Background
Preterm birth is the leading cause of morbidity and mortality in children < 5 years [
1,
2] and is associated with risk of health problems later in life [
3]. However, the identified risk factors for preterm birth are largely absent [
1,
4].
Maternal thyroid hormone is essential for placentation and fetal development [
5,
6]. Gestational thyroid dysfunction is then associated with the risk of adverse fetal outcomes, such as fetal growth restriction and brain underdevelopment [
7,
8]. Despite relationships that are linked between overt hypothyroidism and specific hyperthyroidism (e.g., Graves disease) with the risk of preterm birth [
5,
9], evidence for the association of milder thyroid dysfunction and autoantibody positivity with preterm birth remains controversial [
4,
10‐
13].
Previous studies in estimating these associations had certain limitations since a lack of a unified definition for thyroid dysfunction and a low statistical power [
4]. Recently, a meta-analysis of large individual participant data (47,045 pregnant women) concluded positive associations between mild thyroid dysfunctions and preterm birth [
4]. For example, the odds ratio (OR) was 1.29 (95% confidence interval [CI]: 1.01–1.64) for preterm birth among women with subclinical hypothyroidism relative to euthyroid women [
4]. However, the heterogeneity of the studies included in the meta-analysis might affect the robustness of the results and produce different findings.
Identification of the association of thyroid dysfunction and autoantibody positivity with the risk of preterm birth and whether these associations have specific characteristics (e.g., gestational age or population-specific) are crucial for guiding the necessity and specificity of clinical intervention for women with disorders. Thus, we conducted a cohort study to address these concerns.
Methods
Study design and data collection
A hospital-based cohort study of pregnant women who received prenatal examinations between April 2012 and August 2016 was conducted in 2018 [
14,
15]. All pregnant women who visited during the period were included in the cohort. At the first visit, maternal characteristics, such as maternal age, resident location, parity, and conception model, were investigated. Thyroid function was recommended at recruitment. They were then followed up until the 42nd day after the birth. During this period, pregnancy complications (e.g., gestational diabetes mellitus and preeclampsia) and outcomes (e.g., fetal gender, gestational weeks at delivery, weight, and congenital malformations) were identified [
14,
15]. Inclusion criterions of subjects included for the analysis in the present study referred to women with singleton pregnancy, who received at least a thyroid function test, and those with confirmed birth outcome. Exclusion criterion referred to pregnant women with multiple pregnancies, without a thyroid function test, and those lost to follow-up. Basic characteristics between women included for analysis and those who were lost to follow-up were imbalanced [
15]. Because treatment information was not collected for women with gestational thyroid diseases (e.g., subclinical hypothyroidism, overt hypothyroidism, hyperthyroidism, thyroiditis, thyroid nodules) in the present study, these women were further excluded from the analysis since some drugs (e.g., antithyroid drug) might affect the association between thyroid dysfunction and risk of preterm birth [
16]. The reference ranges for the TSH concentration (uIU/mL) in the hospital were 0.05–5.17, 0.39–5.22, and 0.60–6.84 for women who had the test between 1 and 12, 13–28, and 29–40 weeks of gestation; while 12.91–22.35, 9.81–17.26, and 9.12–15.71 for the FT4 concentration (pmol/L), respectively.
Outcomes
The primary outcome was preterm birth, which was defined as infants born before 37 weeks of gestation. Secondary outcome was very preterm birth (e.g., < 32 weeks of gestation).
Thyroid function test
Thyroid function, including production of TSH, free and total triiodothyronine (FT3, TT3), FT4, TT4, thyroperoxidase (TPO), TSH receptor (TR), and thyroglobulin (Tg) antibodies were measured with an electrochemiluminescence immunoassay (Roche Elecsys, Germany) at the first visit to the hospital (range from 4 to 39 weeks, with a median of 15.4 weeks of gestation) [
14]. The median gestational weeks for the test did not show a significant difference between women with and without a preterm birth baby (
P = 0.633). Since only 33.5% (13971) of the included women received the test twice and 14.2% (5929) had three measurements. These measurements were not included for analysis (e.g., repeated measurements) in the present study.
Gestational age-specific percentiles of the FT4 and TSH concentrations were used for the definition of thyroid function, including euthyroid (both FT4 and TSH within the range between 2.5th and 97.5th percentile), subclinical hypothyroidism (TSH > 97.5th percentile; 2.5th ≤ FT4 ≤ 97.5th percentile), overt hypothyroidism (TSH > 97.5th percentile; FT4 < 2.5th percentile), isolated hypothyroxinemia (2.5th ≤ TSH ≤ 97.5th percentile, FT4 < 2.5th percentile), isolated hyperthyroxinemia (2.5th ≤ TSH ≤ 97.5th percentile, FT4 > 97.5th percentile), subclinical hyperthyroidism (TSH < 2.5th percentile, 2.5th ≤ FT4 ≤ 97.5th percentile), and hyperthyroidism (TSH < 2.5th percentile, FT4 > 97.5th percentile). TPOAb, TgAb, and TRAb were regarded as positive if the concentrations were > 34 IU/mL, 115 IU/mL, and 1.75 IU/L, respectively [
14].
Covariates
Covariates included maternal age (< 25, 25–34, or ≥ 35 years), residence (local or nonlocal), parity (nulliparous or pluriparous), assisted conception (yes or no), fetal sex (male or female), preeclampsia (yes or no), and gestational diabetes mellitus (yes or no) [
14]. Preeclampsia is defined as the onset of hypertension and proteinuria after 20 weeks of gestation [
17]. Gestational diabetes was diagnosed based on a 75-g oral glucose tolerance test conducted between the 24th and 28th weeks of gestation [
18].
Data analysis
Prevalence (95% confidence interval, 95% CI) of the outcomes was estimated. Difference between women with and without a preterm birth baby was compared by Chi-squared/Fisher’s exact test (e.g., category variables) or Mann-Whitney U test (e.g., continuous variables). Odds ratio (OR) for the risk of preterm birth among women with thyroid dysfunction was calculated compared with euthyroid women. Adjusted ORs were estimated after controlling for the covariates.
The risk of outcomes in women positive for thyroid antibodies was also assessed relative to those negative for the marker. In these analyses, women without thyroid antibody tests were also included as an unknown antibody status group. Basic characteristics and outcomes (preterm and very preterm birth) were compared among women with and without antibody tests. We conducted sensitivity analyses restricting to those negative for thyroid antibodies. Preeclampsia is an obvious risk factor for preterm birth and women with a TSH concentration > the upper target of the reference might receive potential treatment. Therefore, we further performed sensitivity analyses limited to women without preeclampsia and those with a TSH concentration ≤ the upper targets to exclude these potential impacts. Stratification analyses were performed according to gestation age at the thyroid test (e.g., first, second, or third trimester).
The associations of TSH and FT4 concentrations with the risk of preterm birth were further probed. Outliers (mean ± 4 SD) of the TSH or FT4 concentrations were removed before the analyses. Interaction term analyses were performed to estimate the potential modification of gestational age at the thyroid test on these associations. Stratification analyses were conducted according to the thyroid testing time (e.g., ≤ or > 18 weeks of gestation) because that the fetal thyroid gland is not functionally mature until 18–20 weeks of gestation [
19] and that the gestational age specific (12–18 weeks of gestation) association between thyroxine concentration and risk of congenital heart defects was identified in our previous study [
15]. Gestational age-specific Z scores or percentiles of the FT4 and TSH were further used in assessing their association with the risk of preterm birth.
All analyses were conducted using Microsoft Office Excel, Stata (version 14.0, Stata Corp., College Station, TX, USA). A two-sided value of P < 0.05 was considered statistically significant.
Discussion
The results in the present study did not support that thyroid dysfunction and autoantibody positivity were significantly associated with a higher risk for preterm birth or very preterm birth. However, some specific thyroid function abnormalities, such as TPOAb positive in the first trimester and overt hypothyroidism restricted to women negative for TPOAb, were related to an increased risk of preterm birth. In addition, a modification of gestational age was identified for the association between FT4 concentration and risk of preterm birth, with a critical cut-off value of 18 weeks of gestation.
In comparison with overt hypothyroidism where there is clear evidence for adverse events, the association between mild thyroid dysfunction (such as isolated hypothyroxinemia and subclinical hypothyroidism) and pregnancy outcomes is unclear[
20], including preterm birth. It is important to determine whether such thyroid dysfunctions are risk factors for preterm birth since that they are much more frequent than overt thyroid disease [
4,
5] and that a clear relationship may provide a basis for possible intervention, such as treatment with levothyroxine.
In the present study, both isolated hypothyroxinemia and subclinical hypothyroidism did not show a significant relation to the risk of preterm birth. These findings were consistent with the pooled results of previous observational studies [
12,
13] and might partly explain why levothyroxine treatment did not significantly reduce the risk of preterm birth [
21,
22]. However, the finding is different from that in an individual participant meta-analysis [
4], in which cohort-specific percentiles were used to define thyroid dysfunction. Pregnancy had a profound impact on the thyroid gland and function [
23]. Physiological changes in maternal hormones (e.g., elevated FT4 and decreased TSH) then happen to meet the demands of both the mother and fetus. The change is apparent before 18 ~ 20 weeks of gestation when the thyroid gland of the fetus is not functionally mature [
5,
23]. Thus, thyroid dysfunction defined by gestational age-specific percentiles for TSH and FT4 concentrations was more reasonable than the cohort-specific percentile method. Such the method taken in that meta-analysis might induce misclassification bias and then affect the validity of their results [
4].
Overt thyroid dysfunction (e.g., overt hypothyroidism or hyperthyroidism) during pregnancy is a known risk factor for preterm birth [
5,
13]. The positive correlations between hyperthyroidism and preterm birth in the previous meta-analysis may be mainly contributed to the inclusion of a large Danish population cohort study, which accounts for a 98.7% weight of the pooled OR [
13,
24]. An obvious flaw of the definition of hyperthyroidism (based on thyroid function tested before, during, or after pregnancy) raised concerns about a potential classification bias of the disorder and then on its real relation to preterm birth since the prevalence of hyperthyroidism in pregnancy varies depending on the timing of screening [
9,
24]. In the present study, we identified the association of overt hypothyroidism with the risk of preterm birth. However, the association was restricted to those negative for TPOAb. Similarly, TPOAb positive only exhibited its impact on preterm birth in the first trimester.
FT4 was deemed a better marker of thyroid function than TSH in pregnancy [
5,
25,
26]. In the present study, the FT4 tested ≤ and > 18 weeks of gestation showed different impacts on preterm birth. This disparity might be attributed to different sources of thyroid hormone. Placental transferred maternal FT4 is the only source for fetal demand ≤18 weeks of gestation since the fetal thyroid gland is not functionally mature until 18–20 weeks of gestation. The potential mechanism remains unclear. However, a high FT4 concentration was found to be associated with an increased risk of fetal brain underdevelopment and fetal growth restriction [
7,
8]. In our previous study, higher FT4 concentration also showed a relationship with the risk of congenital heart defects and this association became stronger when we restricted to the thyroid measurement tested between 12 and 18 weeks of gestation [
15]. Further studies on this disparity effect of FT4 on preterm birth were warranted.
The findings in the present study may partly explain the heterogeneity of the results in the previous studies and might suggest new insights into the potential mechanisms for preterm birth related to thyroid dysfunction [
5,
10‐
13,
27,
28]. Furthermore, these findings may provide evidence for clinical decisions related to a universal thyroid function test and thyroid dysfunction intervention. Clinicians should pay more attention to overt thyroid disease or TPOAb positive at a specific gestational age or in a targeted population. For example, women positive for TPOAb in the first trimester might be the targeted population for the intervention of levothyroxine in reducing the risk of preterm birth. This might explain the failure of levothyroxine in lowing preterm birth risk for women positive for TPOAb before conception in a previous intervention study [
29].
Strengths and limitations
The large sample size in the present study ensured a high statistical power to detect the association between thyroid dysfunction and preterm birth and further probe the specificity of the associations. However, there were some limitations. Firstly, the present study was a single-center study in Shanghai where the iodine status of pregnant women was iodine deficient [
30]. In addition, a high loss to follow-up rate (18.2%), imbalanced basic characteristics between women with and with antibody tests, and some important covariates (e.g., maternal BMI and smoking exposure) were not collected in the present study. These disadvantages might result in a potential selection bias and then affect the results. Therefore, multiple-center studies in regions with various iodine statuses for pregnant women are warranted to identify the generalizability of our results. Secondly, thyroid function was measured at different gestational ages rather than a dynamical observation. Thus, we can’t discriminate such specific associations were consequences of hyperthyroidism-persistent effects or results of gestational age-specific functions. Furthermore, treatment information was lacking in the present study. Thus, studies on the impact of dynamic thyroid function measurements and the treatment of gestational thyroid dysfunction on preterm birth were warranted. Finally, the interpretation of causal relationships should be cautious in observational studies.
Open AccessThis article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit
http://creativecommons.org/licenses/by/4.0/. The Creative Commons Public Domain Dedication waiver (
http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated in a credit line to the data.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.