Perinatal maternal distress and neurobehavioural development in toddlers

Abstract

Background

Maternal perinatal psychological distress (e.g., anxiety, depression) is common and has been linked to adverse child behavioural outcomes. Altered brain function provides a potential mechanism through which perinatal distress may impact child behaviour, but specific associations remain unclear. This study investigated associations between perinatal distress, child brain functional connectivity at 2 years, and child behaviour at 2 and 3 years.

Methods

Neuroimaging and behavioural data were obtained from the Pregnancy during the COVID-19 Pandemic (PdP) study. Maternal distress (anxiety and depression symptoms) was self-reported throughout the perinatal period; analyses included measures from intake survey and 2 years postpartum. Child behaviour at 2 and 3 years was parent-reported using the Ages and Stages Questionnaire (ASQ-3). Resting-state fMRI data was collected at 2 years (n = 62) to quantify connectivity within the default mode, executive control, frontoparietal, and sensorimotor networks. Associations between maternal prenatal and postnatal distress, child functional connectivity, and child behaviour were examined using regression models.

Results

Postnatal, but not prenatal, distress was associated with lower gross motor scores at 2 years (B = -0.5, T = -2.8, p = 0.005, df = 3517, q = 0.01). Higher distress both prenatally and postnatally was associated with lower ASQ-3 scores at 3 years across all domains (q < 0.05). Child brain connectivity was not related to maternal distress or behaviour.

Conclusions

Perinatal distress was associated with worse behaviour at 2 and 3 years, although no associations with connectivity were observed, suggesting other pathways may underlie these effects. Nonetheless, these findings highlight the importance of supporting mental health during pregnancy and postnatally to promote healthy child development.

Graphical Abstract

Highlights

• •We examined perinatal distress and child brain and behaviour at ages 2 and 3 years.
• •Maternal distress at 2 years postpartum was associated with lower child gross motor scores at 2 years.
• •Prenatal and postnatal distress were related to poorer child behaviour scores at 3 years.
• •No associations were found between maternal distress and child functional brain connectivity.
• •Detection and treatment of perinatal distress may help support child development.

1. Background

Maternal psychological distress (i.e., anxiety and depression) is prevalent during pregnancy and the postpartum period, affecting up to 1 in 5 individuals (Fawcett et al., 2019, Gavin et al., 2005, Mateus et al., 2022, O’Hara and Wisner, 2014). Pregnant individuals who experience distress are at a higher risk of continued challenges postnatally (Bayrampour et al., 2016, Obrochta et al., 2020), and maternal mental health throughout the perinatal period can have lasting effects on a child’s emotional, cognitive, and behavioural growth (Bell et al., 2019). The COVID-19 pandemic caused elevated anxiety and depression symptoms among pregnant individuals, raising concerns about potential outcomes on child development, but also offering a unique opportunity to study their effects on the child brain (Lebel et al., 2020).

Infancy and early childhood are critical periods for behavioural development, and recent studies have increasingly focused on how prenatal maternal distress may contribute to emerging behaviours. For example, prenatal maternal depressive and anxiety symptoms in the second and third trimesters are linked to higher risk of developmental delays in children aged 0–24 months (Chen et al., 2025, Duguay et al., 2022, Wei et al., 2023, Zhang et al., 2023). Higher prenatal distress is also related to more emotional and behaviour problems in early childhood (Lahti et al., 2017, O’Connor et al., 2002, O’Donnell et al., 2014).

Postnatal maternal distress may be a continuation of prenatal symptoms or arise independently (Obrochta et al., 2020), and some evidence suggests postnatal symptoms also influence child and adolescent behaviour (Rogers et al., 2020), though findings are mixed. Maternal depression symptoms during and following pregnancy are independently associated with lower child developmental milestone scores (Chorbadjian et al., 2020, Li et al., 2024, Tuovinen et al., 2018), with similar associations reported with maternal anxiety at 3 years of age (Mughal et al., 2019). However, other studies report null findings or positive associations between maternal distress and child behaviour at 1 and 2 years (DiPietro et al., 2006, Jaffer et al., 2024), highlighting the need for continued research to further elucidate these evolving relationships and the factors that may moderate or mediate these effects.

One potential pathway through which both prenatal and postnatal maternal distress could influence child behavioural outcomes is alterations in brain connectivity, particularly within functional networks that support emerging emotional regulation, cognitive processing, motor development, and social capacities. For instance, Hay et al. (2020) found that amygdala-frontal structural connectivity mediated the relationship between depressive symptoms during late pregnancy and externalizing behaviours in preschool-aged boys. Similarly, Uy et al. (2023) found that worse postnatal maternal anxiety and depressive symptoms were associated with more negative fronto-amygdala resting-state connectivity in children, which in turn was linked to more internalizing problems. These findings suggest that alterations in brain connectivity may reflect a process through which maternal distress shapes early child behaviour, highlighting the value of examining large-scale functional networks involved in emotional, cognitive, and motor function during early childhood.

The behavioural findings suggest that the effects of maternal distress are not limited to emotional or behavioural problems but span multiple domains of early development, including motor, problem-solving, and communication skills (Chen et al., 2025, Duguay et al., 2022, Wei et al., 2023, Zhang et al., 2023). This broad pattern of behavioural challenges points to the possibility that maternal distress may influence brain regions and networks beyond those traditionally associated with emotion regulation, such as the amygdala (Donnici et al., 2021, Hay et al., 2020, Hill et al., 2025, Posner et al., 2016, Qiu et al., 2015, Soe et al., 2016, Soe et al., 2017, Uy et al., 2023). Given that early behaviours rely on multiple systems, it is possible that maternal distress may disrupt broader functional brain networks, or the coordination between them, in ways that affect diverse developmental outcomes (Bruchhage et al., 2020, Jaffer et al., 2024, Wang et al., 2020).

Here, we leveraged neuroimaging and behavioural data from a large cohort of Canadian children to examine how prenatal and postnatal maternal distress relate to child communication, gross motor, fine motor, problem solving, and personal-social behaviours at 2 and 3 years of age. We also examined child brain functional connectivity within the default mode, executive control, left frontoparietal, and sensorimotor networks to understand whether brain connectivity may serve as a pathway linking maternal mental health to early child development.

2. Methods

2.1. Study participants

Participants (n = 3679) were part of the ongoing Pregnancy during the COVID-19 Pandemic (PdP) study, a national Canadian study assessing the mental and physical health of pregnant individuals and their children during the COVID-19 pandemic. Eligibility criteria for enrolment in the PdP study were being < 35 weeks pregnant, living in Canada, ≥ 17 years of age, and able to read and write in English or French (Giesbrecht et al., 2021).

Study recruitment ran from April 2020 to April 2021, and prenatal distress measures were collected from then until December 2021. Thus, most of this data was collected during varying stages of lockdowns (which also varied regionally across Canada), and primarily before widespread access to the vaccine. The fMRI imaging data on children was collected between November 8, 2022 and July 5, 2024, and postnatal distress measures were also collected during this time (Table 1).

Table 1.

Demographic characteristics of participants. The full sample providing perinatal distress measures and child behavioural outcomes.

Demographics		Mean ± SD	Range
Children (n = 3679)
Gestational age at birth (weeks)		39.4 ± 1.6	26.6–42.6
Child sex		1914 female (52%) 1765 male (48%)
Child age at 2-year follow-up survey (months)		24.0 ± 0.5	23.5–28.3
Child age at 3-year follow-up survey (months)		36.2 ± 0.7	35.7–46.7
ASQ-3 at 2 years of age (n = 3679)	Communication	49.9 ± 13.0	0.0–60.0
	Gross motor	52.8 ± 9.2
	Fine motor	50.9 ± 7.6
	Problem solving	49.3 ± 9.0
	Personal-social	46.4 ± 9.8
ASQ-3 at 3 years of age (n = 3507)	Communication	52.7 ± 8.1
	Gross motor	54.2 ± 8.3
	Fine motor	43.0 ± 13.9
	Problem solving	53.8 ± 7.4
	Personal-social	52.1 ± 9.6
Mothers
Prenatal distress at survey intake (n = 3679)	EPDS score	9.5 ± 5.2	0.0–27.0
	PROMIS T score	57.5 ± 8.1	36.3–82.7
Postnatal distress at 2 years (n = 3524)	EPDS score	8.4 ± 5.2	0.0–30.0
	PROMIS T score	55.7 ± 8.1	36.3–79.5
Prenatal social support (SSEQ)		56.6 ± 14.1	4.0–80.0
		Group	N (%)
Household income		Less than $20, 000	29 (0.8%)
		$20,000 - $39,999	121 (3.3%)
		$40,000 - $69,999	377 (10.2%)
		$70,000 - $99,999	659 (17.9%)
		$100,000 - $124,999	738 (20.1%)
		$125,000 - $149,999	573 (15.6%)
		$150,000 - $174,999	481 (13.1%)
		$175,000 - $199,999	275 (7.5%)
		$200,000 +	423 (11.5%)
Maternal education		Less than high school diploma	9 (0.2%)
		Completed high school	167 (4.5%)
		Completed trade, technical, vocational school or business/community college	716 (19.5%)
		Bachelor's Degree	1557 (42.3%)
		Master's Degree	880 (23.9%)
		Doctorate (PhD)	152 (4.1%)
		Professional (MD, JD, DDS, ETC)	198 (5.4%)
Ethnicity		White	3230 (87.8%)
		First Nations, Metis, and Inuit	68 (1.9%)
		Black	20 (0.5%)
		West Asian	21 (0.6%)
		South Asian	63 (1.7%)
		Southeast Asian (incl. Filipino)	44 (1.2%)
		East Asian (incl. Chinese, Korean)	72 (2.0%)
		Hispanic/LatinX	56 (1.5%)
		Biracial	99 (2.7%)

2.2. Maternal psychological distress

Maternal depression and anxiety were measured perinatally using the Edinburgh Postnatal Depression Scale (EPDS) and the Patient Reported Outcomes Measurement Information System (PROMIS) Anxiety Adult 7-item short form, respectively (Bergink et al., 2011, Cella et al., 2019); here, we analysed measures from the intake survey and at 2 years postpartum (Cox et al., 1996, Netsi et al., 2018). Prenatal questionnaires were completed at study intake, which occurred across pregnancy, ranging from 3.9 to 35.6 weeks’ gestation. Gestational age at intake was weakly correlated with prenatal maternal distress (r = 0.04, p < 0.001), suggesting that timing of questionnaire completion was not meaningfully associated with prenatal distress levels in this sample. Both measures are self-report assessments validated for use during and after pregnancy, capturing symptom severity over the past seven days. The EPDS consists of 10 questions, each with four response options. The overall EPDS score can range from 0 to 30, with scores > 12 consistent with clinically significant depression symptoms (Bergink et al., 2011). The PROMIS consists of 7 questions, each question has five response options ranging from 1 (never) to 5 (always). PROMIS scores are converted to standardized T-scores, where 50 represents the population average and scores > 59 denote clinically significant anxiety symptoms (Cella et al., 2019). Higher scores on both tools imply greater symptom severity.

Given high correlation between these measures (prenatal R=0.80; postnatal R=0.79) and to reduce the number of variables, EPDS and PROMIS scores at each time point were combined using factor analysis in R to create a single prenatal distress for PdP study intake, and a single postnatal distress score at 2-years postpartum.

2.3. Child behavioural measures

Child behaviour was assessed at 2 and 3 years of age using the Ages and Stages Questionnaire 3rd edition (ASQ-3) (Squires and Bricker, 2009). The ASQ-3 is a 30-question parent report that evaluates five areas of child behaviour: gross motor, fine motor, communication, problem solving, and personal social (Agarwal et al., 2020). Questions are scored using a 0-, 5-, and 10-point system corresponding to ‘Not Yet’, ‘Sometimes’, and ‘Yes’ responses, respectively (Paul H. Brookes Publishing Co., 2009). Scores are then summed within each domain. Children scoring ≥ 2 standard deviations below the mean in any domain are classified as at risk for developmental delay (Paul H. Brookes Publishing Co., 2009). The ASQ-3 demonstrates excellent reliability (test-retest reliability of 0.92 and inter-rater reliability of 0.93), validity (0.82–0.88), sensitivity (0.86), and specificity (0.85) (Paul H. Brookes Publishing Co., 2009).

2.4. Image acquisition

Parents in the PdP study residing in or around Calgary or Vancouver were invited to have their children undergo a brain magnetic resonance imaging (MRI) scan at 2 years. Scans took place at the Alberta Children’s Hospital (ACH) for Calgary participants and at BC Children’s Hospital (BCCH) for those in Vancouver. Children were excluded from scanning if they were born preterm (<37 weeks gestation), experienced major birth complications, or had known neurological or genetic conditions. All children in the imaging sample were from singleton pregnancies.

GE 3 T MR750w scanners with a 32-channel head coil were used at both hospitals to acquire resting state functional MRI (rs-fMRI) data using the following parameters: TR = 2000 ms, TE = 30 ms, FOV = 23 cm, matrix = 64 × 64, voxel size = 3.6 × 3.6 × 3.6 mm³, flip angle = 60°, with 37 slices. Children were awake during scanning and watching videos of their choice. 140 parents consented to MRI for their child; 84 datasets were collected, and after quality checking (see below), a total of 62 scans were included in image analysis (Table 2).

Table 2.

Demographic characteristics of participants in the imaging subset. This subset was used to examine associations between prenatal and postnatal maternal distress, and child brain connectivity and behaviour. Maternal education and household income distributions were comparable to the full sample.

Demographics		Mean ± SD	Range
Children (n = 62)
Child age at scan (months)		30.5 ± 1.6	27.4–33.9
Gestational age at birth (weeks)		39.5 ± 1.2	36.1–41.6
Child sex		24 female (38.7%) 38 male (61.3%)
Site	Calgary	46 (74.2%)
	Vancouver	16 (25.8%)
ASQ-3 at 2 years	Communication	51.8 ± 12.6	15.0–60.0
	Gross motor	53.6 ± 10.7	15.0–60.0
	Fine motor	51.0 ± 8.6	25.0–60.0
	Problem solving	49.0 ± 9.1	25.0–60.0
	Personal-social	50.0 ± 8.9	25.0–60.0
ASQ-3 at 3 years	Communication	51.8 ± 5.6	35.0–60.0
	Gross motor	55.3 ± 8.0	20.0–60.0
	Fine motor	43.2 ± 11.2	20.0–60.0
	Problem solving	51.7 ± 9.9	20.0–60.0
	Personal-social	53.6 ± 5.6	35.0–60.0
Mothers
Prenatal distress at survey intake	EPDS score	9.2 ± 5.4	1.0–25.0
	PROMIS T score	57.9 ± 7.2	42.1–71.5
Postnatal distress at 2 years	EPDS score	8.8 ± 5.3	1.0–24.0
	PROMIS T score	56.4 ± 6.6	42.1–75.8
Prenatal social support (SSEQ)		55.4 ± 13.0	13.0–76.0

2.5. Image processing and analysis

FSL software was used to preprocess neuroimaging data and perform functional connectivity analyses (Jenkinson et al., 2012). Non-brain tissue was removed using the Brain Extraction Tool. Volumes with excessive motion (>0.25 mm relative mean displacement) were excluded. Multivariate Exploratory Linear Optimized Decomposition into Independent Components (MELODIC) Independent Component Analysis (ICA) was then applied to identify and remove noise-components (e.g., high-frequency signals, physiological artifacts) (Griffanti et al., 2017). Participants with fewer than 120 low-motion volumes (i.e., <4 min of low-motion data) were excluded from further analysis. The remaining volumes were concatenated and registered to the 2-year Automated Anatomical Labeling (AAL) atlas from the University of North Carolina (UNC) (Shi et al., 2011).

Functional brain networks were defined based on multi-session ICA performed on the 2-year rs-fMRI data. ICA components corresponding to the default mode, executive control, left frontoparietal, and sensorimotor networks were identified through correlation with the Smith resting-state infant network template (Smith et al., 2009). These brain networks were selected for their fundamental roles in emotional regulation, cognitive function, and motor development during early childhood (Gao, 2025), aligning with the behavioural domains assessed by the ASQ-3 (Bruchhage et al., 2020). Given the developmental variability in functional network organization during toddlerhood, we used a data-driven approach to identify network regions in our 2-year-old sample, rather than relying on adult-defined regions a priori. This allowed for age-appropriate characterization of network regions of interest, as networks in toddlers may not align precisely with adult-derived anatomical boundaries at this stage of development (Chen et al., 2021, Wen et al., 2019).

Two spherical ROIs with 5 mm radii were placed at peak connectivity hubs identified from ICA-derived network maps and localized using the 2-year UNC AAL atlas. Specifically, ROIs were placed at the following coordinates: default mode network (ROI 1: [-1, 43, 5]; ROI 2: [-3, −63, −1]), left frontoparietal network (ROI 1: [-41, 5, 19]; ROI 2: [-37, −43, 19]), executive control network (ROI 1: [-11, 45, 15]; ROI 2: [13, 45, 15]), and sensorimotor network (ROI 1: [-19, −27, 49]; ROI 2: [13, −27, 49]) (Fig. 1). For each participant, average time series were extracted from each 5 mm ROI sphere, and functional connectivity within each network was computed as the Pearson correlation between the two corresponding ROIs (Fig. 1).

Fig. 1.

Regions of interest used in brain network connectivity analysis. Two regions per network were selected for the default mode (DMN), executive control (EC), left frontoparietal (LFP), and sensorimotor (SM) networks. Corresponding ROI coordinates were ROI 1: [-1, 43, 5] and ROI 2: [-3, −63, −1] for the DMN, ROI 1: [-11, 45, 15] and ROI 2: [13, 45, 15] for the EC network, ROI 1: [-41, 5, 19] and ROI 2: [-37, −43, 19] for the LFP network, and ROI 1: [-19, −27, 49] and ROI 2: [13, −27, 49] for the SM network. These regions of interest were used in functional connectivity analyses and are shown on transparent brain templates in sagittal and axial views to illustrate their spatial locations.

To account for potential differences across imaging sites, post-processing data harmonization was performed using the neuroCombat package in R (Fortin et al., 2017). Combat harmonization was applied to adjust for site-related batch effects by modeling differences in the mean and variance of imaging measures across datasets, while preserving variability associated with biological variables of interest (Fortin et al., 2018, Yu et al., 2018). This approach has been validated for use in multi-site neuroimaging studies (Fortin et al., 2018, Yu et al., 2018). Child age and sex were included as covariates to preserve biological variability in the harmonized data.

2.6. Missing data handling

In the full PdP cohort analyses (distress → behaviour), missing data were addressed using listwise deletion.

To enable inclusion of all participants with neuroimaging data, multiple imputation by changed equations (MICE) was used in the imaging analyses (Austin et al., 2021). Missing prenatal (n = 2) and postnatal (n = 9) maternal distress scores, child behavioural scores at 2 years (n = 9) and 3 years (n = 12–15), and prenatal social support measures (n = 3) were imputed using MICE in R. Twenty imputed datasets were generated to account for uncertainty with missing values, and pooled estimates were used in all imaging analyses. To verify the robustness of results, all imaging models were also re-run using listwise deletion.

2.7. Statistical analyses

All statistical analyses were performed in R (version 4.4.0). Regression models were used to assess associations between prenatal and postnatal maternal distress and child behavioural outcomes (ASQ-3 gross motor, fine motor, communication, problem solving, and personal-social skills) at 2 and 3 years of age (separately). The same postnatal distress measure (2 years) was included in models of both 2-year and 3-year behavioural outcomes to assess consistency of associations.

For each behavioural timepoint, analyses included all participants with ASQ-3 data, resulting in overlapping but non-identical samples at 2 and 3 years (some children had data at only one age; overlap occurred where children had data at both ages). Identical samples were analysed in a supplementary analysis (see below). Multiple gestation was an exclusion criterion for the distress–behaviour analyses, and the number of mothers and infants was therefore equivalent. Child age at survey, child sex, maternal education, prenatal social support (Social Support Effectiveness Questionnaire) and gestational age at birth were evaluated for inclusion as covariates, and all were retained in these models. The correlation between prenatal distress and social support was −0.38 (p < 0.001), and between postnatal distress and social support was −0.24 (p < 0.001); showing that lower prenatal social support was correlated with higher maternal distress both before and after pregnancy. Moderation by child sex was also tested across all models but was non-significant and therefore removed, except for the 3-year communication model where the interaction term between prenatal distress and child sex was retained due to statistical significance. Each model was structured as: child behaviour ~ distress score + child age at survey + child sex + maternal education + prenatal social support + gestational age at birth. False discovery rate (FDR) correction (q < 0.05) was applied separately across the five behavioural domains for each analysis of distress predicting behaviour, resulting in a total of 20 models. In a supplementary analysis, we conducted the same analysis on the subset of participants (n = 2925) who had complete ASQ-3 assessments at both 2 and 3 years of age.

In children with neuroimaging, regression models were used to examine associations between prenatal and postnatal maternal distress and child functional connectivity within four brain networks: default mode, executive control, left frontoparietal, and sensorimotor. Child age at scan, child sex, and maternal education were retained as covariates, while prenatal social support and gestational age at birth were evaluated, but excluded as they did not significantly contribute to the models. Each model was structured as: brain connectivity ~ distress score + child age at scan + child sex + maternal education. FDR correction (q < 0.05) was applied separately across the four models of prenatal distress and four models of postnatal distress.

Finally, relationships between child brain functional connectivity and ASQ-3 behavioural outcomes at 2 and 3 years were examined. Child age at scan, child sex, and gestational age at birth were included as covariates; maternal education and prenatal social support were excluded, as they did not contribute significantly to the model. Each model was structured as: child behaviour ~ brain connectivity + child age at scan + child sex + gestational age at birth. FDR correction (q < 0.05) was applied separately across the five behavioural domains at each timepoint (2 and 3 years) within each of the four brain networks, resulting in a total of 40 models.

2.8. Sex and gender considerations

Biological male and female children were included in the study. Since brain structure and function, along with behaviour, can vary between sexes, child sex was included as a covariate in all statistical analyses, and interaction terms were tested to examine whether associations between maternal distress and child outcomes were moderated by child sex. Children at 2 and 3 years of age are generally considered to be too young to inform or advise their own gender identity, therefore gender was not included. All maternal distress data in this sample were from individuals who identify as women.

2.9. Ethics approval

The study received approval from the University of Calgary Conjoint Health Research Ethics Board (CHREB: REB20-0500) and the University of British Columbia Clinical Research Ethics Board and the British Columbia Children’s and Women’s Research Ethics Board (H21-03099). All maternal participants provided written informed consent for themselves and informed consent on behalf of their children for neuroimaging.

3. Results

3.1. Maternal distress and child behaviour at 2 years

Higher postnatal maternal distress was related to lower child gross motor scores on the ASQ-3 (B = −0.5, SE = 0.2, T = -2.8, p = 0.005, df = 3517, q = 0.01, R² = 0.01; Fig. 2). A similar relationship was found with prenatal distress and child gross motor scores, although this association did not remain significant following FDR correction (B = −0.4, SE = 0.2, T = -2.1, p = 0.04, df = 3672, q = 0.06, R² = 0.01). No significant relationships were found between prenatal or postnatal distress and child communication, fine motor, problem solving, or personal-social child behaviours at 2 years.

Fig. 2.

Association between postnatal maternal distress and child gross motor scores at 2 years. Higher postnatal maternal distress was significantly associated with lower gross motor scores (B = −0.5, SE = 0.2, T = -2.8, p = 0.005, df = 3517, q = 0.01, R² = 0.01).

Child age at survey, female sex, maternal education, prenatal social support, and gestational age at birth were all positively associated with ASQ‑3 scores across domains.

3.2. Maternal distress and child behaviour at 3 years

Prenatal and postnatal maternal distress were significantly associated with child ASQ-3 scores across all five domains at 3 years of age.

Higher prenatal distress was related to lower gross motor (B = −0.6, SE = 0.2, T = -3.8, p < 0.001, df = 3500, q < 0.001, R² = 0.01), fine motor (B = −0.6, SE = 0.3, T = -2.4, p = 0.02, df = 3459, q = 0.04, R² = 0.08), problem solving (B = −0.5, SE = 0.2, T = -2.4, p = 0.02, df = 3423, q = 0.04, R² = 0.02), and personal-social scores (B = −0.6, SE = 0.1, T = -4.1, p < 0.001, df = 3438, q < 0.001, R² = 0.07); see Fig. 3. The association between higher prenatal distress and lower child communication was significantly moderated by child sex (B = −1.0, SE = 0.3, T = -3.3, p = 0.001, df = 3497, q = 0.01, R² = 0.05), with a stronger negative relationship observed in males compared to females; see Fig. 3.

Fig. 3.

Associations between prenatal maternal distress and child behaviour at 3 years. Higher prenatal distress was significantly associated with lower scores across all behavioural domains: (A) communication, moderated by child sex (B = −1.0, SE = 0.3, T = -3.3, p = 0.001, df = 3497, q = 0.01, R² = 0.05), (B) gross motor (B = −0.6, SE = 0.2, T = -3.8, p < 0.001, df = 3500, q < 0.001, R² = 0.01), (C) fine motor (B = −0.6, SE = 0.3, T = -2.4, p = 0.02, df = 3459, q = 0.04, R² = 0.08), (D) problem solving (B = −0.5, SE = 0.2, T = -2.4, p = 0.02, df = 3423, q = 0.04, R² = 0.02), and (E) personal-social skills (B = −0.6, SE = 0.1, T = -4.1, p < 0.001, df = 3438, q < 0.001, R² = 0.07).

Similarly, higher postnatal maternal distress was significantly associated with lower scores in communication (B = −0.4, SE = 0.2, T = -2.6, p = 0.01, df = 2986, q = 0.02, R² = 0.04), gross motor (B = −0.6, SE = 0.2, T = -3.7, p < 0.001, df = 2988, q = 0.002, R² = 0.01), fine motor (B = −0.9, SE = 0.3, T = -3.4, p = 0.001, df = 2953, q = 0.004, R² = 0.08), problem solving (B = −0.6, SE = 0.2, T = -3.0, p = 0.003, df = 2926, q = 0.009, R² = 0.02), and personal-social domains (B = −0.4, SE = 0.1, T = -2.8, p = 0.005, df = 2941, q = 0.01, R² = 0.07); see Fig. 4. None of these relationships were moderated by sex.

Fig. 4.

Associations between postnatal maternal distress and child behaviour at 3 years. Higher postnatal distress was significantly associated with lower scores across all behavioural domains at 3 years of age: (A) communication (B = −0.4, SE = 0.2, T = -2.6, p = 0.01, df = 2986, q = 0.02, R² = 0.04), (B) gross motor (B = −0.6, SE = 0.2, T = -3.7, p < 0.001, df = 2988, q = 0.002, R² = 0.01), (C) fine motor (B = −0.9, SE = 0.3, T = -3.4, p = 0.001, df = 2953, q = 0.004, R² = 0.08), (D) problem solving (B = −0.6, SE = 0.2, T = -3.0, p = 0.003, df = 2926, q = 0.009, R² = 0.02), and (E) personal-social skills (B = −0.4, SE = 0.1, T = -2.8, p = 0.005, df = 2941, q = 0.01, R² = 0.07).

As with the 2-year models, age at scan, female sex, maternal education, prenatal social support, and gestational age at birth each showed significant positive associations with ASQ‑3 domain scores.

3.3. Maternal distress and child functional connectivity

After quality checking, 62 scans were retained for analysis. There were no significant associations between prenatal or postnatal maternal distress and child brain functional connectivity across the default mode, executive control, left frontoparietal, and sensorimotor networks (all p > 0.40). Findings were consistent when models were re-run using listwise deletion.

3.4. Child functional connectivity and behavioural outcomes at 2 and 3 years of age

No significant relationships were found between child brain functional connectivity and behavioural outcomes at 2 or 3 years (all p > 0.21). Findings were consistent when models were re-run using listwise deletion (all p > 0.12).

In 2-year models, greater gestational age at birth was positively associated with higher communication and problem solving scores across all brain connectivity models. In 3-year models, greater gestational age at birth was positively associated with higher communication scores across all brain connectivity models. Findings across 2- and 3-year models remained consistent using listwise deletion.

3.5. Supplementary analysis

Supplementary analyses of participants with ASQ-3 data at both 2 and 3 years of age showed largely similar patterns of association between maternal distress and child behaviour as those observed in the full sample.

At 2 years, higher postnatal maternal distress was related to lower child gross motor scores (B = −0.4, SE = 0.2, T = -2.2, p = 0.03, df = 2843, q = 0.04, R² = 0.01).

At 3 years, higher prenatal distress was associated with lower child communication (B = −0.7, SE = 0.2, T = -4.0, p < 0.001, df = 2917, q < 0.001, R² = 0.05), gross motor (B = −0.7, SE = 0.2, T = -3.8, p < 0.001, df = 2918, q < 0.001, R² = 0.01), and personal-social scores (B = −0.5, SE = 0.2, T = -3.1, p = 0.002, df = 2882, q = 0.006, R² = 0.07). Higher prenatal distress was associated with lower problem-solving scores at 3 years (B = −0.4, SE = 0.2, T = -2.2, p = 0.03, df = 2862, q = 0.07, R² = 0.02), though this relationship did not remain significant after FDR correction.

As well, higher postnatal maternal distress was related to lower scores across five ASQ-3 domains at 3 years of age: communication (B = −0.4, SE = 0.2, T = -2.7, p = 0.008, df = 2843, q = 0.02, R² = 0.05), gross motor (B = −0.6, SE = 0.2, T = -3.7, p < 0.001, df = 2843, q = 0.002, R² = 0.01), fine motor (B = −1.0, SE = 0.3, T = -3.6, p < 0.001, df = 2818, q = 0.002, R² = 0.08), problem solving (B = −0.6, SE = 0.2, T = -2.9, p = 0.004, df = 2789, q = 0.01, R² = 0.03), and personal-social domains (B = −0.4, SE = 0.1, T = -2.7, p = 0.008, df = 2807, q = 0.02, R² = 0.07).

3.6. Sensitivity analysis

Sensitivity analyses were conducted to determine whether applying the imaging subsample exclusion criteria to the full behavioural sample altered associations between maternal distress and child behaviour. In total, 85 infants with neurological and/or genetic conditions and 413 infants born preterm (<37 weeks gestation) were removed from the restricted sensitivity sample.

At 2 years, there were no significant associations between prenatal or postnatal maternal distress and child behavioural outcomes after FDR correction in the restricted sensitivity sample. The association between postnatal maternal distress and gross motor scores, which was significant in the main full-sample analysis, was significant before correction in the restricted sample but did not survive FDR correction.

At 3 years, higher prenatal maternal distress was significantly associated with lower communication (B = −0.5, SE = 0.1, T = -3.1, p = 0.002, df = 3254, q = 0.009, R² = 0.04), gross motor (B = −0.4, SE = 0.2, T = -2.7, p = 0.007, df = 3255, q = 0.02, R² = 0.008), and personal-social scores (B = −0.4, SE = 0.1, T = -3.1, p = 0.002, df = 3197, q = 0.009, R² = 0.07). Higher postnatal maternal distress was significantly associated with lower gross motor (B = −0.5, SE = 0.2, T = -3.2, p = 0.001, df = 2794, q = 0.01, R² = 0.01) and fine motor scores (B = −0.8, SE = 0.3, T = -3.0, p = 0.003, df = 2761, q = 0.01, R² = 0.08). Associations between postnatal distress and problem-solving and personal-social scores were significant before correction but did not survive FDR correction.

In the restricted supplementary domain-matched sample, there were no significant associations at 2 years after FDR correction. The postnatal distress–gross motor association, which was significant in the full domain-matched analysis, was significant before correction but did not survive FDR correction in the restricted sample. At 3 years, higher prenatal maternal distress was significantly associated with lower communication (B = −0.4, SE = 0.2, T = -2.8, p = 0.006, df = 2726, q = 0.03, R² = 0.04) and gross motor scores (B = −0.5, SE = 0.2, T = -3.0, p = 0.003, df = 2727, q = 0.03, R² = 0.009). Higher postnatal maternal distress was significantly associated with lower gross motor (B = −0.6, SE = 0.2, T = -3.4, p < 0.001, df = 2659, q = 0.006, R² = 0.01) and fine motor scores (B = −0.9, SE = 0.3, T = -3.2, p = 0.002, df = 2635, q = 0.008, R² = 0.08). Associations between prenatal distress and personal-social scores, and postnatal distress and problem-solving scores, were significant before correction but did not survive FDR correction.

4. Discussion

Here, we show that maternal psychological distress during the perinatal period is associated with multiple domains of child behaviour at 3 years, with some associations evident at 2 years. Child brain functional connectivity showed no significant associations with either maternal distress or behaviour.

The associations between maternal distress and child behavioural outcomes appeared to strengthen with age. At 2 years, higher postnatal maternal distress was significantly associated only with lower gross motor scores, while prenatal distress was not associated with any behavioural domains. By 3 years, however, both prenatal and postnatal distress were robustly associated with lower scores across all five ASQ-3 domains. These findings are consistent with prior research suggesting that the effects of maternal psychological distress may become more pronounced as children get older. For example, Wei et al. (2023) found negative associations between maternal anxiety and depression during late pregnancy and child ASQ-3 scores at 2, 6, 12, and 24 months, with larger effects at later time points. While age-related strengthening of associations is consistent, the distress measures used by Wei et al., the Self-Rating Anxiety Scale and the Center for Epidemiologic Studies Depression Scale, emphasize current somatic and psychological symptoms of distress (Dunstan and Scott, 2020, Carleton et al., 2013), which may be more associated with early behaviours, than the cognitive-emotional aspects of depression and anxiety emphasized by the PROMIS and EPDS scales used in our study (Bergink et al., 2011, Cella et al., 2019). Consistent with this pattern, sustained maternal anxiety from pregnancy through child age 3 years has been linked to developmental delays in communication and personal-social domains (Mughal et al., 2019), while prenatal depression has been associated with lower cognitive, language, and motor outcomes at age 3 (Chen et al., 2025). Taken together, these results suggest that influence of maternal distress on child behaviour becomes increasingly pronounced with age.

At 2 years, postnatal maternal distress was only associated with gross motor development. This early domain-specific association is consistent with evidence showing that motor development is particularly sensitive to the caregiving environment. For example, Nasreen et al. (2013) found that postpartum depression predicted lower gross motor scores at 6–8 months using a milestone-based scale adapted from Yaqoob et al. (1993), and Cornish et al. (2005) reported similar effects in performance at 15 months using the Bayley Scales of Infant Development‑II (BSID‑II). Since motor development relies heavily on caregiver interaction and physical engagement in early life, maternal distress may reduce opportunities for stimulation and play (Pedersen and Hansen, 2022, Riethmuller et al., 2009), which may partly account for the associations we observed. The lack of other developmental domains associated with postnatal distress in our study aligns with some prior work finding no significantly higher risk of developmental delay across ASQ-3 domains at age 2 in children of mothers with postnatal distress (Sindalovsky et al., 2025). Similarly, other research has reported no association between postnatal depression and ASQ-3 scores across the first two years of life (Zhang et al., 2023). In contrast, other studies found that higher postnatal distress was associated with widespread behavioural difficulties at age 2, including externalizing, internalizing, hyperactivity, emotional, and conduct problems (Netsi et al., 2018, Soe et al., 2016). These mixed findings may reflect differences in the sensitively developmental assessment tools. Measures such as the ASQ-3 may be more effective at detecting behavioural differences once children reach a stage where regulatory and social capacities are more developed and observable; at 2 years, some of these capacities may not yet be fully expressed, which could contribute to the lack of significant associations in other domains. Regardless, these findings suggest that motor skills may be especially sensitive to postnatal influences in early childhood, while broader effects of maternal distress may become more evident and observable later in development.

By age 3, the influence of maternal distress extended beyond motor ability, with significant associations found between prenatal distress and lower scores across all ASQ-3 behavioural domains. The association between prenatal distress and communication was moderated by child sex, with a stronger negative relationship observed in males; no such moderation was evident for the other domains. Large-scale studies have shown that higher prenatal distress is linked to lower ASQ-3 scores in children at age 3, affecting communication, motor, problem solving, and personal-social behaviours, consistent with our findings (Chen et al., 2025, Duguay et al., 2022, Wei et al., 2023, Zhang et al., 2023). One possible explanation is that as children mature and demands across cognitive, language, and social domains increase, behavioural differences become more evident and measures such as the ASQ-3 are better able to capture them. Accordingly, by age 3, maternal distress was associated with more widespread behavioural outcomes rather than only domain-specific effects.

Similar to the associations observed with prenatal distress, higher postnatal distress was also associated with lower performance across all behavioural domains at age 3. This result is consistent with prior literature revealing that postnatal depression was associated with increased risk of delays in personal-social behaviours, gross and fine motor ability, adaptive skills, and language production (Lubotzky-Gete et al., 2021, Mughal et al., 2019, Ibanez et al., 2015). Notably, Santos et al. (2024) reported that persistent postnatal depressive symptoms (EPDS ≥ 10 at all three follow-ups in the first three years) were associated with significantly lower child ASQ-3 scores at age 3. Children of mothers with sustained depression scored about 14 points lower on overall developmental scores compared to children of non-depressed mothers, a difference equivalent to 0.27 standard deviations on the ASQ-3 and therefore considered a small effect size. Our findings revealed similarly small effect sizes, with the direction and overall pattern of associations between postnatal distress and lower ASQ-3 performance at age 3 aligning with prior literature. As mentioned, the ASQ-3 may be particularly sensitive to capturing these differences at this age, when cognitive, language, and social demands become more complex, thereby allowing such developmental difficulties to be more readily detected than at younger ages. Evidently, our findings demonstrated that postnatal maternal distress was associated with widespread developmental difficulties at 3 years of age, reinforcing the importance of supporting maternal mental health during the postnatal period. Supplementary analyses in the subset of participants with ASQ-3 data at both 2 and 3 years further supported the primary findings, and suggest discrepancies are not related to sample differences.

When comparing the relative influence of prenatal and postnatal maternal distress, our findings suggest that prenatal distress exerts a more global and widespread effect on early child behaviour. At 2 years, postnatal distress was the only factor significantly associated with gross motor outcomes. By 3 years, however, prenatal distress was consistently related to lower scores across all five ASQ-3 domains, with associations that were both robust and broad in scope. Postnatal distress also predicted lower scores at this age, but the effect sizes were less consistent across domains, whereas prenatal distress exerted a more uniform and statistically robust influence across all behaviours. Kingston et al. (2015) review aligns with this pattern, showing that prenatal distress predicted toddler cognitive outcomes with more consistent medium-sized effects, while findings for postnatal distress were more variable. These findings lend support to the importance of prioritizing screening and intervention during pregnancy, as reducing maternal distress in the prenatal period may be particularly helpful for supporting child development. However, prenatal and postnatal distress were moderately correlated in the full PdP sample (r = 0.52, p < 0.001), with similar correlations observed at 3, 6, and 12 months postpartum, suggesting continuity in distress symptoms across the perinatal period. Therefore, postnatal distress remains highly relevant, as it was still significantly associated with poorer outcomes, highlighting the need for ongoing maternal mental health support across both the prenatal and postnatal periods.

We found no associations between maternal distress and brain connectivity at 2 years. Prior studies of prenatal distress have often focused on amygdala connectivity, showing negative associations with cortical and subcortical regions in infants and children up to 4 years of age (Donnici et al., 2021, Hill et al., 2025, Posner et al., 2016, Scheinost et al., 2020, Soe et al., 2017); other studies found increased connectivity with the amygdala (Rotem-Kohavi et al., 2019, Qiu et al., 2015). Postnatal distress has also been associated with altered connectivity in frontal, sensory, and language-related networks among 4- to 7-year-old children (Uy et al., 2023, Farah et al., 2020, Wang et al., 2019, Zhang et al., 2021), although null findings have also been reported (Donnici et al., 2021, Soe et al., 2017). In an overlapping sample of participants, we previously showed that higher distress symptoms related to higher auditory network connectivity in 3-month-old infants (Jaffer et al., 2024). This finding was specific to the auditory network, which develops early and is largely complete in infancy, and is distinct from the networks assessed at 2 years of age. At 2 years, our analyses examined large-scale brain networks (default mode, executive control, left frontoparietal, and sensorimotor) that do not directly include the amygdala. Connectivity within these networks appeared comparable across children exposed to varying levels of maternal distress, a finding that may be somewhat reassuring, though behavioural difficulties were still observed. This suggests that early organization of these functional networks, as captured in the brain regions examined, may be relatively resilient to prenatal distress, while other pathways likely contribute to the emergence of behavioural challenges. Given that functional connectivity between data-derived brain regions at 2 years of age did not relate to maternal distress or child behavioural outcomes, other factors such as infant sleep (Graven and Browne, 2008, Tham et al., 2017, Jiang, 2019) and maternal physical activity (Yang et al., 2021) may offer alternative explanations and warrant further exploration. Notably, associations between perinatal distress and child behaviour were more consistent and widespread by 3 years, raising the possibility that brain-behaviour relationships may also strengthen with age.

The associations reported here, while modest, contribute to growing evidence supporting the need for systematic identification and support of maternal mental health concerns during both pregnancy and the early postnatal period. The effect sizes in these studies were small, displaying subtle shifts in behavioural scores (B = −0.4 to −0.7 per unit increase in distress), emphasizing the need for continued longitudinal follow-up to assess whether these early behavioural differences persist and to determine which interventions may be most effective in supporting maternal well-being and promoting healthy developmental outcomes in children.

Our study has several limitations. Our sample was predominantly white, highly educated, and from high-income households. Therefore, our findings may not be generalizable to more diverse sociodemographic contexts. In addition, although the full behavioural sample included some children born preterm or with neurological and/or genetic conditions, the imaging subsample was restricted to children without these factors. Sensitivity analyses applying comparable exclusion criteria to the behavioural sample showed that the direction of associations between maternal distress and child behaviour was generally consistent, with higher maternal distress associated with lower ASQ-3 scores. However, fewer associations survived FDR correction in the restricted sample, suggesting that preterm birth and neurological and/or genetic conditions may modestly strengthen some associations in the broader full sample. These factors may independently influence neurodevelopmental outcomes and should be considered in future work examining relationships between maternal distress and child development. Additionally, all measures of maternal distress were based on self-report questionnaires that capture only recent, short-term symptoms (i.e., past 7 days). Behavioural assessments may also be influenced by maternal perceptions, as more distressed or depressed mothers might rate their child’s behaviour differently (Briggs-Gowan et al., 1996, Lee and Hans, 2015), potentially introducing bias. While third-party assessments of child behaviour were not available in the PdP dataset at this age group, parent-report measures remain valuable and can provide clinically useful information, particularly at this age when children are not always easily assessed across multiple settings and domains (Kerr et al., 2007, Miller et al., 2017). Another limitation relates to our analysis strategy. We focused on average connectivity between two major nodes within each network to characterize large-scale functional organization. While this approach enables comparison across networks, it may have lacked sensitivity to detect more region-specific effects, particularly in structures such as the amygdala and hippocampus, which have been the focus of prior seed-based studies. Lastly, other factors, including individual-level variations in network locations and variability related to the videos watched during fMRI, may also have influenced the imaging findings. Future studies using longitudinal measures of maternal distress, performance-based assessments of child behaviour, and broader or more individualized network analyses will enhance our understanding of relationships between maternal distress, brain development, and child outcomes. Longitudinal neuroimaging data may also facilitate future investigation of whether maternal distress influences trajectories of brain maturation across early childhood.

5. Conclusion

In conclusion, elevated maternal anxiety and depression symptoms are associated with small but significantly lower behavioural scores in children at 3 years, while the association between postnatal distress and gross motor function was the only significant relationship at 2 years of age. Prenatal and postnatal distress showed relationships with child outcomes, highlighting the importance of supporting maternal mental health during the perinatal period. The lack of associations between maternal distress and child functional brain connectivity at age 2 highlights a potentially complex relationship between early life experiences and brain-behaviour development, at least in the brain regions analysed. From a public health perspective, our findings reinforce the importance of identifying and addressing maternal psychological distress during the perinatal period. Our findings motivate early, targeted interventions that support maternal mental health may also support better developmental trajectories in children.

CRediT authorship contribution statement

Aliza Jaffer: Writing – review & editing, Writing – original draft, Visualization, Validation, Methodology, Formal analysis, Conceptualization. Kathryn Y. Manning: Writing – review & editing, Validation, Methodology, Conceptualization. Gerald F. Giesbrecht: Writing – review & editing, Methodology, Funding acquisition. Tim F. Oberlander: Writing – review & editing, Supervision, Funding acquisition. Lianne Tomfohr-Madsen: Writing – review & editing, Methodology, Funding acquisition. Catherine Lebel: Writing – review & editing, Supervision, Methodology, Funding acquisition, Conceptualization.

Disclosures

The authors have no relevant financial or non-financial interests.

Funding sources

This research was supported by the Owerko Centre, the Alberta Children’s Hospital Research Institute, and a Canadian Institutes for Health Research grant (PJT-462004). Salary support was provided by the Canada Research Chair program.

Declaration of Competing Interest

CL holds a research grant from GE Healthcare. The other authors declare no real or potential conflicts of interest.

Data availability

Perinatal mental health data is available at: https://osf.io/ha5dp . Neuroimaging data will be made available on request.