Vitamin D insufficiency is common in pregnancy. In a study of pregnant women in Southampton, United Kingdom, during 1990–1991, the prevalence of vitamin D insufficiency was 31% among white Caucasian women1; an even greater frequency has been reported among Asian women.2 In the United Kingdom, vitamin D supplementation is recommended in pregnancy,3, 4 and since 2007, women in receipt of certain benefits have been entitled to free vitamin D supplements through the Healthy Start program. In the United States, the Institute of Medicine (IOM) concluded in 1997 that there was no additional need to increase the vitamin D age-related adequate intake during pregnancy above that required for nonpregnant women5; however, during 2008, the IOM initiated a study to review Dietary Reference Intakes for vitamin D, with a report expected for release in 2010. The vitamin D axis is known to influence the acquisition of bone mineral in utero, and changes in women's calcium homeostasis during pregnancy facilitate calcium supply for bone mineralization in the rapidly growing fetal skeleton.6 Mother-offspring cohort studies have shown that maternal vitamin D insufficiency has a detrimental effect on bone mineral accrual by the fetus, leading to reduced bone mass at birth and in childhood.1, 7
There is, however, little information on the precise consequences of maternal vitamin D insufficiency on the morphology and growth of long bones or whether its influence is apparent in early or late pregnancy. Answering these questions has been hindered by the limitations of ultrasound assessment of fetal bone and the need to avoid exposure to ionizing radiation during pregnancy. The development of high-resolution 3D ultrasound (3DUS) techniques in recent years enables intrauterine bone evaluation, and we have examined the use of 3DUS to look for splaying of the distal femoral metaphyses, such as is seen commonly in childhood rickets. In this study, we address the relationship between low maternal vitamin D status and fetal femur development using this novel approach.
Materials and Methods
The Southampton Women's Survey (SWS)8 is a prospective cohort study that recruited nonpregnant women aged 20 to 34 years in Southampton. Its purpose is to assess how maternal nutrition and lifestyle before and during pregnancy influence development of the offspring. All women were interviewed at home about their lifestyle and social circumstances, including receipt of social security benefits. Height, weight, and skinfold thickness measurements were made. The women attended for ultrasound scans at 11, 19, and 34 weeks' gestation. For this study, we recruited all 525 SWS women with singleton pregnancies who had a 3DUS scan at 19 weeks' gestation between October 2002 and June 2005. Four fetuses had congenital abnormalities, and one was stillborn; one mother withdrew because of severe illness, and two withdrew for other reasons. A total of 491 of the remainder had a further 3DUS at 34 weeks' gestation. At the time of the 34-week scan, a research nurse obtained a venous blood sample; an aliquot of maternal serum frozen at −80°C was available for measurement of maternal vitamin D status for 424 pregnancies (81% of the original 525 recruited). Serum 25-hydroxyvitamin D concentrations were analyzed by radioimmunoassay (Diasorin, Stillwater, Minnesota, USA). This assay measures both vitamin D2 and vitamin D3. The assay met the requirements of the UK National External Quality Assurance Scheme (NEQAS), and intraassay and interassay coefficients of variation were less than 10%. We categorized maternal vitamin D status according to 25-hydroxyvitamin D concentration as follows: ≤25 nmol/L “deficient,” 25 to 50 nmol/L “insufficient,” 50 to 70 nmol/L “borderline,” and >70 nmol/L “sufficient,” as used by the UK National Diet and Nutrition Survey 2004.9
At 19 and 34 weeks' gestation, we performed 3DUS of the fetal thigh using a KretzGE Voluson 730 3DUS system fitted with a curved-array transducer with a frequency range between 4 and 8 MHz and a controllable scan angle between 15 and 75 degrees. Thigh volumes were stored by a single operator (PM), who then used the multiplanar imaging mode to aid identification of structures and orientate anatomic slices. We took a standard linear measure of femur length from the stored scan10 and derived a novel measurement, only possible using 3D ultrasound, of the cross-sectional area of the distal femoral metaphysis (Fig. 1). The distal femur was chosen because it exhibits more evidence of splaying than the proximal end or the midshaft cross-sectional area. Each measurement was performed in triplicate, and mean values were used in subsequent analyses. Precision of the measurements was assessed by replicate examinations in 50 pregnancies at both 19 and 34 weeks. The coefficient of variation for triplicate measurements of femur length was 0.6% at 19 weeks and 0.4% at 34 weeks. Corresponding coefficients of variation for measurements of femoral metaphyseal cross-sectional area were 4.4% at 19 weeks and 3.2% at 34 weeks; these values are similar to those for the reproducibility of measurements obtained in other studies using 3DUS.11 To assess femoral shape, we derived a femoral splaying index as distal femoral metaphyseal cross-sectional area (cm2)/femur length (cm) (see Fig. 1). The index was devised for this study as a means of assessing the degree of metaphyseal splaying in relation to femur length. The duration of gestation was calculated from menstrual information or, when this was uncertain or discrepant with ultrasound assessments, from fetal anthropometry in early pregnancy.
We used linear regression to relate maternal vitamin D status to femoral size and splaying index, together with Pearson's correlation coefficients to show the strength of observed associations and their statistical significance. Where necessary, we transformed the data to satisfy statistical assumptions of normality. Statistical analysis was performed using Stata Statistical Software, Release 10, 2008 (Stata Corporation, College Station, TX, USA).
The study had full approval from the Southampton and Southwest Hampshire Local Research Ethics Committee, and all participants gave written informed consent.
A total of 424 mother-baby pairs were included in the study, of which 215 (50.7%) of the offspring were boys. The mean (SD) age of the mothers at delivery was 31.3 years (3.6); mean maternal height was 163.9 cm, and 59.4% were in their first pregnancy. At prepregnancy recruitment, 13.7% were in receipt of benefits, and 26.4% were smokers; 13.6% were still smoking in late pregnancy (Table 1). Mean [95% confidence interval (CI)] fetal femur length increased from 2.92 (2.90–2.94) cm to 6.36 (6.33–6.39) cm between 19 and 34 weeks' gestation; corresponding geometric mean values for distal femoral cross-sectional area measurements were 0.222 (0.217–0.226) cm2 and 0.765 (0.748–0.783) cm2, a more than threefold increase over this 15-week period. The geometric mean femoral splaying index rose from 0.076 (0.075–0.078) cm2/cm at 19 weeks' gestation to 0.120 (0.118–0.123) cm2/cm at 34 weeks. Fetal femur length, distal cross-sectional area, and splaying index all were similar in boys and girls at both 19 and 34 weeks' gestation.
Table 1. Characteristics of the 424 Mothers
Age at delivery (years)
Prepregnancy sum of skin fold thickness (mm)
Prepregnancy body mass index (BMI) (kg/m2)
34-Week vitamin D concentration (nmol/L)
Smoking status: Smokers (prepregnancy)
Smokers (early pregnancy)
Smokers (late pregnancy)
Parity: nulliparous (first child)
In receipt of benefits
Ethnic origin: white
The median maternal serum 25-hydroxyvitamin D concentration at 34 weeks' gestation was 61 nmol/L (range 8 to 180 nmol/L, interquartile range 41 to 85 nmol/L). Vitamin D concentrations above 70 nmol/L were seen in 171 women (40.3%); a further 98 women (23.1%) had values between 50 and 70 nmol/L (borderline); 130 (30.7%) had values between 25 and 50 nmol/L (insufficient); the remaining 25 (5.9%) had values less than 25 nmol/L (deficient). Thus 36.6% were either insufficient or deficient (Fig. 2). Compared with other women, those in receipt of benefits were more likely to be vitamin D deficient (12.1% versus 4.9%, p = .03) and marginally more likely to be vitamin D insufficient (32.8% versus 30.3). Of the vitamin D–deficient women, 72.0% were not in receipt of benefits, as were 85.4% of the vitamin D–insufficient women.
The panel of histograms on the left side of Fig. 3 shows the associations between maternal serum 25-hydroxyvitamin vitamin D concentration and fetal femur length, distal metaphyseal femoral cross-sectional area, and splaying index at 19 weeks' gestation. Maternal 25-hydroxyvitamin D concentration was not related to 19-week fetal femur length (r = 0.00, 95% CI −0.09 to 0.10), but a lower maternal 25-hydroxyvitamin D concentration was strongly associated with a greater distal metaphyseal cross-sectional area (r = −0.16, 95% CI −0.25 to −0.06) and a higher femoral splaying index (r = −0.17, 95% CI −0.26 to −0.07). Compared with fetuses whose mothers were vitamin D replete, metaphyseal cross-sectional area was 5% greater in those whose mothers were vitamin D insufficient and 14% greater in those whose mothers were vitamin D deficient. Likewise, there was a similar trend in femoral shape, as assessed by the femoral splaying index across the categories of 25-hydroxyvitamin D concentration. When the data were grouped according to 25-hydroxyvitamin D status, the geometric mean femoral splaying index at 19 weeks' gestation was 0.074 (95% CI 0.072–0.076) in fetuses whose mothers had 25-hydroxyvitamin vitamin D concentrations greater than 50 nmol/L; in fetuses whose mothers who were vitamin D insufficient and deficient, splaying index values were 0.078 (95% CI 0.075–0.081) and 0.084 (95% CI 0.076–0.093), respectively. When analyzed by gender, the associations with metaphyseal femoral cross-sectional area and splaying index at 19 weeks were similar in boys and girls. Taking account of the mother's height, prepregnant weight, sum of skinfold thicknesses, smoking, parity, and age had little effect on the association between the mother's vitamin D status and metaphyseal femoral cross-sectional area and femoral splaying index (data not shown). Likewise, season at measurement was not related to bone parameters, and taking season into account had little effect on the associations described earlier.
The panel of histograms on the right side of Fig. 3 demonstrates the association between maternal 25-hydroxyvitamin D concentration and femur length, distal metaphyseal cross-sectional area, and splaying index at 34 weeks' gestation. There was no association between maternal vitamin D concentration and fetal femur length (r = 0.07, 95% CI −0.03 to 0.16); lower maternal vitamin D concentration was, however, associated with greater distal metaphyseal cross-sectional area (r = −0.10, 95% CI −0.20 to 0.00) and a higher femoral splaying index (r = −0.11, 95% CI −0.21 to −0.01). These relationships were similar to those observed at 19 weeks' gestation but slightly weaker.
Our data suggest that lower maternal 25-hydroxyvitamin D status during pregnancy is associated with splaying of the distal metaphysis of the fetal femur. The association we found was already apparent at 19 weeks' gestation and still demonstrable at 34 weeks. We have used 3DUS to evaluate the size and shape of the fetal femur at 19 and 34 weeks' gestation and related this to maternal 25-hydroxyvitamin D concentrations in late pregnancy. Although previous studies of fetal femoral development have focused on femur length, we have derived a reproducible technique for measurement of fetal femoral distal metaphyseal cross-sectional area. The distal femoral metaphyseal splaying we have demonstrated in fetuses of mothers with low vitamin D concentrations is directly analogous to that seen in childhood rickets, and fetal femoral morphology perhaps could be a useful predictor of the risk of childhood rickets. This condition was once thought to have been almost eradicated in Western populations but has again become a public health problem.12, 13 Several reports have been published describing recent cases of infantile rickets.14 The reemergence of this metabolic bone disorder is thought to be due to an epidemic of vitamin D deficiency in mothers and children.15 Neonatal vitamin D stores are completely reliant on maternal supply; thus poor maternal vitamin D status during pregnancy is a major risk factor for neonatal rickets.16 In utero or early life, vitamin D deficiency also has been linked to an increased risk of several disorders, including neonatal craniotabes,17 preterm birth,18 type I diabetes,19 and schizophrenia.20
Strengths and limitations
Strengths of this study were the strong biologic basis for the association that we have demonstrated and the clear a priori hypothesis that we examined. Our study had several potential weaknesses. We solely studied a subset of SWS mothers; however, we have previously demonstrated that participants in this study were of similar age, socioeconomic status, and nutritional status to those who did not participate.7, 8 Second, we measured only total serum concentrations of 25-hydroxyvitamin D; pregnancy is associated with an increase in vitamin D–binding protein, as well as higher 1,25-dihydroxyvitamin D concentrations.21 Third, we measured only maternal vitamin D status in late pregnancy because we thought it more likely that maternal vitamin D insufficiency would influence femoral morphology in late gestation. The demonstration that low maternal vitamin D status is associated with altered femoral morphology as early as 19 weeks' gestation is all the more surprising given the measurement of vitamin D in later gestation. However, in a previous study in Southampton,1 we found that taking account of the month of sample collection, a 1 SD (standard deviation) difference in late-pregnancy maternal 25-hydroxyvitamin D concentration was associated with a 0.33 SD difference in early-pregnancy 25-hydroxyvitamin D concentration (unpublished). Fourth, our study relied on 3DUS to evaluate femoral length and metaphyseal cross-sectional area. Only the first of these measurements has been validated in vitro, but it is difficult to see how systematic differences in the assessment of femoral distal cross-sectional area would have spuriously revealed an association with maternal vitamin D status. Despite adjustment for confounders in our analyses, we cannot exclude the possibility of residual confounding, and causality cannot be assumed from these observational data.
Implications for practice
Evidence that lower prenatal vitamin D status alters fetal skeletal development with persisting effects on bone health has come from a previous Southampton cohort in which a lower maternal concentration of 25-hydroxyvitamin D in late pregnancy was associated with lower whole-body and lumbar spine bone mineral content in the children at aged 9 years, with a suggestion of a threshold effect at 35 nmol/L.1 Both the estimated exposure to ultraviolet B radiation during late pregnancy and the use of vitamin D supplements predicted maternal 25-hydroxyvitamin D concentrations and childhood bone mass. In the current study, we have demonstrated that changes in the distal femoral metaphysis were apparent as early as 19 weeks' gestation. This has important implications for interventions to improve maternal vitamin D status because it supports existing guidelines for the recommendation of vitamin D supplements in pregnancy but suggests that such measures should begin in early gestation. The test of these findings would be through a randomized, controlled trial of vitamin D supplementation in early pregnancy, and we have embarked on such a trial.
Our data reinforce guidance from the UK National Institute for Health and Clinical Excellence that health professionals should take particular care to check that women at greatest risk of vitamin D deficiency are following advice to take a vitamin D supplement during pregnancy and while breast-feeding.22 The women we studied were pregnant during 2002–2005; more recently, pregnant and lactating women in the United Kingdom in receipt of certain benefits have been entitled to free vitamin D supplements through the Healthy Start program. While women receiving benefits were more likely to be vitamin D deficient in our study, most of the deficient women would not have been eligible for free supplements. Our data support the provision of free vitamin D supplements within schemes such as Healthy Start, but they also highlight a need for wider use of vitamin D supplements, as currently recommended in some countries.22
Later life implications
Much evidence now suggests that environmental influences that alter bone development during intrauterine and early postnatal life may increase the risk of osteoporotic fracture in adulthood.23 Longitudinal studies in Finland have shown that poor early growth is associated with higher rates of hip fracture.24 Studies in the United States, the United Kingdom, Sweden, and Australia have shown that low birthweight and weight at 1 year are associated with low bone mass in adult life, as assessed by dual-energy X-ray absorptiometry.25–27 Early-life metaphyseal splaying could have implications for later degenerative joint disease, but as yet, this issue has not been studied. Preliminary data suggest important interactions between birth weight and genes that predict adult bone mass, including the vitamin D receptor gene.28 The relationships between early development and later bone health are independent of known environmental risk factors for osteoporosis, including smoking, alcohol consumption, lack of exercise, and low calcium intake.
Conclusion and future studies
In summary, our study shows that the vitamin D status of mothers during pregnancy is correlated with the morphology of the developing fetal femur. Changes in the distal femoral metaphysis are already apparent at 19 weeks' gestation. Since ultrasound is likely to play an increasing part in antenatal fetal assessment, and since the prevalence of vitamin D insufficiency among pregnant women is known to be high, there is a need for further evaluation of this methodology for the identification of fetuses for whom the skeletal consequences of maternal vitamin D insufficiency might be amenable to correction.
All authors have no conflicts of interest.
We thank the mothers who gave us their time, the midwifery staff for their assistance, and the SWS research team for their assistance. Pamela Mahon conceived and undertook the ultrasound scans and measurements, supervised by Cyrus Cooper and Keith Godfrey; Hazel Inskip, Keith Godfrey, and Sian Robinson initiated the SWS; Sarah Crozier led the statistical analysis; Nicholas Harvey, Nigel Arden, and Rama Swaminathan developed the vitamin D aspects of the study. All authors contributed to analysis and preparation of the manuscript. Keith Godfrey is guarantor for the study.
This research was supported by grants from the Medical Research Council, Arthritis Research Campaign, Dunhill Medical Trust, International Osteoporosis Foundation, and National Osteoporosis Society; a British Medical Ultrasound Society Pump Priming Grant was awarded in 2006. The funding agencies had no role in the conduct or reporting of this research.