Association between maternal thyroid function in early pregnancy and neurodevelopment in children aged 0–24 months: a retrospective cohort study

Abstract

Background

To investigate the influence of maternal thyroid function in early pregnancy on offspring neurodevelopment outcomes.

Methods

We retrospectively analyzed 20,855 mother-offspring pairs from Shenzhen, China, with delivery dates between February 2016 and May 2024. Maternal thyroid hormone levels were measured at 7–13 gestational weeks. Neurodevelopment outcomes were assessed using the Ages and Stages Questionnaire-Third Edition (ASQ-3) at 0–24 months of age.

Results

Compared with euthyroid mothers, overt hypothyroidism was associated with an increased risk of gross motor delay (adjusted odds ratio [aOR] 2.14; 95% confidence interval [CI] 1.17–3.63). Subclinical hypothyroidism was linked to a higher risk of delay in the problem-solving domain (aOR 1.63; 95% CI 1.03–2.49). Hypothyroxinemia was associated with increased risk of delays in both the gross motor (aOR 1.73; 95% CI 1.04–2.70) and fine motor domains (aOR 1.69; 95% CI 1.01–2.66). Subclinical hyperthyroidism was associated with an elevated risk of gross motor delay (aOR 1.97; 95% CI 1.14–3.18). Maternal thyroid-stimulating hormone (TSH) levels exhibited a U-shaped relationship with the risk of neurodevelopmental delay in the gross motor domain. (P-overall = 0.012; P-nonlinear = 0.003)

Conclusions

Maternal overt hypothyroidism, subclinical hypothyroidism, hypothyroxinemia, and subclinical hyperthyroidism in early pregnancy are associated with neurodevelopmental delays in children. Notably, this association exhibits a U-shaped pattern, whereby both low and high TSH levels are linked to an increased risk of neurodevelopmental delay.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1186/s12884-026-09695-w.

Background

Thyroid disorders are more frequent in women, including the ones in reproductive age, and affect up to 10% of pregnancies [1]. This high prevalence warrants clinical attention as thyroid hormones play a crucial role in fetal brain development by regulating neuronal migration, synaptic connectivity, and myelination [2]. Since the fetal thyroid has not become fully functional until approximately 18–20 weeks of gestation, the developing fetus relies primarily on maternal thyroid hormones transferred across the placenta during this critical period [3]. This dependence implies that thyroid dysfunction may impact fetal brain development, potentially resulting in neurodevelopmental delay, a condition characterized by delayed development of communication, motor, cognitive, social, and adaptive skills [4].

Although maternal overt hypothyroidism is known to impair fetal neurodevelopment, and the benefit of levothyroxine treatment in such cases is well established [5], the impact of milder forms of thyroid dysfunction remains controversial. While some studies suggested that subclinical hypothyroidism [6, 7], hypothyroxinemia [8–10], and subclinical hyperthyroidism [11] may adversely affect offspring neurodevelopment, others have found no significant association [12, 13]. Regarding levothyroxine therapy for these milder forms of thyroid dysfunction, the available evidence was insufficient to inform clinical practice [14–16]. These discrepancies may be attributable to methodological limitations, including heterogeneity in diagnostic criteria and neurodevelopmental assessment tools, which hinder the comparability across studies and preclude consistent conclusions. Furthermore, although thyroid-stimulating hormone (TSH) is widely regarded as a key indicator of thyroid function during pregnancy [17], evidence regarding the association between maternal TSH concentrations and offspring neurodevelopmental outcomes remains inconsistent [18]. This may reflect methodological differences, particularly in study populations, which may limit comparability and generalizability across studies [19, 20]. In addition, evidence on the sex-specific effects of maternal thyroid function on offspring neurodevelopment remains inconsistent, as reported in several population-based cohorts [11, 21, 22]. Therefore, further research is warranted to clarify the association between maternal thyroid function and neurodevelopment outcomes in offspring.

In this study, we aimed to assess the impact of five clinical categories of maternal thyroid dysfunction on offspring neurodevelopment, and to evaluate the association of early-pregnancy TSH and free thyroxine (fT4) levels with neurodevelopmental outcomes in offspring.

Methods

Study design and data source

This retrospective, single-center study was conducted at Shenzhen Baoan Women’s and Children’s Hospital in China. The study protocol was approved by the Institutional Review Board (LLSC-2025-03-10-03-KS). All data, including sociodemographic characteristics, maternal medical records spanning from pregnancy to delivery, and children’s routine health examinations were extracted from the hospital information system (HIS), which is a routinely updated electronic healthcare database containing information from the entire hospital. All mothers and children were assigned unique healthcare identification numbers, which enabled accurate matching of mother-offspring pairs and supported the integration of information across different databases. All thyroid function test data used in this study (including TSH and fT4) were extracted from the hospital’s laboratory information system (LIS), which manages the entire lifecycle of sample testing through standardized procedures and interfaces in real time with the HIS.

Study population

A total of 25,719 pregnant women with available thyroid function measurements during early pregnancy were identified from February 2016 to May 2024. Among them, 267 offspring did not have available neurodevelopment assessments. The inclusion criteria for study participants were: (1) pregnant women with thyroid function measurements available during early pregnancy; and (2) children who underwent neurodevelopmental screening using the Chinese version of the Ages & Stages Questionnaire, Third Edition (ASQ-3), a parent-completed developmental screening tool. Exclusion criteria were: (1) non-singleton pregnancies; (2) a history of maternal thyroid disorders before pregnancy; (3) use of medications affecting thyroid function; (4) maternal age > 40 or < 18 years at conception; and (5) conception via assisted reproductive technologies.

Of the remaining 25,452 mother-offspring pairs, 592 pairs were excluded due to non-singleton pregnancies, 1,247 pairs due to the use of assisted reproductive technologies, 1,045 pairs due to a history of maternal thyroid diseases, 1,405 pairs due to the use of medications affecting thyroid function, 225 pairs due to maternal age < 18 or > 40 years. After applying these criteria, 20,938 mother-offspring pairs remained eligible for classification. The study population comprised 18,898 euthyroid mothers (90.3%), 308 with overt hypothyroidism (1.5%), 529 with subclinical hypothyroidism (2.5%), 482 with hypothyroxinemia (2.3%), 280 with overt hyperthyroidism (1.3%), 358 with subclinical hyperthyroidism (1.7%), 82 with isolated hyperthyroxinemia (0.4%), and 1 with unclassified thyroid function (< 0.1%). For the final analysis, the groups with unclassified thyroid function (n = 1) and isolated hyperthyroxinemia (n = 82) were excluded due to lack of research value. Consequently, a total of 20,855 mother-offspring pairs were included in the final analysis (Fig. 1).

Fig. 1.

Selection of study participants

Exposure

Maternal serum levels of TSH, fT4, and thyroid peroxidase antibody (TPOAb) were routinely measured during early pregnancy (defined as 7–13 weeks of gestation). Maternal thyroid function data (TSH, fT4, and TPOAb) and clinical diagnoses of thyroid disorders were extracted from the medical records of maternal prenatal care and delivery. For women who underwent multiple tests during early pregnancy, the result from the most recent assay was recorded. As recommended by the American Thyroid Association [5], we adopted population-based, trimester-specific reference ranges for TSH and fT4, defining the 2.5th and 97.5th percentiles as the lower and upper limits, respectively. Accordingly, thyroid function in early pregnancy was assessed using the following trimester-specific reference values: 0.02–3.06 mIU/L for TSH and 10.15–17.69 pmol/L for fT4. Based on these reference ranges, euthyroidism was defined as maternal TSH and fT4 levels within 2.5th-97.5th percentiles; maternal overt hypothyroidism was defined as a TSH level above the upper limit accompanied by an fT4 level below the lower limit; maternal subclinical hypothyroidism was defined as a TSH level above the upper limit and an fT4 level within the reference range; maternal hypothyroxinemia was defined as a TSH level within the reference range with an fT4 level below the lower limit; maternal overt hyperthyroidism was defined as a TSH level below the lower limit with an fT4 level above the upper limit; maternal subclinical hyperthyroidism was defined as a TSH level below the lower limit and an fT4 level within the reference range. TPOAb positivity was defined as a serum TPOAb level exceeding the upper limit of the reference range.

Outcomes

The primary outcome was neurodevelopmental delay, which was evaluated using the Chinese version of the Ages & Stages Questionnaire-Third Edition (ASQ-3), a parent-completed developmental screening tool with demonstrated reliability and validity in the Chinese population [23]. The ASQ-3 is a 30-item questionnaire designed to assess the neurodevelopment of children aged 1–66 months across five domains: communication, gross motor, fine motor, problem-solving, and personal-social skills. Each domain consists of six questions, with response scored as 0, 5, or 10 points according to parental reports, generating a domain-specific score. The scores were evaluated against the domain-specific cut-off values to determine the child’s neurodevelopmental status. Scores greater than the mean minus 2 standard deviations (mean-2SD) but lower than the mean minus SD (mean-SD) were considered at risk for neurodevelopmental delay, whereas scores below mean-2SD were considered indicative of developmental delay. In this study, neurodevelopmental delay was defined as a score below mean-2SD in any domain, as this threshold has been shown to yield the most robust validity and highest accuracy for developmental delay identification in the Chinese population [24]. Continuous ASQ-3 domain scores were additionally summarized descriptively according to maternal thyroid status and are presented in the Supplementary Table S3.

Covariates

Potential confounders were pre-selected based on previous literature that identified variables associated with both maternal thyroid function and child neurodevelopment. Covariates included maternal age at delivery, education level, pre-pregnancy body mass index (BMI), parity, TPOAb status [25], gestational diabetes mellitus, gestational hypertension, mode of delivery, preterm birth, infant’s birthweight, and infant sex. Pre-pregnancy BMI was calculated by dividing the pre-pregnancy weight (in kilograms) by the square of the measured height (in meters); it was then further categorized into four groups based on the criteria recommended by the Working Group on Obesity in China: <18.5 kg/m², 18.5–23.9 kg/m², 24.0–27.9 kg/m², and ≥ 28 kg/m².

Statistical analyses

The quantile-quantile plot was used to test for normality. Normally distributed continuous variables were presented as mean (SD) and compared using Student’s t test. Categorical variables were presented as frequencies and percentages [n %] and compared using either Pearson’s chi-square test or Fisher’s exact test, as appropriate. First, multivariable logistic regression models were used to assess the association between maternal thyroid dysfunction and neurodevelopmental delay of offspring. Potential confounders were selected on the basis of biological plausibility and prior evidence of confounding, and were incorporated in the multivariable logistic regression models. The variance inflation factor (VIF) was employed to evaluate the multicollinearity among independent variables, with a VIF > 5 indicating significant multicollinearity. Subsequently, restricted cubic spline regression was applied to examine the dose-response relationship between TSH or fT4 levels and the prevalence of neurodevelopmental delay. Sensitivity analyses were then performed to assess the consistency and robustness of the associations. Potential effect modification by infant sex was examined, through sex-stratified analyses. All statistical analyses were conducted using R software, version 4.4.1. A two-sided P-value < 0.05 was considered significant.

Results

Baseline characteristics of the study population

The final study population comprised 20,855 mother-offspring pairs. Among them, 18,898 women (90.6%) were classified as euthyroid, while 1,957 (9.4%) were identified as having thyroid dysfunction, which encompassed overt hypothyroidism, subclinical hypothyroidism, hypothyroxinemia, overt hyperthyroidism, and subclinical hyperthyroidism. Maternal TSH and fT4 measurements across the different thyroid function categories are summarized in Supplementary Table S4. Compared with the euthyroid group, women with thyroid dysfunction were older at delivery (mean age: 30.25 vs. 29.89 years, P < 0.001), and had a higher prevalence of gestational diabetes mellitus (GDM) (16.3% vs. 3.6%, P < 0.001), TPOAb positivity (25.7% vs. 10.8%, P < 0.001), and cesarean delivery (35% vs. 33%, P = 0.017) (Table 1). In contrast, no differences were found between the two groups in pre-pregnancy BMI, education level, parity, gestational hypertension, infant sex, preterm birth, or gestational age at delivery (P > 0.05). Baseline characteristics were also generally comparable between mother-offspring pairs with and without available ASQ-3 data (Supplementary Table S2).

Table 1.

Demographic characteristics of women with euthyroidism and maternal thyroid dysfunction

Demographic variable	Overall N = 20,855	Euthyroidism N = 18,898	Thyroid Dysfunction N = 1,957	P 1
Mothers Age at delivery (years)
As continuous variable (mean [SD])	29.92(3.69)	29.89(3.67)	30.25(3.90)	< 0.001
< 25	2,143 (10%)	1,944 (10%)	199 (10%)	< 0.001
25–29	10,264 (49%)	9,383 (50%)	881 (45%)
30–34	6,775 (32%)	6,091 (32%)	684 (35%)
≥ 35	1,673 (8.0%)	1,480 (7.8%)	193 (9.9%)
Pre-pregnancy BMI
As continuous variable (mean [SD])	21.01(3.92)	21.03(3.93)	20.84(3.82)	0.10
< 18.5	5,655 (27%)	5,113 (27%)	542 (28%)	0.14
18.5–23.9	11,119 (53%)	10,066 (53%)	1,053 (54%)
24.0-27.9	2,928 (14%)	2,652 (14%)	276 (14%)
≥ 28	1,153 (5.5%)	1,067 (5.6%)	86 (4.4%)
Education level				0.8
Below senior high school	1,052 (5.0%)	955 (5.1%)	97 (5.0%)
Senior high school	2,598 (12%)	2,345 (12%)	253 (13%)
College or university and above	17,205 (82%)	15,598 (83%)	1,607 (82%)
Parity				0.11
0	14,044 (67%)	12,758 (68%)	1,286 (66%)
≥1	6,811 (33%)	6,140 (32%)	671 (34%)
Gestational diabetes mellitus	999 (4.8%)	680 (3.6%)	319 (16%)	< 0.001
Gestational hypertension	955 (4.6%)	865 (4.6%)	90 (4.6%)	0.9
TPOAb				< 0.001
Negative	18,312 (88%)	16,857 (89%)	1,455 (74%)
Positive	2,543 (12%)	2,041 (11%)	502 (26%)
Mode of delivery				0.017
Vaginal delivery	13,982 (67%)	12,717 (67%)	1,265 (65%)
Cesarean	6,873 (33%)	6,181 (33%)	692 (35%)
Children
Sex				0.094
Female	9,504 (46%)	8,577 (45%)	927 (47%)
Male	11,351 (54%)	10,321 (55%)	1,030 (53%)
Gestational age at delivery (weeks)
As continuous variable (mean [SD])	38.86(1.37)	38.86(1.37)	38.85(1.42)	0.6
≤ 33	139 (0.7%)	124 (0.7%)	15 (0.8%)	0.4
34–36	801 (3.8%)	714 (3.8%)	87 (4.4%)
37–38	5,733 (27%)	5,221 (28%)	512 (26%)
39–40	12,673 (61%)	11,469 (61%)	1,204 (62%)
≥ 41	1,509 (7.2%)	1,370 (7.2%)	139 (7.1%)
Birth weight (grams)
As continuous variable (mean [SD])	3,221.22(416.48)	3,219.56(416.72)	3,237.28(413.98)	0.045
< 2,500	769 (3.7%)	700 (3.7%)	69 (3.5%)	0.15
2,500-2,999	4,702 (23%)	4,269 (23%)	433 (22%)
3,000–3,499	10,140 (49%)	9,221 (49%)	919 (47%)
3,500-3,999	4,542 (22%)	4,071 (22%)	471 (24%)
≥ 4,000	702 (3.4%)	637 (3.4%)	65 (3.3%)

BMI  body mass index, SD  standard deviation, TPOAb  thyroid peroxidase antibodies

¹Continuous variables were compared using Student’s t test; categorical variables were compared using Pearson’s Chi-squared test or Fisher’s exact test, as appropriate

Bolded values indicate P < 0.05 compared with the euthyroid group

The association between the prevalence of maternal thyroid dysfunction and ASQ-3 outcomes

The overt hypothyroidism group exhibited a higher prevalence of neurodevelopmental delay in the gross motor domain (3.9% vs. 2.3%, P = 0.010; Table 2). The subclinical hypothyroidism group exhibited a higher prevalence of delay in the problem-solving domain (4.5% vs. 2.9%, P = 0.029). In the hypothyroxinemia group, a higher prevalence of neurodevelopmental delay was observed in both the gross motor (3.9% vs. 2.3%, P = 0.019) and the fine motor domains (3.7% vs. 2.3%, P = 0.045). The subclinical hyperthyroidism group showed a higher prevalence of delay in the gross motor domain (4.5% vs. 2.3%, P = 0.007). There were no differences in the overt hyperthyroidism group across any ASQ-3 domain (P > 0.05).

Table 2.

Prevalence of neurodevelopmental delay in each ASQ-3 domain according to maternal thyroid status

	Euthyroidism	Overt hypothyroidism	Subclinical hypothyroidism	Hypothyroxinemia	Overt hyperthyroidism	Subclinical hyperthyroidism
ASQ-3 Outcomes	18,898	308	529	482	280	358
Communication, n (%)1	999 (5.3%)	19 (6.2%)	27 (5.1%)	27 (5.6%)	17 (6.1%)	13 (3.6%)
Gross motor, n (%)1	436 (2.3%)	12 (3.9%)	9 (1.7%)	19 (3.9%)	5 (1.8%)	16 (4.5%)
Fine motor, n (%)1	440 (2.3%)	9 (2.9%)	15 (2.8%)	18 (3.7%)	6 (2.1%)	12 (3.4%)
Problem-solving, n (%)1	549 (2.9%)	7 (2.3%)	24 (4.5%)	17 (3.5%)	11 (3.9%)	10 (2.8%)
Personal-social, n (%)1	658 (3.5%)	5 (1.6%)	17 (3.2%)	16 (3.3%)	10 (3.6%)	18 (5.0%)

ASQ-3, Ages and Stages Questionnaire, third edition

ASQ-3 outcomes represent the proportion of children with neurodevelopmental delay (scores below mean-2SD) in the corresponding domain

1 Data are presented as n (%)

2 Comparisons for each ASQ-3 domain were performed using the Pearson’s Chi-squared test or Fisher’s exact test, as appropriate

Bolded values indicate P < 0.05 compared with the euthyroid group

Adjusted associations of maternal thyroid dysfunction with ASQ-3 outcomes

Compared with the euthyroid group, maternal overt hypothyroidism was associated with a higher risk of neurodevelopmental delay in the gross motor domain (adjusted odds ratio[aOR] 2.14; 95% CI 1.17–3.63; P = 0.008; Table 3). Regarding maternal subclinical hypothyroidism, an increased risk of developmental delay was observed in the problem-solving domain (aOR 1.63; 95% CI 1.03–2.49; P = 0.029; Table 3). Furthermore, maternal hypothyroxinemia was associated with a higher risk of delays in both the gross motor (aOR 1.73; 95% CI 1.04–2.70; P = 0.023; Table 3) and fine motor domains (aOR 1.69; 95% CI 1.01–2.66; P = 0.034; Table 3). Subclinical hyperthyroidism was also associated with an elevated risk in the gross motor domain (aOR 1.97; 95% CI 1.14–3.18; P = 0.009; Table 3). However, overt hyperthyroidism did not exhibit association with neurodevelopmental delays in any domain. This analysis revealed a VIF < 5 for all variables, suggesting no multicollinearity concerns. Sensitivity analyses, performed by excluding participants with GDM, positive TPOAb, preterm birth, or low birth weight infants, affirmed the robustness of the primary findings (Supplementary Table S1).

Table 3.

Estimated risks of ASQ-3 outcomes in association with maternal thyroid dysfunction from logistic regression models

	Communication		Gross motor		Fine motor		Problem-solving		Personal-social
ASQ-3 outcomes	Crude OR (95% CI)	aOR (95% CI)	Crude OR (95% CI)	aOR (95% CI)	Crude OR (95% CI)	aOR (95% CI)	Crude OR (95% CI)	aOR (95% CI)	Crude OR (95% CI)	aOR (95% CI)
Euthyroidism	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref
Overt hypothyroidism	1.18 (0.71, 1.83)	1.13 (0.68, 1.79)	2.02 (1.12, 3.34)	2.14 (1.17, 3.63)	1.26 (0.60, 2.32)	1.36 (0.63, 2.54)	0.78 (0.33, 1.53)	0.79 (0.33, 1.57)	0.46 (0.16, 1.00)	0.41 (0.15, 0.91)
Subclinical hypothyroidism	0.96 (0.64, 1.40)	0.88 (0.57, 1.30)	0.73 (0.35, 1.34)	0.76 (0.36, 1.42)	1.22 (0.70, 1.99)	1.29 (0.72, 2.15)	1.59 (1.02, 2.36)	1.63 (1.03, 2.49)	0.92 (0.54, 1.45)	0.80 (0.46, 1.28)
Hypothyroxinemia	1.11 (0.73, 1.60)	1.09 (0.72, 1.58)	1.74 (1.05, 2.70)	1.73 (1.04, 2.70)	1.63 (0.97, 2.55)	1.69 (1.01, 2.66)	1.30 (0.78, 2.03)	1.28 (0.76, 2.00)	0.95 (0.55, 1.52)	0.93 (0.54, 1.49)
Overt hyperthyroidism	1.18 (0.72, 1.81)	1.27 (0.78, 1.96)	0.66 (0.24, 1.45)	0.67 (0.24, 1.47)	0.93 (0.39, 1.82)	0.97 (0.41, 1.91)	1.29 (0.68, 2.20)	1.32 (0.70, 2.27)	0.97 (0.50, 1.70)	1.00 (0.51, 1.75)
Subclinical hyperthyroidism	0.68 (0.37, 1.13)	0.71 (0.39, 1.19)	1.98 (1.14, 3.19)	1.97 (1.14, 3.18)	1.45 (0.77, 2.49)	1.48 (0.78, 2.53)	0.96 (0.48, 1.71)	0.97 (0.48, 1.73)	1.47 (0.88, 2.30)	1.52 (0.90, 2.39)

aOR adjusted odds ratio, ASQ-3 Ages and Stages Questionnaire, third edition, CI confidence interval, Ref reference

Bolded values indicate P < 0.05 compared with the euthyroid group

Sex-stratified associations of maternal thyroid dysfunction with ASQ-3 outcomes

In male offspring (Table 4), maternal overt hypothyroidism was associated with a higher risk of neurodevelopmental delay in the gross motor domain (aOR 3.11; 95% CI 1.49–5.81; P < 0.001). Maternal subclinical hypothyroidism was associated with an increased risk of delay in the problem-solving domain (aOR 2.68; 95% CI 1.58–4.32; P < 0.001). Maternal hypothyroxinemia was also linked to a greater risk of delay in the gross motor domain (aOR 2.61; 95% CI 1.48–4.28; P < 0.001). Maternal overt hyperthyroidism was associated with an increased risk of delay in the problem-solving domain (aOR 2.18; 95% CI 1.02–4.11; P = 0.026). In addition, maternal subclinical hyperthyroidism was associated with a higher risk of delay in the gross motor domain (aOR 2.84; 95% CI 1.43–5.07; P = 0.001).

Table 4.

Estimated risks of ASQ-3 outcomes in association with maternal thyroid dysfunction, stratified by infant sex

ASQ-3 outcomes	Communication		Gross motor		Fine motor		Problem-solving		Personal-social
	Crude OR (95% CI)	aOR (95% CI)	Crude OR (95% CI)	aOR (95% CI)	Crude OR (95% CI)	aOR (95% CI)	Crude OR (95% CI)	aOR (95% CI)	Crude OR (95% CI)	aOR (95% CI)
	Boys
Euthyroidism	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref
Overt hypothyroidism	1.25 (0.69, 2.10)	1.16 (0.63, 1.98)	2.58 (1.26, 4.71)	3.11 (1.49, 5.81)	1.15 (0.41, 2.55)	1.22 (0.42, 2.77)	0.56 (0.14, 1.49)	0.60 (0.15, 1.62)	0.57 (0.17, 1.35)	0.50 (0.15, 1.20)
Subclinical hypothyroidism	1.04 (0.62, 1.61)	0.93 (0.55, 1.48)	1.08 (0.46, 2.14)	1.27 (0.53, 2.60)	1.58 (0.81, 2.79)	1.68 (0.83, 3.06)	2.44 (1.48, 3.80)	2.68 (1.58, 4.32)	0.80 (0.38, 1.47)	0.70 (0.33, 1.33)
Hypothyroxinemia	1.14 (0.69, 1.76)	1.12 (0.68, 1.74)	2.65 (1.51, 4.32)	2.61 (1.48, 4.28)	1.64 (0.84, 2.90)	1.64 (0.83, 2.92)	1.74 (0.96, 2.89)	1.68 (0.92, 2.83)	1.02 (0.52, 1.79)	0.98 (0.50, 1.73)
Overt hyperthyroidism	0.93 (0.44, 1.74)	0.95 (0.45, 1.77)	1.19 (0.36, 2.84)	1.30 (0.40, 3.13)	1.09 (0.33, 2.61)	1.13 (0.35, 2.72)	2.09 (0.98, 3.90)	2.18 (1.02, 4.11)	0.85 (0.30, 1.88)	0.86 (0.30, 1.91)
Subclinical hyperthyroidism	0.51 (0.20, 1.06)	0.52 (0.20, 1.08)	2.86 (1.44, 5.09)	2.84 (1.43, 5.07)	1.88 (0.84, 3.62)	1.95 (0.87, 3.76)	1.14 (0.45, 2.38)	1.15 (0.45, 2.40)	1.79 (0.94, 3.11)	1.82 (0.95, 3.18)
	Girls
Euthyroidism	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref	Ref
Overt hypothyroidism	0.98 (0.35, 2.17)	1.09 (0.38, 2.46)	1.30 (0.40, 3.13)	1.22 (0.36, 3.03)	1.42 (0.43, 3.41)	1.67 (0.50, 4.13)	1.09 (0.33, 2.61)	1.08 (0.32, 2.65)	0.25 (0.01, 1.13)	0.24 (0.01, 1.09)
Subclinical hypothyroidism	0.85 (0.38, 1.62)	0.78 (0.35, 1.53)	0.35 (0.06, 1.09)	0.31 (0.05, 1.01)	0.76 (0.23, 1.81)	0.79 (0.24, 1.96)	0.58 (0.18, 1.38)	0.55 (0.17, 1.36)	1.12 (0.50, 2.15)	0.95 (0.41, 1.88)
Hypothyroxinemia	1.00 (0.45, 1.92)	1.02 (0.46, 1.97)	0.61 (0.15, 1.63)	0.62 (0.15, 1.64)	1.59 (0.67, 3.18)	1.70 (0.71, 3.42)	0.69 (0.21, 1.63)	0.69 (0.21, 1.64)	0.82 (0.29, 1.81)	0.82 (0.29, 1.81)
Overt hyperthyroidism	1.68 (0.85, 2.97)	1.67 (0.84, 2.98)	0.24 (0.01, 1.08)	0.23 (0.01, 1.04)	0.80 (0.20, 2.12)	0.80 (0.20, 2.13)	0.61 (0.15, 1.62)	0.60 (0.15, 1.58)	1.18 (0.46, 2.46)	1.13 (0.44, 2.37)
Subclinical hyperthyroidism	1.00 (0.42, 1.99)	1.01 (0.42, 2.01)	1.18 (0.42, 2.62)	1.13 (0.40, 2.52)	1.03 (0.31, 2.45)	0.98 (0.30, 2.35)	0.79 (0.24, 1.87)	0.75 (0.23, 1.80)	1.13 (0.44, 2.36)	1.10 (0.43, 2.30)

aOR, adjusted odds ratio, ASQ-3, Ages and Stages Questionnaire, third edition, CI confidence interval, Ref reference

Bolded values indicate P < 0.05 compared with the euthyroid group

In female offspring, no association was observed between maternal thyroid dysfunction and neurodevelopmental delays (Table 4).

Non-linear association of TSH and fT4 levels with ASQ-3 outcomes

Figure 2 illustrates the relationship between maternal TSH and fT4 levels in early pregnancy and the risk of neurodevelopmental delay, modeled using restricted cubic splines regression. A non-linear association was found between maternal TSH levels and neurodevelopmental delay in the gross motor domain (P = 0.003), whereas no association was found for fT4 (P > 0.05). As shown in Fig. 2A, TSH levels in early pregnancy exhibited a U-shaped association with the OR of neurodevelopmental delay (overall P = 0.012; non-linear P = 0.003), where both low and high TSH levels were associated with an increased risk.

Fig. 2.

Restrictive cubic spline curves depicting the relationship between maternal (A) thyroid-stimulating hormone (TSH) and (B) free thyroxine (fT4) with offspring gross motor developmental delay. Red lines represent estimated odds ratios (ORs); pink shading indicates 95% confidence intervals (CIs). Reference values (vertical pink lines) were set at TSH (1.704 mIU/L) and fT4 (10.280 pmol/L) levels, where OR = 1.0 (dashed line). P-overall tests the overall association; P-non-linear tests non-linearity

Discussion

In this large retrospective cohort study, maternal overt hypothyroidism in early pregnancy was associated with impaired outcome in the gross motor domain of the ASQ-3, while subclinical hypothyroidism was associated with poorer performance in the problem-solving domain. Furthermore, maternal hypothyroxinemia was associated with a higher risk of neurodevelopmental delay in both the gross motor and fine motor domains. Maternal subclinical hyperthyroidism was also associated with an increased risk of delay in the gross motor domain. In contrast, overt hyperthyroidism showed no significant associations with neurodevelopmental delays in any domain. Notably, we identified a U-shaped association between maternal TSH levels and neurodevelopmental delay in the gross motor domain, indicating that both low and high TSH concentrations may adversely affect neurodevelopment. However, this non-linear association was not observed for fT4. Moreover, male infants appeared more vulnerable to the impact of maternal thyroid dysfunction across several domains, whereas female infants remained unaffected.

Although significant associations were observed between maternal thyroid dysfunction and offspring neurodevelopmental delay, the absolute increases in prevalence were relatively modest. Therefore, the clinical implications of these findings at the individual level should be interpreted with caution. Nevertheless, given the high prevalence of maternal thyroid dysfunction and the potential consequences of early neurodevelopmental impairment, even modest increases in absolute risk may have important population-level implications from a public health perspective. Early neurodevelopmental differences identified by screening tools such as the ASQ-3 may also represent indicators of later cognitive or behavioral outcomes, despite the need for further validation of their long-term predictive value. Further studies with longitudinal follow-up are needed to determine the long-term clinical significance of these findings.

This study has contributed to the growing body of research on the impact of maternal thyroid dysfunction on neurodevelopmental outcomes in children. Our findings indicate that maternal overt hypothyroidism is associated with neurodevelopmental delays in infants, which is consistent with previous studies [11, 15, 22, 26, 27]. The landmark study by Haddow et al. [15] first demonstrated the association between untreated maternal hypothyroidism and reduced cognitive performance in children, and this association has since been corroborated by subsequent cohort studies [11, 17]. Notably, our study extends these findings by additionally adjusting for key confounders, including TPOAb status, thereby reducing residual confounding and further strengthening the observed association. In contrast, long-term prospective cohort study, such as that reported by Nelson et al. [13] did not identify associations between maternal thyroid dysfunction and later academic performance. This discrepancy may reflect differences in developmental stage, outcome assessment, and compensatory neurodevelopmental processes occurring between infancy and school age. Further studies using different measurement scales and long-term neurodevelopmental outcomes are warranted.

In this present study, we also found a negative association between subclinical hypothyroidism and neurodevelopment during infancy. Although inconsistencies exist across studies, the adverse effects of maternal subclinical hypothyroidism on progeny development have been well reported in previous studies and clinical guidelines [5, 18, 27–29]. Our findings therefore add to the growing evidence suggesting that even milder degrees of maternal thyroid dysfunction may influence early neurodevelopment. Unlike overt hypothyroidism, the impact of subclinical hypothyroidism was primarily evident in the problem-solving domain rather than the gross motor domain, suggesting a divergent behavioral pattern. One possible explanation is that the spatial and temporal distribution of iodothyronine deiodinase activity differs across brain regions during development, potentially resulting in differential susceptibility of specific neurodevelopmental domains to varying degrees of maternal thyroid hormone insufficiency [30]. In line with our findings, Samuels et al. [31] reported that subclinical hypothyroidism was associated with impaired working memory, while motor learning memory remained unaffected. Together, these findings suggest that maternal thyroid hormone insufficiency may have domain-specific effects on offspring neurodevelopment.

Maternal hypothyroxinemia (isolated fT4) was associated with delayed motor development in offspring. Consistent with our findings, previous studies have reported adverse effects of maternal hypothyroxinemia on psychomotor, cognitive, and executive function in children [32, 33]. Moreover, the relationship between maternal thyroid function and neurodevelopmental outcomes appears complex. The Generation R Study reported an inverted U-shaped association between maternal fT4 levels and offspring IQ and brain morphology, suggesting that both low and high fT4 concentrations may adversely affect fetal brain development [17]. In the present study, we also observed a U-shaped association for TSH rather than fT4. This inconsistency may be explained by several methodological and biological factors. First, the neurodevelopment outcomes differed between studies. We assessed early neurodevelopment using the ASQ-3, whereas previous studies primarily relied on long-term IQ measures. Second, TSH and fT4 reflect different aspects of maternal thyroid physiology during pregnancy. In the first trimester, human chorionic gonadotropin (hCG) shares structural homology with TSH and exerts weak thyrotropic activity, resulting in increased fT4 concentrations and suppression of maternal TSH levels. Consequently, TSH and fT4 may exhibit different associations with offspring neurodevelopment.

Evidence on maternal subclinical hyperthyroidism and neurodevelopment remains limited and inconsistent. Our finding of adverse gross motor outcomes is consistent with Chen et al. [11], who reported that maternal subclinical hyperthyroidism was associated with a 6.9-point reduction in language scores in girls. However, Nelson et al. [13] found no association between subclinical hyperthyroidism and offspring educational attainment. Together with our TSH findings, this further supports a broader U-shaped relationship between maternal thyroid function and offspring neurodevelopment, suggesting that both hypothyroid and hyperthyroid states may adversely affect the developing brain [34]. This pattern may partly reflect the complex regulation of local thyroid hormone availability in the developing brain by transporters and iodothyronine deiodinases, which varies across brain regions and developmental stages [35]. It is also plausible that such delays are transient and resolve with catch-up development, underscoring the need for longitudinal follow-up.

Our findings also revealed sexually dimorphic effects of maternal thyroid dysfunction on child neurodevelopment. In our study, boys exhibited greater susceptibility to adverse neurodevelopmental outcomes linked to maternal overt hypothyroidism, subclinical hypothyroidism, hypothyroxinemia, overt hyperthyroidism, and subclinical hyperthyroidism with more pronounced vulnerability observed in the motor and problem-solving domains, whereas girls demonstrated greater resilience to neurodevelopmental impairments associated with maternal thyroid dysfunction. This pattern of sex-specific vulnerability aligns with findings from a previous prospective cohort study [11]. An observational study in China has also reported an association between low cord blood fT4 levels and impaired neurodevelopment in males, but not in females [36]. The underlying mechanisms may involve the neuroprotective effect of estrogen [37, 38], X-linked gene O-linked N-acetylglucosamine transferase (OGT)-mediated signaling in the placenta [39], and hormone-mediated DNA demethylation [40], although the specific mechanisms remain to be established.

The strengths of the present study include its large-scale retrospective cohort design; the inclusion of both continuous (TSH/fT4 levels) and categorical (five types of clinical diagnoses) exposure variables, allowing a comprehensive assessment of dose-response relationships; and the use of sex-stratified analyses to explore potential sex-specific effects. Despite these strengths, several limitations should be noted. First, this was a single-center study, which might have limited the generalizability of the findings to broader populations. Second, because only children with available neurodevelopmental assessments were included, selection bias related to loss to follow-up cannot be entirely ruled out. Third, the multiple comparisons and statistical tests may increase the risk of chance findings. Fourth, while our study revealed several statistically significant associations, the underlying biological mechanisms have yet to be fully elucidated. In addition, although we adjusted for a range of important maternal and perinatal factors, residual confounding from unmeasured variables cannot be completely excluded.

Conclusions

Maternal overt hypothyroidism, subclinical hypothyroidism, hypothyroxinemia, and subclinical hyperthyroidism in early pregnancy are associated with an increased risk of neurodevelopmental delay in children aged 0–24 months. Notably, this relationship exhibits a U-shaped pattern, whereby both low and high TSH levels in early pregnancy are linked to a greater risk. These findings support the importance of careful monitoring of maternal thyroid function during early pregnancy. Further prospective studies are needed to confirm these associations and clarify their long-term clinical implications.

Abbreviations

aOR

Adjusted odds ratio

ASQ-3

Ages and stages questionnaire, third edition

BMI

Body mass index

CI

Confidence interval

DNA

Deoxyribonucleic acid

fT4

Free thyroxine

GDM

Gestational diabetes mellitus

HIS

Healthcare (hospital) information system

IQ

Intelligence quotient

IRB

Institutional review board

LIS

Laboratory information system

OGT

O-linked N-acetylglucosamine transferase

OR(s)

Odds ratio(s)

SD

Standard deviation

TPOAb

Thyroid peroxidase antibody

TSH

Thyroid-stimulating hormone

VIF

Variance inflation factor

Authors’ contributions

CKW, YY, MW, and LYM designed the study. CKW, YY, and HL analyzed the data. CKW and wrote the manuscript. ZW and HL took part in data collection and management. WWL, ZHL, XY, JHR, and GHL reviewed the manuscript and contributed to manuscript revision. All authors contributed to the article and approved the submitted version. All authors reviewed the manuscript.

Funding

The study was funded by the Science and Technology Innovation Program of Hunan Province (2024RC3048 to Mi Wang), and the National Natural Science Foundation of China (82301738).

Data availability

The data presented in this article are not readily available due to ethical restrictions related to human participants but are available from the corresponding author upon reasonable request (E-mail: mi.wang@csu.edu.cn).

Declarations

Ethics approval and consent to participate

This study was performed in accordance with the Declaration of Helsinki. As this research was a retrospective cohort study using only medical records without any interventions or direct contact with participants, the need for consent to participate was waived by the Ethics Committee of Shenzhen City Baoan District Women’s and Children’s Hospital. The study protocol was approved by the Institutional Review Board (IRB) of Shenzhen City Baoan District Women’s and Children’s Hospital (LLSC-2025-03-10-03-KS).

This study involves human participants and was approved by the Ethics Committee of Shenzhen City Baoan District Women’s and Children’s Hospital (LLSC-2025-03-10-03-KS). All patient data were de-identified and anonymized prior to analysis.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.