1. Widespread Vitamin D Deficiency in Urban
Massachusetts Newborns and Their Mothers
WHAT’S KNOWN ON THIS SUBJECT: Vitamin D deficiency seems AUTHORS: Anne Merewood, MPH, IBCLC,a Supriya D.
to be more widespread than has been previously understood, and Mehta, PhD, MHS,b Xena Grossman, RD, MS,a Tai C. Chen,
newborns and their mothers are at high risk. PhD,c Jeffrey S. Mathieu, MS,c Michael F. Holick, PhD, MD,c
and Howard Bauchner, MDa
a
WHAT THIS STUDY ADDS: Information about predisposing Division of General Pediatrics, Boston Medical Center, and
factors for vitamin D deficiency in newborns and mothers and Boston University School of Medicine, Boston, Massachusetts;
bDivision of Epidemiology and Biostatistics, University of Illinois
about the relationship between maternal and infant vitamin D
Chicago School of Public Health, Chicago, Illinois; cDivision of
status. The results also suggest that prenatal vitamins are not Endocrinology, Diabetes and Nutrition, Department of Medicine,
adequate sources of vitamin D for pregnant women. Boston University School of Medicine, Boston, Massachusetts
KEY WORDS
vitamin D, newborns, urban, minority health, pregnancy, birth
ABBREVIATIONS
25(OH)D—25-hydroxyvitamin D
abstract WIC—Supplemental Nutrition Program for Women, Infants, and
Children
OBJECTIVE: To determine vitamin D status and associated factors in a CI— confidence interval
cohort of newly delivered infants and their mothers in Boston, aOR—adjusted odds ratio
Massachusetts. www.pediatrics.org/cgi/doi/10.1542/peds.2009-2158
PATIENTS AND METHODS: Enrollment in this cross-sectional study took doi:10.1542/peds.2009-2158
place from 2005 to 2007 in an urban Boston teaching hospital with 2500 Accepted for publication Jan 14, 2010
births per year. A questionnaire and medical-record data were used to Address correspondence to Anne Merewood, MPH, IBCLC,
identify variables that are potentially associated with vitamin D deficiency Division of General Pediatrics, Boston Medical Center, 88 E
(25-hydroxyvitamin D [25(OH)D] 20 ng/mL). Infant and maternal blood Newton St, Vose 3, Boston, MA 02118. E-mail:
was obtained by venipuncture within 72 hours of birth. The main outcome anne.merewood@bmc.org
measure was infant and maternal 25(OH)D status, assessed by competi- PEDIATRICS (ISSN Numbers: Print, 0031-4005; Online, 1098-4275).
tive protein binding. Copyright © 2010 by the American Academy of Pediatrics
RESULTS: We enrolled 459 healthy mother/infant pairs. After subsequent FINANCIAL DISCLOSURE: The authors have indicated they have
exclusions, analyses were performed on 376 newborns and 433 women. no financial relationships relevant to this article to disclose.
The median infant 25(OH)D level was 17.2 ng/mL (95% confidence interval
[CI]: 16.0 –18.8; range: 5.0 to 60.8 ng/mL). The median maternal 25(OH)D
level was 24.8 ng/mL (95% CI: 23.2–25.8; range: 5.0 to 79.2 ng/mL). Over-
all, 58.0% of the infants and 35.8% of the mothers were vitamin D deficient
(25[OH]D 20 ng/mL); 38.0% of the infants and 23.1% of the mothers were
severely deficient (25[OH]D 15 ng/mL). Risk factors for infant vitamin D
deficiency included maternal deficiency (adjusted odds ratio [aOR]: 5.28
[95% CI: 2.90 –9.62]), winter birth (aOR: 3.86 [95% CI: 1.74 – 8.55]), black
race (aOR: 3.36 [95% CI: 1.37– 8.25]), and a maternal BMI of 35 (aOR: 2.78
[95% CI: 1.18 – 6.55]). Maternal prenatal-vitamin use throughout the sec-
ond and third trimesters was protective against infant deficiency (aOR:
0.30 [95% CI: 0.16 – 0.56]). Similarly, prenatal-vitamin use of 5 times per
week in the third trimester was protective for mothers (aOR: 0.37 [95% CI:
0.20 – 0.69]). Despite this, 30% of the women who took prenatal vitamins
were still vitamin D deficient at the time of birth.
CONCLUSIONS: A high proportion of infants and their mothers in New
England were vitamin D deficient. Prenatal vitamins may not contain
enough vitamin D to ensure replete status at the time of birth. Pediatrics
2010;125:640–647
640 MEREWOOD et al
Downloaded from www.pediatrics.org. Provided by Siu School Of Medicine on March 25, 2010

2. ARTICLES
In 1898, 80% of inpatient children English, Spanish, or French; was white, Skin color was categorized as “light,”
younger than 2 years of age at a Bos- black, or Hispanic (because of insuffi- “medium,” and “dark” according to
ton, Massachusetts, hospital showed cient numbers of potential subjects categories derived from the Fitz-
physical signs of rickets.1 The disease, from other races/ethnicities); was not patrick et al skin-type matrix.17 Be-
ubiquitous in industrialized nations in using illegal drugs; and had no history cause race/ethnicity is not necessarily
the early 20th century, virtually disap- of parathyroid, renal, or liver disease. an accurate measure of skin color,
peared after its causal pathway was Race was self-identified through the analyses were conducted according to
recognized and the hormone “vitamin” enrollment questionnaire. Pairs were skin color as well as by race/ethnicity.
D was added to milk for population- ineligible if the infant was premature,
level supplementation.2 Recent re- had a congenital anomaly, or was ad- Diagnostic Test
ports, however, have suggested that mitted to the NICU or if the mother had Serum 25-hydroxyvitamin D (25[OH]D),
rickets is reemerging1,3 and demon- spent more than half the pregnancy accepted as the indicator of vitamin D
strated that vitamin D deficiency outside the greater Boston area (be-
status in individuals, was measured by
is widespread in industrialized na- cause of the potential for different
competitive protein binding as de-
tions.1,4–8 In 2008, the American Acad- sunlight-exposure levels from women
scribed by Chen et al.18 The competitive
emy of Pediatrics issued new guide- who were pregnant in Boston). The
protein-binding assay that was per-
lines for vitamin D intake in children, study was approved by the Boston Uni-
formed used the vitamin D– binding
from infancy through adolescence, versity Medical Center institutional re-
protein that recognizes 25(OH)D3
and called for an increase in previous view board.
equally as well as 25(OH)D2. This assay
supplementation recommendations.9 After signing informed-consent forms, was validated previously by liquid
Vitamin D deficiency has been associ- women completed a questionnaire on chromatography tandem mass spec-
ated with a broader range of adverse demographic, lifestyle, and behavioral troscopy.19 The lower limit of detection
health outcomes than was previously factors; additional data were obtained was 5 ng/mL, and the intraassay and
acknowledged7,10,11 and seems to be from the medical record. Before dis- interassay coefficients of variation
prevalent in women of childbearing charge from the hospital, a blood sam- were 5.0% to 10% and 10% to 15%, re-
age and young children.12–15 The goal of ple was obtained by venipuncture from
spectively. The reference range was 20
our study was to examine vitamin D the mother and infant.
to 100 ng/mL. Serum was frozen at
status in mothers and newborns in Variables assumed to be pertinent to 80°C and stored, and tests were run
Boston and to determine factors that vitamin D status included season of in batches approximately every 2
were associated with vitamin D defi- birth (vitamin D is made primarily by weeks.
ciency in that population, such as the skin in sunlight, and research re-
prenatal-vitamin use. Although dark- sults have suggested that sunlight is Statistical Analyses
skinned people who live in the inner only intense enough between March
city are at the highest risk of defi- We conducted 2 separate analyses to
and October, in Boston, to produce vi-
ciency, few studies have examined examine risks for vitamin D deficiency
tamin D316); use of prenatal vitamins
infants and their mothers in this in infants and their mothers. The out-
(most of which contain 400 IU per daily
population. come was vitamin D deficiency (dichot-
dose); maternal prepregnancy BMI
omized as 20 vs 20 ng/mL) and
(obtained from the medical record);
METHODS was examined separately for new-
milk consumption; clothing habits; and
The study was conducted at an urban other measures of sunlight exposure. borns and mothers. Newborns and
Boston teaching hospital with 2500 Numerous questions were asked to as- mothers with 25(OH)D levels below the
births per year and a primarily low- sess clothing habits. For example, level of detection ( 5 ng/mL) were
income black and Hispanic population. women were asked to recall for each categorized as being at 20 ng/mL
Between January 2005 and December trimester how frequently they went for models examining predictors of
2007, women were screened for eligi- outside in clothing that revealed their deficiency.
bility and approached in the maternity legs from below the knees and their Explanatory variables in the infant
service; those who agreed to partici- arms from the shoulders down. Other model were season of birth, mother’s
pate were enrolled within 72 hours of measures to determine sunlight expo- skin color and 25(OH)D status, mater-
the birth of their child. Mother/infant sure included reported use of sun- nal sociodemographics, gestational
pairs were eligible if the mother spoke screen and time spent in the sun. age, birth weight and length, gender,
PEDIATRICS Volume 125, Number 4, April 2010 641
Downloaded from www.pediatrics.org. Provided by Siu School Of Medicine on March 25, 2010

3. Apgar score, and mother’s sun expo- analysis at a P value of .20. In addi- women; because we were indepen-
sure and prenatal-vitamin use. tion, we used the generalized estimat- dently interested in a woman’s sta-
Explanatory variables in the maternal ing equations (GEE) extension of gener- tus at the time of delivery, women
model were maternal sociodemo- alized linear models to account for the were not excluded from maternal
graphics (age, educational attainment, within-subject correlation among the analyses because their infant’s
marital status, employment status, repeated measures, assuming bino- blood was not available.
health insurance, Supplemental Nutri- mial distributions with log link. None of The median infant 25(OH)D level was
tion Program for Women, Infants, and the repeated-measure variables were 17.2 ng/mL (95% confidence interval
Children [WIC] status, and place of statistically significant in GEE20 except [CI]: 16.0 –18.8; range: 5.0 to 60.8 ng/
birth), parity, mode of delivery, hospi- for vitamin use. In addition, variables
mL). The median maternal 25(OH)D
talization, bed rest, weight gain during that assessed sun exposure had low
level was 24.8 ng/mL (95% CI:
pregnancy, BMI, and skin color (light, correlation between the first and third
23.2–25.8; range: 5.0 to 79.2 ng/mL).
medium, or dark). For each trimester trimesters (r 0.27, all measures),
According to the standard of 20
of pregnancy, we examined prenatal- which was expected because of sea-
ng/mL as deficient,14 58.0% of the in-
vitamin use and sun exposure (fre- sonal variation in weather. The
fants and 35.8% of the mothers were
quency and duration of time spent out- correlation between first- and third-
vitamin D deficient. With a 25(OH)D
side, sun-protective-factor use). Intake trimester vitamin use was 0.38. There-
level 15 ng/mL as a marker of severe
of foods high in vitamin D content, in- fore, composite variables were cre-
deficiency, 38% of the infants and
cluding vitamin D–fortified milk and ated as needed, and standard logistic
regression modeling was used to iden- 23.1% of the mothers were severely vi-
orange juice, cod liver oil, certain
tify factors associated with maternal tamin D deficient.
fish, fortified cereals, and eggs, was
ascertained from a food-frequency and infant vitamin D deficiency.
Infant Vitamin D Status
questionnaire. Variables significant at the P .20
level in exploratory analysis were ex- In bivariate analyses, variables associ-
We estimated a sample size of 417 in-
amined by univariate logistic regres- ated with infant vitamin D status at
fants and mothers to observe a 20%
sion. Variables significant at the P birth included season of birth and the
difference in the proportion of infants
and mothers with vitamin D deficiency .20 level by likelihood-ratio testing mother’s vitamin D status, skin color,
according to race and season with were entered into a multivariable lo- use of prenatal vitamins, and clothing
80% power (2-sided test of signifi- gistic regression model by using for- habits (Table 1). Although duration of
cance: .017 for multiple compari- ward stepwise entry. Wald-test P val- sun exposure was statistically signifi-
sons). The racial makeup of the popu- ues are presented for the final cantly associated with maternal vita-
lation was known in advance, and we multivariable model. Data were ana- min D status, it was not associated
assumed that births would be evenly lyzed by using Stata/SE 9.2 for Win- with infant vitamin D status. After mul-
distributed over the seasons. However, dows (Stata Corp, College Station, TX). tivariable logistic regression, risk fac-
to further ensure that equal numbers tors for infant vitamin D deficiency that
of individuals of the desired races RESULTS remained statistically significant in-
were enrolled in the desired season, We enrolled 458 mother/infant pairs cluded maternal deficiency (adjusted
we tracked enrollment over the course between January 1, 2005, and Decem- odds ratio [aOR]: 5.28 [95% CI: 2.90 –
of the study. When numbers of individ- ber 31, 2007. Of these, 12 pairs were 9.62]), winter birth (aOR: 3.86 [95% CI:
uals of specific races were not evenly later excluded for the following rea- 1.74 – 8.55]), black race (aOR: 3.36
distributed, we proactively enrolled to sons: 1 spent 7 months of the preg- [95% CI: 1.37– 8.25]), and a maternal
meet the desired quota. nancy outside of the United States, 2 BMI of 35 (aOR: 2.78 [95% CI: 1.18 –
Differences between explanatory vari- withdrew consent, and for 9 pairs we 6.55]). Maternal use of prenatal vita-
ables and the outcome were assessed failed to obtain either infant or mater- mins 5 days/week during at least the
by 2 test for categorical variables. In nal blood samples. In addition, 70 en- second and third trimesters (com-
cross-sectional analysis, there were rolled infants and 13 women were pared with use of prenatal vitamins
no first-trimester variables regarding subsequently excluded because of 5 days week in at least 2 trimesters)
sun exposure and vitamin use that failure to obtain an adequate blood was protective of vitamin D deficiency
were associated with infant or mater- sample. Thus, analyses were per- in the infant (aOR: 0.30 [95% CI: 0.16 –
nal vitamin D status in exploratory formed on 376 newborns and 433 0.56]) (Table 2).
642 MEREWOOD et al
Downloaded from www.pediatrics.org. Provided by Siu School Of Medicine on March 25, 2010

4. ARTICLES
Maternal Vitamin D Status TABLE 1 Factors Associated With Infant 25(OH)D Status
Variable Total (N 376), Infant 25(OH)D Status P
Variables associated with maternal vi-
n (%)
tamin D status in bivariate analysis in- 20 ng/mL 20 ng/mL
(n 158), (n 218),
cluded season of birth, skin color, bed n (%) n (%)
rest during pregnancy, use of vitamins Maternal 25(OH)D, ng/mL .001
in the third trimester, drinking milk, 20 230 (63.4) 129 (56.1) 101 (43.9)
and time spent outside (Table 3). Use 20 133 (36.6) 24 (18.0) 109 (82.0)
Type of delivery .29
of prenatal vitamins in the first and Vaginal 284 (75.5) 115 (40.5) 169 (59.5)
second trimesters was not associated Cesarean 92 (24.5) 43 (46.7) 49 (53.3)
with vitamin D status, nor was con- Infant gender .24
Female 192 (51.1) 75 (39.1) 117 (60.9)
sumption of vitamin D–fortified orange
Male 184 (48.9) 83 (45.1) 101 (54.9)
juice, cod liver oil, fish, or fortified ce- Season of birth
reals (data not shown). Women who Summer 97 (25.8) 54 (55.7) 43 (44.3) .001
reported consuming eggs had a lower Fall 89 (23.7) 38 (42.7) 51 (57.3)
Winter 88 (23.4) 18 (20.5) 70 (79.5)
prevalence of vitamin D deficiency; Spring 102 (27.1) 48 (47.1) 54 (52.9)
however, egg consumption was less Maternal race/ethnicitya .002
common among US-born mothers and White 40 (10.6) 25 (62.5) 15 (37.5)
Hispanic 195 (51.9) 87 (44.6) 108 (55.4)
those who took vitamins less fre- Black 141 (37.5) 46 (32.6) 95 (67.4)
quently during the third trimester. Maternal skin color .02
Light 124 (33.0) 64 (51.6) 60 (48.4)
In multivariable logistic regression, Medium 129 (34.3) 51 (39.5) 78 (60.5)
the strongest predictor of vitamin D Dark 123 (32.7) 43 (35.0) 80 (65.0)
deficiency was giving birth in winter Maternal age, y .08
20 40 (10.6) 11 (27.5) 29 (72.5)
compared with summer (aOR: 4.78
20 to 30 221 (58.8) 92 (41.6) 129 (58.4)
[95% CI: 2.39 –9.55]). Other statistically 30–43 115 (30.6) 55 (47.8) 60 (52.2)
significant risk factors for maternal de- Maternal place of birth .33
ficiency were dark skin color (aOR: 2.74 Non-US born 244 (64.9) 107 (43.9) 137 (56.1)
US born 132 (35.1) 51 (38.6) 81 (61.4)
[95% CI: 1.53– 4.88]), fall birth (aOR: 2.73 Maternal education .54
[95% CI: 1.41–5.32]); and being born in Less than high school 156 (41.6) 63 (40.4) 93 (59.6)
the United States (aOR: 2.03 [95% CI: High school graduate/GED 111 (29.6) 44 (39.6) 67 (60.4)
More than high school 108 (28.8) 50 (46.3) 58 (53.7)
1.25–3.30]). Frequent vitamin use in the WIC recipient .18
third trimester, drinking milk during No 56 (14.9) 28 (50.0) 28 (50.0)
pregnancy, and spending at least 1 hour/ Yes 319 (85.1) 129 (40.4) 190 (59.6)
Maternal health insurance .92
day outside during the second or third Public 331 (88.7) 139 (42.0) 192 (58.0)
trimesters were protective against vita- Private, self-pay 42 (11.3) 18 (42.9) 24 (57.1)
min D deficiency (Table 4). Maternal bed rest during pregnancy .38
No 342 (91.4) 145 (42.4) 197 (57.6)
Yes 32 (8.6) 11 (34.4) 21 (65.6)
DISCUSSION Frequency of prenatal-vitamin use in first trimester .002
We found that, in a population of 5 d/wk 118 (32.2) 35 (29.7) 83 (70.3)
5 d/wk 248 (67.8) 119 (48.0) 129 (52.0)
largely Hispanic and black urban Frequency of prenatal-vitamin use in second .001
mothers and newborns, vitamin D de- trimester
ficiency was present in 58% of the new- 5 d/wk 84 (22.8) 21 (25.0) 63 (75.0)
5 d/wk 285 (77.2) 136 (47.7) 149 (52.3)
borns and 35.8% of the mothers, and Frequency of prenatal-vitamin use in third .001
severe deficiency was present in 38% trimester
of the infants and 23.1% of the moth- 5 d/wk 92 (24.9) 23 (25.0) 69 (75.0)
5 d/wk 278 (75.1) 134 (48.2) 144 (51.8)
ers. Maternal and infant status were
Frequency of prenatal-vitamin useb .001
closely associated, and winter birth as 5 d/wk in 1 trimester 84 (24.1) 20 (23.8) 64 (76.2)
well as a BMI of 35 were risk factors 5 d/wk during at least second and third 264 (75.9) 129 (48.9) 135 (51.1)
for mothers and infants. These results trimester
are consistent with findings from
other studies.21,22
PEDIATRICS Volume 125, Number 4, April 2010 643
Downloaded from www.pediatrics.org. Provided by Siu School Of Medicine on March 25, 2010

5. TABLE 1 Continued D. Although maternal vitamin D status
Variable Total (N 376), Infant 25(OH)D Status P was an important factor in predicting
n (%)
20 ng/mL 20 ng/mL infant vitamin D status, the correlation
(n 158), (n 218), between maternal and infant vitamin D
n (%) n (%)
deficiency is not 1-to-1: 18% of the in-
Mother ever drank milk during pregnancy .01
No 51 (13.6) 13 (25.5) 38 (74.5) fants born to vitamin D– deficient
Yes 324 (86.4) 144 (44.4) 180 (55.6) mothers were not deficient, whereas
Maternal employment status during pregnancy .37
44% of the infants born to vitamin
No/part-time 264 (70.6) 114 (43.2) 150 (56.8)
Full-time 110 (29.4) 42 (38.2) 68 (61.8) D–replete mothers were deficient. For
Maternal frequency of exposing arms or legs .001 example, a maternal BMI of 35 was
during third trimester
associated with increased risk of defi-
0–4 d/wk 197 (52.7) 66 (33.5) 131 (66.5)
5–7 d/wk 177 (47.3) 91 (51.4) 86 (48.6) ciency in the infants but not in the
Maternal smoking status .34 mothers, which suggests that obesity
Never/before pregnancy 333 (88.6) 137 (41.1) 196 (58.9)
During pregnancy 43 (11.4) 21 (48.8) 22 (51.2)
may possibly impede transference of
Any maternal alcohol during pregnancy .86 micronutrients during pregnancy; this
No 355 (94.7) 149 (42.0) 206 (58.0) is a novel association that is worth ad-
Yes 20 (5.3) 8 (40.0) 12 (60.0)
Any maternal medical problem .10
ditional examination.
No 281 (74.7) 125 (44.5) 156 (55.5) We were able to identify modifiable fac-
Yes 95 (25.3) 33 (34.7) 62 (65.3)
Maternal BMI .001
tors that could decrease risk. For in-
35 327 (87.0) 148 (45.3) 179 (54.7) fants, the strongest association ob-
35 49 (13.0) 10 (20.4) 39 (79.6) served was with maternal deficiency,
GED indicates General Educational Development.
a Maternal race and skin color were highly correlated (0.76). Therefore, only 1 of these variables was entered in multivariate
which implies that improved vitamin D
regression. Maternal race/ethnicity was chosen, because it explained a greater amount of variance than did skin color. status in mothers will have beneficial
b Frequency of vitamin use in the first trimester was correlated with the second (0.51) and third (0.38) trimesters, and
second-trimester vitamin use was correlated with third-trimester use (0.76). Therefore, a composite variable was created
effects on infants. Frequent prenatal-
to summarize use over all trimesters. vitamin use reduced the infant’s risk of
deficiency if the mother took them
TABLE 2 Results of Univariate and Multivariate Logistic Regression for Risks for Infant 25(OH)D throughout the second and third tri-
Deficiency at Term mesters. Third-trimester vitamin use
Variable Crude OR (95% CI) aOR (95% CI) alone may not be sufficient to reduce
Maternal 25(OH)D, ng/mL infant deficiency; rather, a longer
20 Referent Referent
duration of use may be needed for
20 5.80 (3.47–9.69) 5.28 (2.90–9.62)
Season of birth adequate stores. Although prenatal-
Summer Referent Referent vitamin use 5 times per week in the
Fall 1.69 (0.94–3.01) 1.66 (0.82–3.34)
Winter 4.88 (2.54–9.40) 3.86 (1.74–8.55)
third trimester was protective for
Spring 1.41 (0.81–2.47) 1.14 (0.58–2.21) women, 30% of the women who took
Maternal race/ethnicity the prenatal vitamin this often were
White Referent Referent
Hispanic 2.07 (1.03–4.16) 2.47 (1.05–5.80)
still vitamin D deficient at the time of
Black 3.44 (1.66–7.15) 3.36 (1.37–8.25) birth. One possible explanation for this
Frequency of prenatal-vitamin use result is that the prenatal vitamin does
5 d/wk in 1 trimester Referent Referent
5 d/wk during at least the second and third trimesters 0.33 (0.19–0.57) 0.30 (0.16–0.56) not contain enough vitamin D to affect
Maternal BMI an apparent chronic deficiency in
35 Referent Referent these women of childbearing age. For
35 3.22 (1.56–6.68) 2.78 (1.18–6.55)
the mothers, drinking milk and spending
All parameter estimates were adjusted for other covariates presented. Variables entered from univariate analysis that
were not statistically significant in multivariate analysis were maternal age, clothing, ever drinking milk during pregnancy, time outside were protective of vitamin D
and maternal medical problem. deficiency. Although a BMI of 35 was a
risk factor for both mothers and infants,
Although sun exposure was protective ferred to their infants, which suggests when compared with a BMI of 35, the
of vitamin D deficiency in the mothers, that other factors may influence risk of deficiency did not increase pro-
this protective effect was not trans- maternal-fetal transference of vitamin portionally with increased BMI and was
644 MEREWOOD et al
Downloaded from www.pediatrics.org. Provided by Siu School Of Medicine on March 25, 2010

6. ARTICLES
not evident at lower categorizations of TABLE 3 Factors Associated With Maternal 25(OH)D Status
BMI (data not shown). Variable Total, n (%) Maternal 25(OH)D Status P
The rates of deficiency in this study, 20 ng/mL 20 ng/mL
(n 278), (n 155),
especially in infants, were particularly n (%) n (%)
high. The use of 20 ng/mL as an indica- Season of birth .001
tor of deficiency is now standard in Summer 112 (25.9) 91 (81.3) 21 (18.7)
adult populations. We used it in our in- Fall 105 (24.3) 66 (62.9) 39 (37.1)
fant population, because it is the clini- Winter 99 (22.9) 45 (45.5) 54 (54.4)
Spring 117 (27.0) 76 (65.0) 41 (35.0)
cally accepted standard used in hospi- Maternal race/ethnicitya .001
tals in our area, including our own; Black 164 (37.9) 87 (53.0) 77 (47.0)
children with a 25(OH)D level of 20 White 46 (10.6) 35 (76.1) 11 (23.9)
Hispanic 223 (51.5) 156 (70.0 67 (30.0)
ng/mL are treated for deficiency. Re- Maternal skin color .001
cently, using 20 ng/mL as deficient in Light 140 (32.3) 107 (76.4) 33 (23.6)
a Boston population of infants and us- Medium 146 (33.7) 92 (63.0) 54 (37.0)
Dark 147 (34.0) 79 (53.7) 68 (46.3)
ing a comparable assay, Gordon et al14 Maternal age, y .999
found a 12% rate of deficiency. How- 20 53 (12.2) 34 (64.2) 19 (35.8)
ever, we suggest that our rates were 20 to 30 254 (58.7) 163 (64.2) 91 (35.8)
30–43 126 (29.1) 81 (64.3) 45 (35.7)
higher because we measured new-
Place of birth .001
born levels, whereas Gordon et al in- Non-US born 280 (64.7) 196 (70.0) 84 (30.0)
cluded infants and older infants up to 1 US born 153 (35.3) 82 (53.6) 71 (46.4)
year of age. Newborn status is related Education .008
Less than high school 179 (41.4) 129 (72.1) 50 (27.9)
to maternal status and likely changes High school graduate/GED 133 (30.8) 82 (61.7) 51 (38.3)
in the first year of life. More than high school 120 (27.8) 66 (55.0) 54 (45.0)
WIC recipient .25
A recent review by the American Acad- No 69 (16.0) 40 (58.0) 29 (42.0)
emy of Pediatrics suggested that ade- Yes 363 (84.0) 237 (65.3) 126 (34.7)
quate vitamin D status is important for Health insurance .15
Public 379 (88.1) 247 (65.2) 132 (34.8)
a range of developmental factors in in-
Private, self-pay 51 (11.9) 28 (54.9) 23 (45.1)
fants and children, including skeletal Bed rest in pregnancy .004
development, higher birth weight, and No 388 (90.0) 257 (66.2) 131 (33.8)
improved bone-mineral content and Yes 43 (10.0) 19 (44.2) 24 (55.8)
Frequency of prenatal-vitamin use in third trimester .001
bone mass in children at 9 years of Never or 1 d/wk 67 (15.7) 34 (50.7) 33 (49.3)
age.9 In addition, we recently re- 1–4 d/wk 41 (9.6) 20 (48.8) 21 (51.2)
ported an association between ma- 5 d/wk 319 (74.7) 221 (69.3) 98 (30.7)
Maternal calcium supplement .81
ternal vitamin D deficiency at birth Never 196 (64.1) 110 (35.9)
and an elevated risk of primary ce- Sometimes 53 (62.4) 32 (37.6)
sarean delivery.23 This is consistent Often 28 (68.3) 13 (31.7)
Ever drank milk during pregnancy .004
with data of the impact of vitamin D
No 57 (13.2) 27 (47.4) 30 (52.6)
deficiency on skeletal and smooth Yes 374 (86.8) 250 (66.8) 124 (33.2)
muscle function. Employment status during pregnancy .03
No/part-time 304 (70.7) 204 (67.1) 100 (32.9)
It should be noted that skin color was Full-time 126 (29.3) 71 (56.3) 55 (46.7)
a risk for deficiency in mothers, Average time spent outside during second or third .03
whereas race/ethnicity was a statisti- trimestersb
1 h/d in second and third trimesters 199 (48.0) 118 (59.3) 81 (40.7)
cally clearer risk for deficiency in in- 1 h/d in second or third trimester 216 (52.0) 150 (69.4) 66 (30.6)
fants. It is possible that the variable Frequency of exposing arms or legs during third .001
race/ethnicity includes unmeasured trimesterc
0–4 d/wk 228 (52.9) 128 (56.1) 100 (43.9)
confounders that have important bear-
5–7 d/wk 203 (47.1) 149 (73.4) 54 (26.6)
ing on infant vitamin D status but not Ever used sunscreen .61
on maternal vitamin D status. No 282 (65.4) 183 (64.9) 99 (35.1)
Yes 149 (34.6) 93 (62.4) 56 (37.6)
We acknowledge some limitations.
Data regarding variables in pregnancy
PEDIATRICS Volume 125, Number 4, April 2010 645
Downloaded from www.pediatrics.org. Provided by Siu School Of Medicine on March 25, 2010

7. TABLE 3 Continued size needed (376 vs 417) because of
Variable Total, n (%) Maternal 25(OH)D Status P difficulty obtaining blood samples.
20 ng/mL 20 ng/mL Baseline values of 25(OH)D were miss-
(n 278), (n 155), ing for some infants; however, this was
n (%) n (%)
a result of inevitable difficulties draw-
Smoking status .008
Never/before pregnancy 386 (89.2) 256 (66.3) 130 (33.7) ing blood on young infants and was not
During pregnancy 47 (10.8) 22 (46.8) 25 (53.2) associated with the mother’s 25(OH)D
Any alcohol during pregnancy .99 level, skin color, or season of birth (re-
No 407 (94.2) (64.1) 146 (35.9)
Yes 25 (5.8) 16 (64.0) 9 (36.0) sults not shown).
Maternal BMI .01 In addition, although children with a
35 380 (87.8) 252 (66.3) 128 (33.7)
35 53 (12.2) 26 (49.1) 27 (50.9)
25(OH)D level of 20 ng/mL are
GED indicates General Educational Development. treated with supplementation in the
a Maternal race and skin color were highly correlated (0.76). Therefore, only 1 of these variables was entered in multivariate
clinical setting, biomarkers of vitamin
regression. Maternal race/ethnicity was chosen, because it explained a greater amount of variance than did skin color.
b The average amount of time spent outside daily was highly correlated between the second and third trimesters (0.76), so D deficiency in young children particu-
they were combined into a single variable. larly are lacking. A need for research
c Frequency of exposing arms and legs during the first and second trimesters differed significantly according to the
mother’s 25(OH)D status. However, the direction of association was clearly confounded by season (ie, results were not in
to determine validity of commonly ac-
keeping with the pattern seen in the third trimester of decreasing prevalence of vitamin D deficiency with increasing time cepted measures of deficiency has
spent outside).
been identified.21,24
TABLE 4 Results of Univariate and Multivariate Logistic Regression for Risks for Maternal 25(OH)D CONCLUSIONS
Deficiency at Term
Variable Crude OR (95% CI) aOR (95% CI)
We found that more than half of the
Season of birth
infants and approximately one third of
Summer Referent Referent the mothers who gave birth in this Bos-
Fall 2.56 (1.38–4.75) 2.73 (1.41–5.32) ton study were vitamin D deficient at
Winter 5.20 (2.80–9.64) 4.78 (2.39–9.55)
Spring 2.34 (1.27–4.29) 1.87 (0.96–3.65)
the time of delivery. Although prenatal-
Maternal skin color vitamin use was protective for both
Light Referent Referent infants and mothers, considerable
Medium 1.90 (1.14–3.18) 2.46 (1.36–4.42)
Dark 2.79 (1.68–4.63) 2.74 (1.53–4.88)
proportions of infants and mothers re-
Maternal place of birth mained deficient even when prenatal
Non-US born Referent Referent vitamins were taken regularly. It is
US born 2.02 (1.34–3.04) 2.03 (1.25–3.30)
time to rethink our approach to ensur-
Frequency of prenatal-vitamin use in third trimester
Never or 1 d/wk Referent Referent ing vitamin D sufficiency in newborns
1–4 d/wk 1.08 (0.50–2.35) 0.79 (0.32–1.91) and their mothers.
5 d/wk 0.46 (0.27–0.78) 0.37 (0.20–0.69)
Ever drank milk during pregnancy
No Referent Referent ACKNOWLEDGMENTS
Yes 0.45 (0.25–0.78) 0.45 (0.24–0.87)
Time spent outside in second and third trimester
This work was funded by Health Re-
1 h/d second and third trimester Referent Referent sources and Services Administration/
1 h/d second or third trimester 0.64 (0.43–0.96) 0.55 (0.34–0.87) Bureau of Maternal and Child Health
All parameter estimates were adjusted for the other covariates presented. Variables entered from univariate analysis that grant R40 MC03620-01-00 and US De-
were not statistically significant in multivariate analysis were maternal race (correlated with skin color r 0.66 ), bed rest
during pregnancy, smoking during pregnancy, educational attainment, BMI, and employment status. partment of Agriculture/Cooperative
State Research, Education, and Exten-
sion Service grant 2005-35200-15260
were collected retrospectively. We did min D levels at the time they were and supported by grant M01-RR00533
not have dietary intake of vitamin D questioned, knowledge of the study from the General Clinical Research
and calcium, only the frequency or topic may have influenced some re- Centers Program of the National Cen-
amount of consumption. Although sponses. The number of infants ana- ter for Research Resources, National
mothers were unaware of their vita- lyzed was 90% of the estimated sample Institutes of Health.
646 MEREWOOD et al
Downloaded from www.pediatrics.org. Provided by Siu School Of Medicine on March 25, 2010

8. ARTICLES
REFERENCES
1. Wharton B, Bishop N. Rickets. Lancet. 2003; diatrics. 2009;123(1):197]. Pediatrics. 2008; las and Synopsis of Clinical Dermatology:
362(9393):1389 –1400 122(5):1142–1152 Common and Serious Diseases. 4th ed.
2. Harrison HE. A tribute to the first lady of 10. Holick MF. Vitamin D deficiency. N Engl New York, NY: McGraw-Hill Professional
public health (Martha M. Eliot). V. The disap- J Med. 2007;357(3):266 –281 Publishing; 2000
pearance of rickets. Am J Public Health Na- 11. Greer FR. Vitamin D deficiency: it’s more than 18. Chen TC, Turner AK, Holick MF. A method for
tions Health. 1966;56(5):734 –737 rickets. J Pediatr. 2003;143(4):422– 423 the determination of the circulating con-
3. Holick MF. Resurrection of vitamin D defi- 12. Bodnar LM, Simhan HN, Powers RW, Frank centration of 25-hydroxyvitamin D. J Nutr
ciency and rickets. J Clin Invest. 2006; MP, Cooperstein E, Roberts JM. High preva- Biochem. 1990;1(6):320 –327
116(8):2062–2072 lence of vitamin D insufficiency in black and 19. Holick MF, Siris ES, Binkley N, et al. Preva-
4. Calvo MS, Whiting SJ. Prevalence of vitamin D white pregnant women residing in the lence of vitamin D inadequacy among post-
insufficiency in Canada and the United States: northern United States and their neonates. menopausal North American women receiv-
importance to health status and efficacy of J Nutr. 2007;137(2):447– 452 ing osteoporosis therapy. J Clin Endocrinol
current food fortification and dietary supple- 13. Lee JM, Smith JR, Philipp BL, Chen TC, Metab. 2005;90(6):3215–3224
ment use. Nutr Rev. 2003;61(3):107–113 Mathieu J, Holick MF. Vitamin D deficiency in 20. Hardin J, Hilbe J. Generalized Estimating
5. Hanley DA, Davison KS. Vitamin D insuffi- a healthy group of mothers and newborn Equations. Boca Raton, FL: Chapman and
ciency in North America. J Nutr. 2005; infants. Clin Pediatr (Phila). 2007;46(1): Hall; 2003
135(2):332–337 42– 44
21. Greer FR. 25-Hydroxyvitamin D: functional
6. Holick MF. Vitamin D: a millennium perspec- 14. Gordon CM, Feldman HA, Sinclair L, et al.
outcomes in infants and young children.
tive. J Cell Biochem. 2003;88(2):296 –307 Prevalence of vitamin D deficiency among
Am J Clin Nutr. 2008;88(2):529S–533S
7. Holick MF. The vitamin D epidemic and its healthy infants and toddlers. Arch Pediatr
Adolesc Med. 2008;162(6):505–512 22. Bowyer L, Catling-Paull C, Diamond T, Homer
health consequences. J Nutr. 2005;135(11):
C, Davis G, Craig ME. Vitamin D, PTH and cal-
2739S–2748S 15. Newhook LA, Sloka S, Grant M, Randell E,
Kovacs CS, Twells LK. Vitamin D insufficiency cium levels in pregnant women and their
8. Kumar J, Muntner P, Kaskel FJ, Hailpern SM,
common in newborns, children and preg- neonates. Clin Endocrinol (Oxf). 2009;70(3):
Melamed ML. Prevalence and associations of
nant women living in Newfoundland and La- 372–377
25-hydroxyvitamin D deficiency in US children:
NHANES 2001–2004. Pediatrics. 2009;124(3). brador, Canada. Matern Child Nutr. 2009; 23. Merewood A, Mehta SD, Chen TC, Bauchner
Available at: www.pediatrics.org/cgi/content/ 5(2):186 –191 H, Holick MF. Association between vitamin D
full/124/3/e362 16. Webb AR, Kline L, Holick MF. Influence of sea- deficiency and primary cesarean section.
9. Wagner CL, Greer FR; American Academy of son and latitude on the cutaneous synthesis J Clin Endocrinol Metab. 2009;94(3):
Pediatrics, Section on Breastfeeding; Amer- of vitamin D3: exposure to winter sunlight in 940 –945
ican Academy of Pediatrics, Committee on Boston and Edmonton will not promote vita- 24. Brannon PM, Yetley EA, Bailey RL, Picciano
Nutrition. Prevention of rickets and vitamin min D3 synthesis in human skin. J Clin Endo- MF. Summary of roundtable discussion on
D deficiency in infants, children, and adoles- crinol Metab. 1988;67(2):373–378 vitamin D research needs. Am J Clin Nutr.
cents [published correction appears in Pe- 17. Fitzpatrick TB, Wolf K, Johnson RA. Color At- 2008;88(2):587S–592S
PEDIATRICS Volume 125, Number 4, April 2010 647
Downloaded from www.pediatrics.org. Provided by Siu School Of Medicine on March 25, 2010