liu X, Garnett SP, Brock KE. Vitamin D levels in childhood and adolescence and cardiovascular risk factors in a cohort of healthy Australian children. J Steroid Biochem Mol Biol. 2018;177:270–7. [DOI] [PubMed] [Google Scholar]
- 3.Kumar J, Muntner P, Kaskel FJ, Hailpern SM, Melamed ML. Prevalence and associations of 25-hydroxyvitamin D deficiency in US children: NHANES 2001–2004. Pediatrics. 2009;124:e362–70. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Newberry SJ, Chung M, Shekelle PG, Booth MS, Liu JL, Maher AR, Motala A, Cui M, Perry T, Shanman Ret al. Vitamin D and calcium: a systematic review of health outcomes (update). Evid Rep Technol Assess (Full Rep). 2014;(217):1–929. [DOI] [PubMed] [Google Scholar]
- 5.Cediel G, Corvalán C, López de Romaña D, Mericq V, Uauy R. Prepubertal adiposity, vitamin D status, and insulin resistance. Pediatrics. 2016;138:e20160076. [DOI] [PubMed] [Google Scholar]
- 6.Gilbert-Diamond D, Baylin A, Mora-Plazas M, Marin C, Arsenault JE, Hughes MD, Willett WC, Villamor E. Vitamin D deficiency and anthropometric indicators of adiposity in school-age children: a prospective study. Am J Clin Nutr. 2010;92:1446–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Cranney A, Horsley T, O'Donnell S, Weiler H, Puil L, Ooi D, Atkinson S, Ward L, Moher D, Hanley Det al. Effectiveness and safety of vitamin D in relation to bone health. Evid Rep Technol Assess (Full Rep). 2007;(158):1–235. [PMC free article] [PubMed] [Google Scholar]
- 8.Gordon CM, Feldman HA, Sinclair L, Williams AL, Kleinman PK, Perez-Rossello J, Cox JE. Prevalence of vitamin D deficiency among healthy infants and toddlers. Arch Pediatr Adolesc Med. 2008;162:505–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Holick MF. Resurrection of vitamin D deficiency and rickets. J Clin Invest. 2006;116:2062–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Institute of Medicine . Dietary Reference Intakes for calcium and vitamin D. Washington (DC): The National Academies Press; 2011. [PubMed] [Google Scholar]
- 11.Rovner AJ, O'Brien KO. Hypovitaminosis D among healthy children in the United States: a review of the current evidence. Arch Pediatr Adolesc Med. 2008;162:513–9. [DOI] [PubMed] [Google Scholar]
- 12.Houghton LA, Szymlek-Gay EA, Gray AR, Ferguson EL, Deng X, Heath A-LM. Predictors of vitamin D status and its association with parathyroid hormone in young New Zealand children. Am J Clin Nutr. 2010;92:69–76. [DOI] [PubMed] [Google Scholar]
- 13.Soininen S, Eloranta A-M, Lindi V, Venäläinen T, Zaproudina N, Mahonen A, Lakka TA. Determinants of serum 25-hydroxyvitamin D concentration in Finnish children: the Physical Activity and Nutrition in Children (PANIC) study. Br J Nutr. 2016;115:1080–91. [DOI] [PubMed] [Google Scholar]
- 14.Sochorová L, Hanzlíková L, Černá M, Vosátková M, Grafnetterová AP, Fialová A, Kubínová R. Assessment of vitamin D status in Czech children. Cent Eur J Public Health. 2018;26:260–4. [DOI] [PubMed] [Google Scholar]
- 15.Hansen L, Tjønneland A, Køster B, Brot C, Andersen R, Cohen AS, Frederiksen K, Olsen A. Vitamin D status and seasonal variation among Danish children and adults: a descriptive study. Nutrients. 2018;10:1801. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Cairncross CT, Stonehouse W, Conlon CA, Grant CC, McDonald B, Houghton LA, Eyles D, Camargo CA Jr, Coad J, von Hurst PR. Predictors of vitamin D status in New Zealand preschool children. Matern Child Nutr. 2017;13:e12340. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Science M, Maguire JL, Russell ML, Smieja M, Walter SD, Loeb M. Prevalence and predictors of low serum 25-hydroxyvitamin D levels in rural Canadian children. Paediatr Child Health. 2017;22:125–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.Karin Z, Gilic B, Supe Domic D, Sarac Z, Ercegovic K, Zenic N, Uljevic O, Peric M, Markic J. Vitamin D status and analysis of specific correlates in preschool children: a cross-sectional study in southern Croatia. Int J Environ Res Public Health. 2018;15:2503. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Nwosu BU, Kum-Nji P. Tobacco smoke exposure is an independent predictor of vitamin D deficiency in US children. PLoS One. 2018;13:e0205342. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Mandlik R, Kajale N, Ekbote V, Patwardhan V, Khadilkar V, Chiplonkar S, Khadilkar A. Determinants of vitamin D status in Indian school-children. Indian J Endocrinol Metab. 2018;22:244–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Oken E, Baccarelli AA, Gold DR, Kleinman KP, Litonjua AA, De Meo D, Rich-Edwards JW, Rifas-Shiman SL, Sagiv S, Taveras EMet al. Cohort profile: Project Viva. Int J Epidemiol. 2015;44:37–48. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Provenzano AM, Rifas-Shiman SL, Herring SJ, Rich-Edwards JW, Oken E. Associations of maternal material hardships during childhood and adulthood with prepregnancy weight, gestational weight gain, and postpartum weight retention. J Womens Health (Larchmt). 2015;24:563–71. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Blum RE, Wei EK, Rockett HR, Langeliers JD, Leppert J, Gardner JD, Colditz GA. Validation of a food frequency questionnaire in Native American and Caucasian children 1 to 5 years of age. Matern Child Health J. 1999;3:167–72. [DOI] [PubMed] [Google Scholar]
- 24.US Department of Agriculture, Agricultural Research Service . USDA National Nutrient Database for Standard Reference, Release 14. Methods and Application of Food Composition Laboratory home page. [Internet]. 2001; [cited 2020 Jun 2]. Available from: https://www.ars.usda.gov/ARSUserFiles/80400535/DATA/sr14/sr14_doc.pdf. [Google Scholar]
- 25.Willett WC, Howe GR, Kushi LH. Adjustment for total energy intake in epidemiologic studies. Am J Clin Nutr. 1997;65:1220S–8S.; discussion 1229S–31S. [DOI] [PubMed] [Google Scholar]
- 26.Mackerras D. Energy adjustment: the concepts underlying the debate. J Clin Epidemiol. 1996;49:957–62. [DOI] [PubMed] [Google Scholar]
- 27.Rifas-Shiman SL, Willett WC, Lobb R, Kotch J, Dart C, Gillman MW. PrimeScreen, a brief dietary screening tool: reproducibility and comparability with both a longer food frequency questionnaire and biomarkers. Public Health Nutr. 2001;4:249–54. [DOI] [PubMed] [Google Scholar]
- 28.Kuczmarski RJ, Ogden CL, Grummer-Strawn LM, Flegal KM, Guo SS, Wei R, Mei Z, Curtin LR, Roche AF, Johnson CL. CDC growth charts: United States. Adv Data. 2000;(314):1–27. [PubMed] [Google Scholar]
- 29.Boeke CE, Oken E, Kleinman KP, Rifas-Shiman SL, Taveras EM, Gillman MW. Correlations among adiposity measures in school-aged children. BMC Pediatr. 2013;13:99. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.National Institute of Standards & Technology (NIST) . Certificate of Analysis, Standard Reference Material 972a: vitamin D metabolites in frozen human serum. [Internet]. Gaithersburg (MD): NIST; 2017; [cited 2020 Jul 14]. Available from: https://www-s.nist.gov/srmors/certificates/972A.pdf. [Google Scholar]
- 31.Singh RJ, Taylor RL, Reddy GS, Grebe SKG. C-3 epimers can account for a significant proportion of total circulating 25-hydroxyvitamin D in infants, complicating accurate measurement and interpretation of vitamin D status. J Clin Endocrinol Metab. 2006;91:3055–61. [DOI] [PubMed] [Google Scholar]
- 32.Holick MF, Binkley NC, Bischoff-Ferrari HA, Gordon CM, Hanley DA, Heaney RP, Murad MH, Weaver CM, Endocrine Society . Evaluation, treatment, and prevention of vitamin D deficiency: an Endocrine Society clinical practice guideline. J Clin Endocrinol Metab. 2011;96:1911–30. [DOI] [PubMed] [Google Scholar]
- 33.Herrick KA, Storandt RJ, Afful J, Pfeiffer CM, Schleicher RL, Gahche JJ, Potischman N. Vitamin D status in the United States, 2011–2014. Am J Clin Nutr. 2019;110:150–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Rajakumar K, de las Heras J, Chen TC, Lee S, Holick MF, Arslanian SA. Vitamin D status, adiposity, and lipids in black American and Caucasian children. J Clin Endocrinol Metab. 2011;96:1560–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Tolppanen A-M, Fraser A, Fraser WD, Lawlor DA. Risk factors for variation in 25-hydroxyvitamin D3 and D2 concentrations and vitamin D deficiency in children. J Clin Endocrinol Metab. 2012;97:1202–10. [DOI] [PubMed] [Google Scholar]
- 36.White IR, Royston P, Wood AM. Multiple imputation using chained equations: issues and guidance for practice. Stat Med. 2011;30:377–99. [DOI] [PubMed] [Google Scholar]
- 37.Rubin DB. Multiple imputation for nonresponse in surveys. Hoboken (NJ): Wiley-Interscience; 2004. [Google Scholar]
- 38.Switkowski KM, Camargo CA Jr, Perron P, Rifas-Shiman SL, Oken E, Hivert M-F. Cord blood vitamin D status is associated with cord blood insulin and C-peptide in two cohorts of mother-newborn pairs. J Clin Endocrinol Metab. 2019;104:3785–94. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Karppinen T, Ala-Houhala M, Ylianttila L, Kautiainen H, Lakkala K, Hannula H-R, Turunen E, Viljakainen H, Reunala T, Snellman E. The effect of vernal solar UV radiation on serum 25-hydroxyvitamin D concentration depends on the baseline level: observations from a high latitude in Finland. Int J Circumpolar Health. 2017;76:1272790. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Lagunova Z, Porojnicu AC, Aksnes L, Holick MF, Iani V, Bruland OS, Moan J. Effect of vitamin D supplementation and ultraviolet B exposure on serum 25-hydroxyvitamin D concentrations in healthy volunteers: a randomized, crossover clinical trial. Br J Dermatol. 2013;169:434–40. [DOI] [PubMed] [Google Scholar]
- 41.Matsuoka LY, Wortsman J, Haddad JG, Kolm P, Hollis BW. Racial pigmentation and the cutaneous synthesis of vitamin D. Arch Dermatol. 1991;127:536–8. [PubMed] [Google Scholar]
- 42.Armas LAG, Dowell S, Akhter M, Duthuluru S, Huerter C, Hollis BW, Lund R, Heaney RP. Ultraviolet-B radiation increases serum 25-hydroxyvitamin D levels: the effect of UVB dose and skin color. J Am Acad Dermatol. 2007;57:588–93. [DOI] [PubMed] [Google Scholar]
- 43.Dong Y, Pollock N, Stallmann-Jorgensen IS, Gutin B, Lan L, Chen TC, Keeton D, Petty K, Holick MF, Zhu H. Low 25-hydroxyvitamin D levels in adolescents: race, season, adiposity, physical activity, and fitness. Pediatrics. 2010;125:1104–11. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 44.Wortsman J, Matsuoka LY, Chen TC, Lu Z, Holick MF. Decreased bioavailability of vitamin D in obesity. Am J Clin Nutr. 2000;72:690–3. [DOI] [PubMed] [Google Scholar]
- 45.Abboud M, Rybchyn MS, Rizk R, Fraser DR, Mason RS. Sunlight exposure is just one of the factors which influence vitamin D status. Photochem Photobiol Sci. 2017;16:302–13. [DOI] [PubMed] [Google Scholar]
- 46.Lippi G, Bonelli P, Buonocore R, Aloe R. Birth season and vitamin D concentration in adulthood. Ann Transl Med. 2015;3:231. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 47.Lippi G, Mattiuzzi C, Aloe R. The impact of seasonality and other determinants on vitamin D concentration in childhood and adulthood: still an unresolved issue. Ann Transl Med. 2016;4:21. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 48.Zostautiene I, Jorde R, Grimnes G. No association between birth season and vitamin D concentration in adults in a North Norwegian population—the Tromsø study. Ann Transl Med. 2016;4:20. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 49.US Preventive Services Task Force (USPSTF) . Vitamin D deficiency: screening, final recommendation statement. [Internet]. Rockville (MD): USPSTF; 2014; Nov. [Cited 2020 Jun 2]. Available from: https://www.uspreventiveservicestaskforce.org/uspstf/document/RecommendationStatementFinal/vitamin-d-deficiency-screening#bootstrap-panel–8. [Google Scholar]
- 50.Hyppönen E, Boucher BJ. Adiposity, vitamin D requirements, and clinical implications for obesity-related metabolic abnormalities. Nutr Rev. 2018;76:678–92. [DOI] [PubMed] [Google Scholar]
- 51.Vranić L, Mikolašević I, Milić S. Vitamin D deficiency: consequence or cause of obesity?. Medicina (Kaunas). 2019;55:541. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.