Recruitment and study participants
At Kätilöopisto Maternity Hospital, Helsinki, Finland, 987 families were recruited into the VIDI study between January 2013 and June 2014, after delivery during the mother’s hospital stay. According to the inclusion criteria, the mothers were of Caucasian origin without regular medication and with singleton pregnancy. Exclusion criteria for the newborns were: nasal continuous positive airway pressure treatment >one day, intravenous glucose infusion, seizures, duration of phototherapy >three days and need for nasogastric tube >one day. The infants were born between 37 + 0 and 42 + 0 weeks of gestation, and newborn’s birth weight was appropriate for gestational age (SD-score [SDS] between −2.0 and +2.0). Of the recruited eligible families, 29% (987/3408) agreed to participate in the VIDI study. For the present cross-sectional study, we included mothers who had a record from a community prenatal clinic or baseline questionnaire and both two maternal 25(OH)D measurements. Two infants with a congenital disease (Down syndrome and Rieger syndrome) were excluded. Thus, the total number of subjects was 723. Of the infants, 367 were girls and 356 boys. Number of subjects in each analysis is reported in Tables.
25(OH)D analyses
Maternal pregnancy serum samples for 25(OH)D measurements were collected at community prenatal clinics at gestational weeks 7 to 25 between June 2012 and November 2013 as part of the mothers’ normal follow-up [hereafter referred to as pregnancy 25(OH)D]. At birth, umbilical cord blood (UCB) for 25(OH)D measurement was obtained at gestational weeks 37 to 42 between January 2013 and May 2014 [hereafter referred to as UCB 25(OH)D]. Both pregnancy serum and UCB plasma 25(OH)D were analysed simultaneously using the IDS-iSYS fully automated immunoassay system with chemiluminescence detection (Immunodiagnostic Systems Ltd., Bolton, UK). Detailed information on 25(OH)D analysis has been previously reported [
5]. The quality and accuracy of the serum 25(OH)D analysis is validated on an ongoing basis by participation in the vitamin D External Quality Assessment Scheme (DEQAS, Charing Cross Hospital, London, UK).
Both 25(OH)D concentrations were corrected by applying a linear regression equation (Oct 2014 value (nmol/L) = [(early 2014 value) – 8.2] / 0.99) provided by the manufacturer because of methodological changes in the IDS-iSYS system between 2014 and 2016 (see Additional file
1). We re-analysed a subsample of 77 samples and verified the correction (adjusted R
2 = 0.922, SEE = 9.2 nmol/l).
We employed UCB 25(OH)D to reflect both the maternal vitamin D status at the end of pregnancy and the newborn’s vitamin D status at birth [
1]. We defined vitamin D deficiency as 25(OH)D < 50 nmol/L, and vitamin D sufficiency as 25(OH)D ≥ 50 nmol/L, since a concentration of ≥50 nmol/L is considered sufficient for bone health [
17]. Suboptimal vitamin D status was defined as 25(OH)D < 80 nmol/L, and optimal vitamin D status as 25(OH)D ≥ 80 nmol/L, as has been suggested based on calcium absorption studies [
18].
Maternal and newborn data
Maternal data were obtained from a self-administered baseline questionnaire, filled in after delivery, and from medical records. Maternal height (cm) and weight (kg) before pregnancy and parity were collected primarily from the prenatal maternity card or, if missing, from our baseline questionnaire. Gestation was determined by first trimester ultrasound examination. Maternal age was determined at delivery. Parity was categorised into nullipara, secundipara and multipara (>two deliveries). Prepregnancy body mass index (BMI) (kg/m2) was categorised into underweight (<18.5), normal weight (18.5–24.9), overweight (25.0–29.9) and obese (>30.0).
Prepregnancy weight and weight recorded in prenatal clinics were utilised to calculate gestational weight gain (GWG) (kg). We recorded GWG at first measurement, at approximately the 12th, 20th and 30th gestational weeks, and at last measurement. GWG was adjusted for consecutive gestational week. Concerning the analysis of total GWG, absolute values were used, and women who had their final weight recorded more than three weeks before the delivery were omitted (
n = 7). Total GWG was categorised into inadequate, adequate and excessive based on national recommendations by prepregnancy BMI: recommended GWG for underweight mothers was 12.5–18.0 kg, for normal weight 11.5–16.0 kg, for overweight 7.0–11.5 kg, and for obese 5.0–9.0 kg [
19].
Education level was graded from one (=comprehensive school/lower secondary education) to six (university degree/first or second stage of tertiary education). Education was re-categorised into ‘lower’ and ‘higher’ education (lower = lower or upper secondary or post-secondary non-tertiary education, higher = first or second stage of tertiary education), due to a low number of subjects in other education categories. Prepregnancy smoking status was assessed as number of cigarettes per day. Maternal use of supplements, specific brand names, dosing, and date of commencement were recorded. We calculated the average daily intake of vitamin D from supplementation during the last two months of pregnancy.
Birth size, including birth weight (kg), length (cm), and head circumference (cm), was measured by midwives according to standard procedure. These data and the duration of pregnancy were retrospectively collected from birth records. Birth size measures were transformed into SDS by using Finnish sex-specific normative data for fetal growth [
20]. Ponderal index was calculated (birth weight (kg) / birth length (m)
3) and standardised into sex-specific z-score.
Assessment of GDM
The diagnosis of GDM was based on a two-hour 75 g oral glucose tolerance test (OGTT). According to the national guidelines, GDM was diagnosed if the OGTT results exceeded cut-offs for one or more values: fasting plasma glucose ≥5.3 mmol/l, 1-h ≥ 10.0 mmol/l and 2-h ≥ 8.6 mmol/l [
19]. An OGTT was performed at gestational weeks 10 to 40 between October 2012 and March 2014, and the results were collected from prenatal maternity cards or the hospital laboratory database. In general, screening for GDM depends on a presence of risk factors according to national recommendations [
19] and based on these, OGTT was performed on 490 (54.5%) of the participating mothers. None of the pregnant women in our study received insulin therapy nor other regular medication, but mothers with GDM obtained dietary counselling at community prenatal clinics [
19].
Statistical analysis
The normality of the variables was visually inspected. Outliers (n = 18) of 25(OH)D concentrations were identified with Normal probability plot of residuals, Leverage and Cook’s Distance diagnostic tests, and omitted from the analyses. Season with four categories affected maternal 25(OH)D concentrations. Thus, season at pregnancy blood sampling and at birth was coded using dummy variables (with autumn as a reference) in ANCOVA and used as a covariate.
The data included partially missing information. Imputation of missing values for education (
n = 12) and parity (
n = 2) were conducted using the median value in subgroups by GDM status. Missing data on prepregnancy smoking as number of cigarettes daily were imputed as a median value (= zero) by GDM status according to smoking status (
n = 22). Imputation of missing values for GWG at 12 gestational weeks was conducted using a mean value of two consecutive measurements (
n = 12). Missing values of other variables were not imputed. Maternal characteristics in Table
1 are described only as un-imputed values.
Table 1
Maternal characteristics in GDM and non-GDM mothers. P-values refer to differences between the groups
Age at delivery (y) | 642 | 31.4 ± 4.3 | 81 | 32.7 ± 4.5 | 0.018 |
Level of educationa
| 632 | 5.0 ± 1.3 | 79 | 4.7 ± 1.4 | 0.051 |
Parity | 640 | 1.5 ± 0.7 | 81 | 1.5 ± 0.7 | 0.275 |
Prepregnancy smoking, number of cigarettes daily | 625 | 1.2 ± 3.7 | 76 | 2.0 ± 4.6 | 0.064 |
Alcohol consumption before pregnancy, portion/wk. | 625 | 1.9 ± 2.0 | 78 | 2.3 ± 3.5 | 0.755 |
Prepregnancy height (cm) | 642 | 166.3 ± 6.0 | 81 | 165.9 ± 5.4 | 0.398 |
Prepregnancy weight (kg) | 631 | 63.5 ± 10.2 | 81 | 72.1 ± 13.5 | <0.001 |
Prepregnancy BMI (kg/m2) | 639 | 23.0 ± 3.5 | 81 | 26.2 ± 4.8 | <0.001 |
Duration of gestation at OGTT (wk) | 323 | 26.3 ± 4.3 | 80 | 25.9 ± 5.9 | 0.774 |
Duration of gestation at pregnancy blood sampling (wk) | 642 | 11.3 ± 1.9 | 81 | 11.2 ± 2.2 | 0.089 |
Duration of gestation at delivery (wk) | 642 | 40.2 ± 1.1 | 81 | 40.1 ± 1.1 | 0.410 |
Pregnancy 25(OH)D (nmol/L) | 642 | 81.9 ± 19.5 | 81 | 80.0 ± 21.2 | 0.417 |
UCB 25(OH)D (nmol/L) | 642 | 80.1 ± 20.0 | 81 | 78.4 ± 18.8 | 0.448 |
Supplemental vitamin D intake during pregnancy (μg/d) | 621 | 15.5 ± 16.6 | 76 | 13.5 ± 10.8 | 0.162 |
Cumulative gestational weight gain atb (kg) | 0.093 |
first measurement | 580 | 1.7 ± 0.5 | 74 | 1.6 ± 0.5 | |
12th gestational week | 580 | 3.8 ± 1.0 | 74 | 3.6 ± 1.1 | |
20th gestational week | 580 | 6.3 ± 0.8 | 74 | 6.2 ± 0.8 | |
30th gestational week | 580 | 9.7 ± 0.5 | 74 | 9.7 ± 0.4 | |
last measurement | 580 | 13.8 ± 0.4 | 74 | 13.7 ± 0.3 | |
Independent sample t-tests, Mann-Whitney U-tests, repeated measures ANOVA or the Pearson Chi-Square test, when appropriate, were applied to compare maternal characteristics between GDM and non-GDM mothers. The difference in 25(OH)D between GDM and non-GDM mothers was investigated with ANCOVA adjusted for season, maternal age, education and prepregnancy BMI. Association between pregnancy 25(OH)D and OGTT results/birth weight, and between UCB 25(OH)D and head circumference at birth were tested with univariate linear regression. Prevalence of vitamin D deficiency in non-GDM and GDM mothers were tested with Fisher’s Exact test.
Newborn birth size was investigated in categories of maternal prepregnancy BMI (underweight, normal weight, overweight, obese), GWG (inadequate, adequate, excessive), prepregnancy smoking status (yes, no), maternal education (higher, lower), parity (nullipara, secundipara, multipara), GDM status (yes, no), vitamin D status (suboptimal, optimal) in pregnancy and in UCB, with ANCOVA with Bonferroni correction when applicable, and adjusted for maternal height. Changing covariates appearing in the ANCOVA models were: prepregnancy BMI, GWG at last measurement, smoking, education, parity, GDM and 25(OH)D concentrations. Using both 25(OH)D concentrations as covariates in the GWG analysis induced a multicollinearity problem based on Cook’s Distance and Levene’s test, but excluding these covariates from the model did not change the results.
Results are shown as means or adjusted means with SD or SEM. The means and medians were similar in both vitamin D concentrations. Associations were considered significant at P < 0.05. All statistical analyses were conducted using the IBM SPSS program for Windows, version 22 (IBM, Chicago, IL, USA).