Birth weight has been considered a proxy for fetal growth and development and is associated with future risk of adult metabolic diseases, particularly in low birth weight(Reference Barker1). It is well known that small for gestational age (SGA) newborns have an increased risk of neonatal and infant mortality, short stature, neurocognitive impairment and other outcomes, including chronic disease in adulthood(Reference Clayton, Cianfarani and Czernichow2–Reference Albertsson-Wikland, Boguszewski and Karlberg6).
Vitamin D (VD) is an essential hormone for Ca homoeostasis and bone mineralisation of the developing fetus(Reference Thorne-Lyman and Fawzi7). Previous observational studies have suggested that lower maternal 25-hydroxyvitamin D (25(OH)D) concentrations were associated with lower birth weight(Reference Eckhardt, Gernand and Roth8,Reference Miliku, Vinkhuyzen and Blanken9) , others have not found such an association(Reference Ong, Quah and Tint10,Reference Rodriguez, García-Esteban and Basterretxea11) . Moreover, a recent meta-analysis of thirteen randomised controlled trials indicated that maternal VD supplementation significantly increased birth weight and attenuated the risk of low birth weight and SGA compared with the no intervention group(Reference Maugeri, Barchitta and Blanco12). Despite these findings, a limited number of randomised controlled trials could not completely confirm the positive effect of VD supplementation during pregnancy on fetal growth(Reference Hossain, Kanani and Ramzan13–Reference Hewison, Wagner and Hollis16). These conflicting results suggest that underlying factors relevant to this potential association need to be discovered.
Evidence from the results of the ‘Mamma & Bambino’ cohort and a meta-analysis suggested that decreased birth weight may be affected by vitamin D receptor gene polymorphisms and mutated alleles, but the specific effect still needs to be further corroborated(Reference Barchitta, Maugeri and La Rosa17). Another under-recognised biological factor that may contribute to this conflicting literature is parathyroid hormone (PTH). Indeed, as the key upstream hormonal regulator of VD, PTH stimulates the kidneys to produce 1,25-(OH)2D3 and enhances the tubular reabsorption of Ca. It also activates osteoblasts, which stimulate the transformation of pre-osteoclasts into mature osteoclasts(Reference Miliku, Vinkhuyzen and Blanken9,Reference DeLuca18,Reference Holick19) . Moreover, the 25(OH)D concentration above an individual threshold could provide maximal PTH suppression(Reference Saliba, Barnett and Rennert20). Thus, VD helps maintain appropriate serum Ca concentrations by regulating the release of PTH to enhance Ca absorption. It is possible that the combined serum status of VD and PTH during pregnancy is a better determinant of fetal growth.
Therefore, we hypothesised that elevated PTH may further increase the risk of adverse fetal development in mothers with lower maternal 25(OH)D concentrations. In this study, we aimed to estimate the role of PTH during pregnancy on the relationship between maternal vitamin D deficiency (VDD) and fetal growth restriction (FGR) in a prospective cohort study from China.
Methods
Study population
From March 2015 to December 2017, a total of 4768 consecutive healthy pregnancies in the second trimester (gestational age at 14–27 weeks) were recruited from three different antenatal clinics in Hefei (Hefei First People’s Hospital, Hefei Maternal and Child Care Hospital and the First Affiliated Hospital of Anhui Medical University), China. The exclusion criteria of the study participants were multiple gestation (n 113), assisted conception (n 108), drug or alcohol abuse (n 90) and residence outside of Hefei (n 251). We followed up the participants until delivery and excluded serious pregnancy complications (e.g. gestational hypertension and preeclampsia, moderate and severe eclampsia, acute or chronic liver or renal diseases and chronic diseases, n 276), abortion (n 85), stillbirth (n 51), birth defection (congenital anomalies, genetic abnormalities or skeletal dysplasia, n 143) and loss to follow-up (birth outcome information not obtained, n 244). Finally, we obtained the full data of 3407 mother–infant pairs (Fig. 1). The Ethics Committee of the Anhui Medical University (reference number 2015002; date of approval 22 April 2015) approved the study and all participants provided written informed consent before enrolment.
Flow diagram of study participants enrolment, follow-up and analysis. 25(OH)D, 25-hydroxyvitamin D; PTH, parathyroid hormone.
Scatter plot between parathyroid hormone (PTH) and 25-hydroxyvitamin D (25(OH)D) concentrations in the second trimester (n 1536).
Data collection
Face-to-face interviews at enrolment were used to obtain information about the socio-economic and demographic characteristics, health status and lifestyle relevant to VD status. Socio-economic and demographic characteristics include maternal age, gestational age at enrolment, maternal education level, place of residence, parity and household income level. Health status includes anthropometric measurements (height and weight) and medical history. Variables on lifestyle relevant to VD status include the frequency of consuming milk, eggs, Ca and vitamin D supplements. Complications of pregnancy were obtained from medical records. The pre-pregnancy BMI was calculated based on self-reported pre-pregnancy weight and measured height(Reference Engstrom, Paterson and Doherty21). The season of blood sample collection at enrolment was defined as two periods of the year: winter/spring (December–May) and summer/autumn (June–November). Gestational weight gain was calculated based on the pre-pregnancy weight and the weight recorded at delivery or the last prenatal visit before delivery.
The birth outcome indexes were obtained from medical records, and birth weight was measured by study nurses. The accuracy of the scales used to measure birth weight was checked using standard weights at the beginning of the study. Gestational age (in completed weeks) was calculated based on the history of the last menstrual period. SGA was defined based on the method developed by Mikolajczyk et al. (Reference Mikolajczyk, Zhang and Betran22) as a birth weight lower than the 10th percentile for the gestational age-specific value of the study cohort (n 3407).
Serum 25-hydoxyvitamin D, parathyroid hormone and calcium
At enrolment, fasting maternal venous blood samples were drawn from participants after 15 min of rest and were processed by study nurses. Serum samples were stored to −80°C freezers for long-term storage until analysis. Circulating 25(OH)D is the best biomarker of VD status because it represents VD obtained from both dietary sources and UV skin synthesis(Reference Michael and Holick23). The quantification of serum 25(OH)D concentration was measured using commercial radioimmunoassay kits (DiaSorin Stillwater). The intra-assay and inter-assay coefficients of variation were 8·8 and 11·1 %, respectively. Serum concentrations of intact PTH were determined using a direct chemiluminescence immunoassay (ADVIA Centaur XP; Siemens Healthcare Diagnostics), with intra- and inter-assay coefficients of variations of 4·2 and 11·9 %, respectively. The levels of ionised Ca in the serum sample were determined by flame atomic absorption spectrometry (using a Zeiss AAS-3 spectrometer with deuterium background correction). The accuracy of the method was verified using certified reference material (Hum Asy Control 2, Randox) and was 95 %.
Second-trimester 25(OH)D and Ca concentrations were measured in 3407 pregnancies. We randomly selected half of the total sample (n 1703) and assayed for PTH concentrations. However, 167 samples failed to detect PTH; therefore, we obtained 1536 serum PTH concentrations. Attrition analyses showed that the distributions of socio-economic and demographic characteristics, health status, lifestyle, birth outcomes and VD levels in 1536 pregnant women with PTH measurements did not differ from those of all participants.
Serum 25(OH)D concentrations were categorised into groups as severe VDD (<30 nmol/l), moderate VDD (≥30–<50 nmol/l) and non-VDD (≥50 nmol/l)(Reference Holick, Binkley and Bischoff-Ferrari24). In the absence of pregnancy-specific reference ranges, elevated PTH levels (≥16·7 ng/l) were defined by a concentration that exceeded the upper reference limit for 75th percentiles (P75). Non-elevated PTH levels were defined as concentrations within the upper reference limit for P75 (<16·7 ng/l).
Statistical analysis
The socio-economic and demographic characteristics and the laboratory statistics of 3407 mother–infant pairs are shown as means and standard deviations or as percentages. The associations of serum 25(OH)D and Ca supplement frequency with PTH were tested with linear regression or binary logistic analysis.
The individual relationships of serum 25(OH)D or PTH status with birth weight and SGA risk were evaluated by multivariate linear regression analysis and binary logistic regression after adjustment for confounders, including gestational age at enrolment, household income, parity, pre-pregnancy BMI, gestational diabetes, gestational weight gain, VD supplement, Ca supplement, milk supplement, sunshine exposure, blood sampling season, maternal anaemia and fetal sex. To identify the unique contribution of each factor of maternal 25(OH)D and PTH status, the other factor was controlled in multivariate linear regression models.
Multiple linear regression and binary logistic regression were accordingly used to calculate the association of the combined status of VD and PTH in the second trimester with birth weight and the risk of SGA, with adjustment for confounders. Similarly, the association of the combined status of VD and PTH with serum Ca and Ca supplement frequency was analysed by multiple linear regression and binary logistic regression. Multiple linear regression analysis generated β-coefficients and 95 % CI. Likewise, binary logistic regression analysis generated adjusted risk ratios (RR) and 95 % CI. Sensitivity analyses were performed to confirm the consistency of the results for different percentile cut-offs for elevated PTH. We performed all analyses using IBM SPSS version 21.0 software (IBM Corp.).
Results
Socio-economic, demographic and laboratory characteristics of the study population
A total of 3407 mother–infant pairs were included in the study. The average age of the study population was 29·2 (SD 4·4) years, and the gestational age upon entry into the study was 24·2 (SD 2·4) weeks. The mean concentrations of serum 25(OH)D and Ca were 39·9 (SD 16·3) nmol/l and 1·52 (SD 0·19) μmol/l, respectively. The percentages of the women who had severe, moderate VDD and non-VDD were 29·4, 46·9 and 23·7 %, respectively. The mean serum PTH concentration was 12·9 (SD 7·6) ng/l, and the threshold concentration of P75 was 16·7 ng/l. The percentage of less Ca supplementation (<3 times/week) was 25·1 %. The mean birth weight was 3402·4 (SD 397·6) g, and the gestational age at delivery was 39·1 (SD 1·2) weeks (Table 1).
Characteristics of the study population, enrolment through to birth (n 3407)
(Mean values and standard deviations; numbers and percentages)
GWG, gestational weight gain; PTH, parathyroid hormone; 25(OH)D, 25-hydroxyvitamin D; SGA, small for gestational age.
* n was 1536.
Associations of vitamin D and calcium supplement frequency with parathyroid hormone levels
Associations of VD and Ca supplement frequency with PTH levels during the second trimester of pregnancy are shown in online Supplementary Table S1. Serum PTH concentrations (β = 1·69 ng/l, 95 % CI 0·63, 2·74, P < 0·01) and the risk of elevated PTH (RR = 1·66, 95 % CI 1·20, 2·31, P < 0·01) were significantly increased in the severe VDD group compared with the non-VDD group. There was a weak but statistically significant inverse association (r −0·068, P = 0·008) between 25(OH)D and PTH concentration (Fig. 2). Similarly, PTH concentrations (β = 1·09 ng/l, 95 % CI 0·24, 1·94, P < 0·05) and the risk of elevated PTH (RR = 1·41, 95 % CI 1·08, 1·85, P < 0·05) were significantly increased in pregnancies with a lower frequency of Ca supplementation (<3 times/week) compared with the reference group (≥5 times/week).
Individual association of maternal vitamin D deficiency and elevated parathyroid hormone with fetal growth
The individual relationships of serum VD or PTH status with neonatal birth weight and the risk of SGA are presented in Table 2. In the multiple linear regression model, the neonatal birth weight in pregnancies with severe or moderate VDD was, respectively, −79·8 g (95 % CI −127·2, −32·5, P < 0·01) and −49·4 g (95 % CI −91·1, −7·8, P < 0·05) lower than that in non-VDD pregnancies after adjustments were made for potential confounders. Similarly, the neonatal birth weight of pregnancies with elevated PTH decreased by 41·8 g (95 % CI −80·3, −3·2, P < 0·05) compared with that of pregnancies with non-elevated PTH. However, the risk of SGA significantly increased only in pregnancies with severe VDD (adjusted RR = 2·24, 95 % CI 1·22, 4·10, P < 0·01) but not elevated PTH compared with the reference group.
Individual association of vitamin D deficiency or elevated parathyroid hormone (PTH) with fetal growth*
(Mean values and standard deviations; β-coefficients and 95 % confidence intervals; numbers and percentages; risk ratios (RR))
SGA, small for gestational age; 25(OH)D, 25-hydroxyvitamin D; Ref, reference.
* Adjusted for gestational age at delivery, gestational age at enrolment, maternal age at enrolment, household income, nulliparous, pre-pregnancy BMI, gestational diabetes, gestational weight gain, vitamin D supplement, Ca supplement, milk supplement, sunshine exposure, blood sampling season, sex, anaemia, vitamin D status (for PTH levels) or PTH status (for 25(OH)D status).
Neonatal birth weight: the influence of the combined status of maternal serum 25-hydroxyvitamin D and parathyroid hormone
Table 3 shows the role of PTH during the second trimester on the relationship between maternal VDD and neonatal birth weight. Compared with the reference group (non-VDD and non-elevated PTH levels), pregnancies with severe or moderate VDD were significantly associated with lower birth weight, regardless of PTH status. Neonatal birth weight progressively and significantly decreased (adjusted β = −20·8 g, 95 % CI −32·0, −9·5) across the different combined statuses of descending VD and ascending PTH. Infants whose mothers had severe VDD and elevated PTH had the lowest birth weight (adjusted β = −124·7 g, 95 % CI −194·6, −54·8, P < 0·001). Among pregnancies with severe VDD, the neonatal birth weight of pregnancies with non-elevated PTH significantly increased (adjusted β = 67·5 g, 95 % CI 1·2, 133·7) compared with pregnancies with elevated PTH. In addition, among pregnancies with severe or moderate VDD, the birth weight of pregnancies with non-elevated PTH significantly increased (adjusted β = 42·3 g, 95 % CI 17·2, 67·5) compared with pregnancies with elevated PTH.
Association of combined status of maternal serum 25-hydroxyvitamin D (25(OH)D) and parathyroid hormone (PTH) with infant birth weight*
(Mean values and standard deviations; β-coefficients and 95 % confidence intervals)
VD, vitamin D; Ref, reference.
* Adjusted for gestational age at delivery, gestational age at enrolment, maternal age at enrolment, household income, nulliparous, pre-pregnancy BMI, gestational diabetes, gestational weight gain, VD supplement, Ca supplement, milk supplement, sunshine exposure, blood sampling season, sex and anaemia.
The risk of small for gestational age: the influence of the combined status of serum 25-hydroxyvitamin D and parathyroid hormone
In the multiple regression model for the association of the combined status of serum 25(OH)D and PTH with the risk of SGA (Table 4), the risk of SGA (adjusted RR = 1·23, 95 % CI 1·07, 1·41) progressively and significantly increased across the different combined statuses of descending VD and ascending PTH. Pregnancies with severe VDD and elevated PTH had a significantly higher risk of SGA (adjusted RR = 3·36, 95 % CI 1·41, 8·03, P < 0·01). Among pregnancies with severe or moderate VDD, the SGA risk of pregnancies with non-elevated PTH significantly decreased (adjusted RR = 0·68, 95 % CI 0·25, 0·98) compared with that of pregnancies with elevated PTH.
Influence of combined status of maternal serum 25-hydroxyvitamin D (25(OH)D) and parathyroid hormone (PTH) on the risk of small for gestational age*
(Numbers and percentages; risk ratios (RR) and 95 % confidence intervals)
VD, vitamin D; Ref, reference.
* Adjusted for gestational age at delivery, gestational age at enrolment, maternal age at enrolment, household income, nulliparous, pre-pregnancy BMI, gestational diabetes, gestational weight gain, VD supplement, Ca supplement, milk supplement, sunshine exposure, blood sampling season, sex and anaemia.
Serum calcium metabolism and calcium supplement: the association of the combined status of serum 25-hydroxyvitamin D and parathyroid hormone
Serum Ca concentrations progressively and significantly increased (adjusted β = 0·007 μmol/l, 95 % CI 0·001, 0·014) across the different combined statuses of descending VD and ascending PTH. Pregnancies with severe VDD and elevated PTH had significantly higher serum Ca concentrations (adjusted β = 0·050 (95 % CI 0·011, 0·089) μmol/l, P < 0·05) compared with non-VDD and non-elevated PTH pregnancies. Consistently, the proportion of less Ca supplement frequency (<3 times/week) during pregnancy progressively and significantly increased (adjusted RR = 1·34, 95 % CI 1·23, 1·46) across the different combined statuses of descending VD and ascending PTH. Pregnancies with severe VDD and elevated PTH had a significantly higher prevalence (38·2 %) of lower Ca intake frequency (<3 times/week) compared with non-VDD and non-elevated PTH pregnancies (Table 5).
Associations of combined status of vitamin D (VD) and parathyroid hormone (PTH) with serum calcium metabolism and frequency of calcium supplementation during pregnancy*
(β-Coefficients and 95 % confidence intervals; numbers and percentages; risk ratios (RR))
Ref, reference.
* Adjusted for gestational age at delivery, gestational age at enrolment, maternal age at enrolment, household income, nulliparous, pre-pregnancy BMI, gestational diabetes, gestational weight gain, VD supplement, milk supplement, sunshine exposure, blood sampling season, sex and anaemia.
† Less frequency of Ca supplement: Ca supplement <3 times/week.
Sensitivity analyses: the consistency of analysis results for different percentile cut-offs for elevated parathyroid hormone
Sensitivity analyses were performed to confirm the consistency of the analysis results for different percentile cut-offs for elevated PTH. We found that the associations of the combined status of VDD and elevated PTH (≥P80) with birth weight, SGA, maternal Ca and Ca supplementation frequency were similar to the aforementioned analysis results in which elevated PTH was defined as PTH levels ≥75th percentiles. Pregnancies with severe VDD and elevated PTH (≥P80) had the lowest neonatal birth weight and the highest risk of SGA. Consistently, the highest serum Ca levels and the highest proportion of lower Ca supplementation were observed in pregnancies with severe VDD and elevated PTH (≥P80) (online Supplementary Tables S2 and S3).
Discussion
In this study, we evaluated the effect of the combined serum status of VD and PTH during pregnancy on FGR in a large sample and prospective birth cohort. The results showed that severe VDD in the second trimester could independently lead to decreased neonatal birth weight and an increased risk of SGA. Unlike VD, elevated PTH was significantly associated with decreased neonatal birth weight but not an increased risk of SGA. However, elevated PTH might further aggravate the risk of FGR for pregnant women with VDD, especially those with severe VDD. This result highlights the role of PTH on the association between maternal VD and fetal growth and implicates PTH as a potential factor that may be relevant to both conflicting studies to date regarding maternal VD and fetal growth and clinical trials of VD supplementation in pregnancy.
Several previous reports have suggested an association between VD insufficiency in pregnancy and FGR. A recent meta-analysis(Reference Aghajafari, Nagulesapillai and Ronksley25) of thirty-one studies demonstrated that pregnant women with low serum 25(OH)D levels had an increased risk of low birth weight and SGA infants. Conversely, in two cohort studies(Reference Ong, Quah and Tint10,Reference Hanieh, Ha and Simpson26) , there were no associations between maternal VD status and infant birth weight. Moreover, the U-shaped association between 25(OH)D and FGR has also been reported in several observational studies(Reference Bodnar, Catov and Zmuda27,Reference Zhu, Tong and Hu28) . In this context of conflicting findings, our study reinforces the independent association between maternal VD status during the second trimester and FGR in a cohort of 3407 mother–infant pairs and strictly controls a large number of confounding variables. However, it remains unclear whether VDD exerts a causal effect on FGR, which warrants further investigation in the context of a randomised trial.
It was a distinct advantage that our study simultaneously evaluated the individual and combined role of maternal PTH on FGR, which has been largely ignored in nearly all previous studies. There are few reports relevant to the association between PTH and fetal growth. A previous study(Reference Brunvand, Quigstad and Urdal29) found that maternal serum PTH at delivery was inversely related to the crown-heel length among thirty pregnant women. In contrast, an observational study(Reference Morley, Carlin and Pasco30) of 374 pregnancies demonstrated that maternal PTH concentrations at 28–32 weeks were weakly positively associated with neonatal birth weight. However, these results should be interpreted with caution because of the small study sample or the absence of adjustments for confounders. In this large sample study with strict control of confounders, elevated PTH was significantly associated with decreased neonatal birth weight but not an increased risk of SGA, which was partly supported by another recent large sample study(Reference Hemmingway, Kenny and Malvisi31). However, the most valuable finding in our study was the combined effect of PTH and VD during pregnancy on FGR.
To date, few studies have reported the combined effect of maternal VD and PTH on fetal development. As there is a lack of clarity with regard to PTH threshold levels in pregnancy, elevated PTH was usually defined according to a percentile cut-off. A status of VDD (<30 nmol/l) and elevated PTH was identified as functional VDD(Reference Scholl, Chen and Stein32,Reference Scholl, Chen and Stein33) . Scholl et al. (Reference Morley, Carlin and Pasco30) reported that mothers with functional VDD were 2- to 3-fold more likely to develop SGA and that neonatal birth weight was reduced by 102–128 g. Hemmingway et al. (Reference Hemmingway, Kenny and Malvisi31) observed in a large sample cohort study that the prevalence of SGA was highest (P < 0·05) in mothers with functional VDD; however, the SGA risk (adjusted RR 1·53; 95 % CI 0·80, 2·93) was attenuated in a fully adjusted model. A similar trend was observed in our study, such that infants whose mothers with functional VDD (severe VDD with elevated PTH) had the lowest birth weight (reduced by 127·7 g) and the highest SGA risk (increased by 3·19-fold). The gradual effects across the different combined statuses of descending VD and ascending PTH on FGR were also observed in our study. Although ethnic differences in the sample population between our study and the above two studies were evident, our study strongly identified the enormous adverse effect of functional VDD during pregnancy on fetal growth.
At present, the mechanism through which functional VDD during pregnancy was linked to FGR remains unclear. VD has well-defined classical functions related to Ca metabolism(Reference Liu and Hewison34,Reference Luk, Torrealday and Neal Perry35) . Ca metabolic stress, in which adverse effects of VDD are mediated through a functional impact on the Ca metabolic system, was considered as the critical mechanism by Scholl et al. (Reference Scholl, Chen and Stein33) and Hemmingway et al. (Reference Hemmingway, Kenny and Malvisi31). However, in their studies, they did not evaluate serum Ca or Ca supplementation. Uniquely, in our study, we simultaneously assessed Ca supplementation during pregnancy and maternal serum Ca concentrations. We found that compared with the reference group, Ca concentrations significantly increased in mothers with functional VDD (severe VDD with elevated PTH) but not in mothers with severe VDD and non-elevated PTH, which was considered a sign of stress to Ca metabolism of functional VDD(Reference Schnatz and Curry36). Consistently, the proportion of lower Ca supplement frequency (<3 times/week) during pregnancy is highest with functional VDD. Maternal Ca homoeostasis is associated with fetal Ca homoeostasis to support fetal development(37). Therefore, our finding suggested that elevated PTH levels induced by less Ca supplementation might further aggravate FGR for pregnancies with severe VDD through impaired maternal Ca metabolism homoeostasis.
Strengths and limitations
Our study was a prospective cohort study. The number of participants was relatively large, and a large number of confounding factors relevant to VD status and fetal development were strictly controlled. Our findings are strengthened by the consistency of the analysis results for different fetal outcome indexes (birth weight and SGA), the sensitivity analyses for different percentile cut-offs for elevated PTH and the evaluation of maternal Ca metabolism stress (Ca supplement and serum Ca). However, we should acknowledge that there are some limitations in our study. First, as a lack of clarity elevated PTH threshold concentrations in pregnancy, a percentile cut-off is used to define elevated PTH, which might lead to misclassification. Therefore, the clinical significance of elevated PTH is difficult to interpret. The harmonisation and standardisation of PTH analysis are required to define reference ranges for pregnancy in future research in this context. Second, we did not evaluate fetal Ca metabolism and other markers related to fetal growth, although the inclusion of maternal serum Ca concentrations may partly reflect fetal Ca metabolism status. The mechanism through which maternal functional VDD was linked to fetal growth still has not been illuminated.
Conclusions
In conclusion, maternal PTH may be a potential risk factor relevant to the association between maternal VD and fetal growth. Elevated PTH levels induced by less Ca supplementation might further aggravate fetal growth in pregnant women with severe VDD through impaired maternal Ca metabolism homoeostasis. Functional VDD in the second trimester, defined as VDD (<30 nmol/l) with elevated PTH (>P75), predicted the enormous adverse effect on fetal growth. Treating VDD in pregnancy (and not only Ca supplements) may contribute to optimal fetal growth. Evidence from our study may warrant further investigation to illuminate the clinical significance in the context of a randomised trial.
Acknowledgements
The authors appreciate the contributions of the staff of the Hefei First People’s Hospital, the Maternal and Child Care Hospital and the First Affiliated Hospital of Anhui Medical University, who contributed to providing the birth outcome data. We also acknowledge the selfless and dedicated pregnant women who donated biological samples.
This work was supported by the National Natural Science Foundation of China (81472991, 81872631); the Key Project of National Development Plan of China (SQ2018YFC100243); the Key Projects of Excellent Young Talents Fund in Universities of Anhui Province (gxyqZD2018025); the Project of Academic and Technical Leaders of Anhui Province (2017H141) and the Project of Applied Medicine in the Municipal Health Planning Commission of Hefei (hwk2018zd002). The funders had no role in the study design, data collection and analysis, decision to publish or preparation of the manuscript.
We were grateful to the participating pregnant women and neonatal units in Hefei, China. The authors’ responsibilities were as follows – P. Z. and R.-X. T. designed the study; D.-H. M., Y. Z., H.-L. H, J.-J. L., W.-J. Y., S.-S. M. and R.-X. T. conducted research and participated in data collection; D.-H. M. and Y. Z. conducted the statistical analysis and drafted the manuscript; S.-S. M. drafted the revision of the manuscript; P. Z. critically reviewed the manuscript for accuracy and intellectual content and all authors read and approved the final draft of the manuscript.
All of the authors have no conflicts of interest or compensation in relation to this article to disclose. The ICMJE disclosure forms are available to view as online supporting information.
Supplementary material
For supplementary material referred to in this article, please visit https://doi.org/10.1017/S0007114520001105
