COVID-19 infection during pregnancy and infant neurodevelopment

Abstract

INTRODUCTION:

The impact of maternal SARS-CoV-2 infection on fetal brain development during pregnancy remains unclear. Prior research has associated other antenatal infections with adverse neurodevelopmental outcomes in offspring.

OBJECTIVE:

To compare neurodevelopmental outcomes in infants born to mothers infected with SARS-CoV-2 during pregnancy (COVID+) to infants without congenital exposure (COVID−).

METHODS:

This study included 77 COVID+ infants and 157 COVID− infants assessed at 6 and/or 12 months. Outcomes were based on maternal self-report, observed infant behavior and brain fMRI.

RESULTS:

Overall, COVID+ and COVID− infant groups showed no significant differences across a range of neurobehavioral measures. However, analyses not adjusted for multiple comparisons revealed differences: fewer night awakenings at 6 (t(154) = 2.24, p < 0.03) and 12 months (t(107) = 1.94, p < 0.05), and reduced duration of orienting at 12 months (t(55.38) = 2.15, p < 0.04) in COVID+ infants. Neural differences were noted in posterior-anterior midline, insular-frontal, insular-posterior cingulate, and frontal-cingulate regions at an uncorrected threshold of p < 0.01.

CONCLUSION:

This study of multi-level infant development suggests that infants born to mothers infected with COVID during pregnancy are not experiencing harmful effects of that exposure.

INTRODUCTION

The potential for neurodevelopmental sequalae in offspring prenatally exposed to SARS-CoV-2 is a significant concern. Even a slight increase in the risk of adverse neurodevelopmental outcomes would have a considerable public health impact, due to the high number of exposed individuals, which as of May 2025, was over 103 million in the United States and over 778 million worldwide.¹ Obstetric populations are a critical focus, as heightened inflammatory response in both mother and fetus in response to SARS-CoV-2 has been reliably documented,^2,3 and maternal immune activation (MIA) is a known risk factor for offspring neurodevelopmental morbidity.⁴

Over the past four years, studies around the world have addressed the biological and medical impacts of the virus in pregnancy. Studies show that vertical transmission to the fetus is extremely rare.^5,6 However, compared to non-pregnant people, pregnant individuals tend to be more ill when infected, less responsive to treatment, experience worse outcomes, and mount a less robust immune response to the virus.^7,8 Fortunately, however, the placenta appears to serve as an effective barrier against direct transmission of the virus to the developing fetus.⁹ Evidence of placental resistance to the virus comes from studies showing a lack of virus in the placenta and from studies showing that SARS-CoV-2 receptor angiotensin-converting enzyme 2 (ACE2) and spike transmembrane serine protease 2 (TMPRSS2) expression patterns do not indicate support for viral cell entry.¹⁰ Along with this, the placenta mounts a robust inflammatory response to the virus, which is seen, for example, in elevated immune and interferon-stimulated genes.⁷ Cord blood and neonatal biological samples corroborate a similar pattern, with robust immune activation paired with lack of virus in cord blood or neonatal biospecimens. Specifically, congenitally exposed neonates show increased innate immune cells and cytokine expression and a response pattern that suggests mild, non-specific immune imprinting. That is, cord blood transplantal antibody transfer is significantly lower than is observed in other pathologies, and while maternal immunoglobulin G (IgG) transfer occurs, there is a lack of SARS-CoV-2-specific IgM.¹¹ Because IgG reflects past or long-term immunity, whereas IgM reflects acute or recent infection, this pattern is more likely to reflect carry-over immune activation rather than a newly mounted immune response. Thus, the overall picture is robust immune activation in the fetal compartment without direct viral exposure or mounted viral response in the developing fetus or neonate.

Despite concern that MIA may result in developmental problems in infants congenitally exposed to SARS-CoV-2, evidence for this is mixed at best. A large-scale, electronic medical record study led by Edlow and colleagues (2022) reported higher incidence of neurodevelopmental diagnoses in N=222 infants born to SARS-CoV-2 positive mothers compared to controls.¹² However, the authors highlighted that confounds such as higher rates of prematurity in the congenitally exposed group may have contributed to observed differences. They concluded that further research was needed. Another study by Ayed and colleagues (2022) reported favorable outcomes in 298 infants born to mothers with SARS-CoV-2 infections.¹³ They observed a lower prevalence of developmental delay, as reported by the Ages and Stages Questionnaire, Third Edition (ASQ-3), in their congenitally-exposed sample than would be expected in a general population, 10% versus 12–16%.¹⁴ A study by Aldrete-Cortez and colleagues (2022) reported abnormal levels of fidgety movements in congenitally exposed infants and concluded that these infants may be at risk for future developmental disorders, but it should also be noted that this was a modestly sized sample of 28 exposed infants, so further examination of this topic is warranted. A study by Shuffrey and colleagues (2022) that directly compared ASQ-3 outcomes in infants with (N=114) and without (N=141) congenital SARS-CoV-2 exposure reported no significant differences between groups and no relationship between illness severity and ASQ-3 outcome measures.¹⁵ In a separate study of the same cohort, no significant differences in Developmental Assessment of Young Children, second edition (DAYC-2) outcomes at 5–11 months, and no effect of illness severity or trimester of exposure were reported.¹⁶ A study by Werchan and colleagues (2022) similarly found no significant main effect of congenital exposure on either observed infant attention or Brief Infant-Toddler Social and Emotional Assessment (BITSEA) measures at both 6 and 12 months (N=167).¹⁷ Together, these studies indicate that developmental problems are an unlikely result of congenital SARS-CoV-2 exposure; however, the studies described above focused on a rather circumscribed set of variables, and there are many remaining knowledge gaps across a wider array of infant outcomes.

Another important question is whether brain structure or function is impacted by congenital SARS-CoV-2 exposure. Available studies using fetal neurosonography¹⁸ and fetal MRI¹⁹ have shown no fetal brain differences when comparing pregnancies complicated to those not complicated by SARS-CoV-2 infection. A study of fetal umbilical vein flow, similarly, showed lack of differences between groups.²⁰ However, a recent study of infant neuroanatomical development reported differences in cortical gray matter volume and sulcal depth in 47 congenital SARS-CoV-2 infants. This study, by Andescavage and colleagues, reported increased cortical gray matter volume and greater sulcal depth in congenitally exposed infants compared to controls.²¹ Thus, available neuroimaging evidence suggests little or no apparent differences in brain structure or cerebral blood flow between groups, and where one study observed differences, the pattern was a more mature developmental profile in congenitally exposed infants. Whether SARS-CoV-2 infection during pregnancy relates to infant brain functional architecture has yet to be examined and is one of the novel focuses of the current study.

The present study addresses the question: Does congenital SARS-CoV-2 exposure relate to variation in infant neural and behavioral outcomes? We report on 234 mother-infant dyads who provided complete prenatal data and participated in infant neurobehavioral assessment(s). Seventy-seven infants in the sample, or 33%, were born to mothers who were infected with SARS-CoV-2 during pregnancy. This work is designed to provide a comprehensive basis of evidence across fundamental domains of infant development, including milestones, attention, sleep, temperament, and socioemotional development. We evaluate these domains using both parent report and behavioral observation-based measures that are standards within the developmental psychology literature. We also evaluate potential contributions of sociodemographic factors, maternal psychological health factors, differences in testing conditions, illness severity, timing of infection and infant birth outcomes. Our MRI analysis targets whole-brain functional connectivity with a focus on major subnetworks and their within and between network connectivity. Like the behavioral data, this approach is comprehensive given the exploratory nature of the work. Our goal in taking this broad bottom-up approach is to provide a comprehensive evidence base for future studies that aim to replicate existing findings or generate novel hypotheses regarding specific neurobehavioral domains that may be impacted.

MATERIALS AND METHODS

Study population

The New York COVID-19 and Perinatal Experiences (NYU-COPE) cohort is one of a collection of prospective pregnancy cohorts around the world that were initiated between March 2020 and May 2020,²² many of which were harmonized to use common measures.²³ Participant recruitment occurred predominately through the New York University Langone Medical Center electronic health records (EHR), with recruitment beginning April 7, 2020. Here we report on a sample of 234 mother-infant dyads who provided complete prenatal data and participated in at least one infant neurobehavioral assessment. Families opted into neurobehavioral assessments that occurred at ages: newborn (MRI), 6 months (development) and 12 months (MRI and development). Eighty-three dyads completed newborn (M age = 2.53 months, SD = 0.86 months; N=53) and/or infant 12-month MRI (M age = 12.36 months, SD = 1.55 months; N=30). 234 dyads completed neurobehavioral assessment at infant age 6-months (N=194; Mage=6.57 months, SD=0.97 months) and/or 12-months (N=167; M age=12.57 months, SD=0.58 months). Congenital SARS-CoV-2 exposure was defined using EHR (ICD-10 code U07), self-report, and/or as disclosed on the Novel Coronavirus Illness Patient Report (NCIPR)²⁴ measure. Women never infected with the virus and those who were infected before or after pregnancy comprise our comparison group. For the full sample, 77 infants meet criteria for congenital SARS-CoV-2 exposure (33%), and 157 comprise our comparison group (67%). Table 1 provides overview of the sample and compares exposed and comparison groups across demographic factors, psychological health, testing conditions, and MRI data quality. Supplemental Material, Figs. S1 and S2 provide dates of infection and dates of study enrollment, respectively. The study was approved by the Institutional Review Board of New York University Grossman School of Medicine. All mothers provided informed consent to participate in the study.

Table 1.

Group comparisons across demographic factors, psychological health, testing conditions, and MRI data quality.

	Full sample (n = 234)	COVID+ (n = 77)	COVID− (n = 157)	COVID+ vs. COVID−
Variable	M (SD) or N (%)	M (SD) or N (%)	M (SD) or N (%)	Statistical resultsa
Demographics
Maternal age (years)	33.77 (4.15)	33.33 (4.33)	33.99 (4.05)	p = 0.26
Family incomeb	10 (5)	10 (4)	11 (5)	p = 0.86
Infant sex at birth (% male)	122 (52%)	38 (49%)	84 (54%)	p = 0.76
Infant gestational age at birth (weeks)	38.87 (1.85)	39.01 (1.61)	38.81 (1.96)	p = 0.45
Infant birthweight (pounds)	7.34 (1.41)	7.44 (1.12)	7.31 (1.53)	p = 0.64
Infant age at 6-month assessment (months)	6.57 (.97)	6.50 (0.98)	6.60 (0.97)	p = 0.52
Infant age at 12-month assessment (months)	12.57 (0.58)	12.58 (0.61)	12.57 (0.58)	p = 0.95
Infant age at MRI visit (months)	5.89 (4.81)	6.21 (5.26)	5.81 (4.74)	p = 0.78
Infant ethnicity (% Hispanic/Latine)	55 (24%)	16 (21%)	39 (25%)	p = 0.58
Infant race
White	138 (59%)	55 (71%)	83 (53%)	p = 0.02
More than one race	38 (16%)	9 (12%)	29 (18%)	p = 0.26
Black	17 (7%)	5 (6%)	12 (8%)	p = 0.83
Other	17 (7%)	0 (0%)	15 (10%)	p=1.0
Asian	13 (6%)	2 (3%)	11 (7%)	p = 0.30
Unreported	11 (5)	6 (8%)	7 (4%)	p=1.0
Maternal psychological health
Prenatal BSI anxiety symptoms average	1.64 (0.69)	1.56 (0.77)	1.69 (0.53)	p = 0.20
Prenatal BSI depressive symptoms average	1.58 (0.65)	1.50 (0.71)	1.63 (0.55)	p = 0.21
6-month postpartum EPDS score	6.32 (4.52)	6.13 (3.91)	6.41 (4.84)	p = 0.78
Visit completion
6-month behavioral visit participation	194 (83%)	67 (87%)	127 (81%)	p = 0.32
12-month behavioral visit	167 (71%)	53 (69%)	114 (73%)	p = 0.65
MRI data	73 (32%)	16 (21%)	57 (36%)	p = 0.02
Testing parameters
Behavioral assessment device (% smartphone)	58 (28%)	21 (31%)	36 (26%)	p = 0.51
MRI motion	0.001 (0.005)	0.004 (0.004)	0.001 (0.006)	p = 0.54
MRI frame count	402 (82)	395 (90)	405 (81)	p = 0.40

BSI Brief Symptom Inventory, CESD Center for Epidemiologic Studies Depression, EPDS Edinburgh Postnatal Depression Scale.

^aT-tests was used for continuous variables and chi-square tests were used for categorical variables (race/ethnicity)

^bIncome was coded as 1 ≤ 10k, 2= 10–20k, 3 = 20–30k, 4 = 30–40k, 5 = 40–50k, 6 = 50–60k, 7 = 60–80k, 8 = 80–100k, 9= 100–120k, 10= 120–140k, 11 = 140–160k, 12= 160–180k, 13 = 180–200k, 14 = 200–220k, 15 = 220–250k, 16 = 250k+.

Maternal measures

Questionnaires completed by mothers addressed demographic information, household income, psychological health, and COVID illness in self and in family members. COVID-19 symptoms, medical complications, home and hospital treatments, lasting symptoms, anxiety about illness, employment impacts, quarantine behaviors, vaccine-related behaviors and effects, and illness of other family/household members were assessed using the Novel Coronavirus Illness Patient Report (NCIPR) Survey.²⁴ Included in the NCIPR is also a question about perceived illness severity, where higher values represent more severe illness: 1=very mild, 2=mild to moderate; 3=moderate to severe, 4=severe to extreme. Only a subset of participants in the sample completed the NCIPR, and, as such, comparisons using the NCIPR, include smaller sample sizes. Psychological health was assessed using the Brief Symptom Inventory short form (BSI-18), a measure of psychological distress, inclusive of depression, anxiety, and somatization,²⁵ the Center for Epidemiologic Studies Depression (CESD) Scale, a measure of depressive symptomatology,²⁶ and the Edinburgh Postnatal Depression Scale (EPDS), designed to assess depression during the peripartum period.²⁷

Neurodevelopmental assessment

Children’s neurodevelopment was assessed at 6- and 12-months using a remote testing protocol that included parent-report surveys and behavioral testing conducted over Zoom. Families completed the Zoom-based behavioral visit using either laptops (73%) or smartphones (27%). Observed tasks were performed in the same order for all subjects, and testing staff was blinded to COVID exposure status. Behavioral assessments at 6 months included a Visual Attention Task,¹⁷ Still-Face Task,²⁸ and Arm Restraint Task.^7,28 At 12-months the Arm Restraint Task was administered. Questionnaires completed by parents, included the Ages and Stages Questionnaire (fine motor, gross motor, communication, problem-solving, and personal-social scales of development),¹⁴ the Brief Infant-Toddler Socioemotional Assessment (BITSEA; socioemotional competence, problems, and autism risk),²⁹ the Brief Infant Sleep Questionnaire (BISQ; night sleep duration and awakenings),³⁰ and the Infant-Behavior-Questionnaire-Revised Very-Short-Form (IBQ-R VSF; negative affect, surgency, and regulatory capacity, as well as duration of orienting and soothability exploratory subdomains of the regulatory capacity dimension, see Supplemental Materials for subscale reliability statistics, Table S1).^31,32 Supplemental Materials also include more detailed overview of the self-report measures of infant development (Table S2) and behavioral assessments (Table S3) used at 6 and 12 months.

Infant behavioral coding

Infant visual attention and emotional regulation behaviors were coded using artificial intelligence (AI) and manual evaluation, respectively. Specifically, Online Webcam-Linked Eye Tracker (OWLET), an algorithm developed by our group to automatically quantify Infant gaze behavior from video, was used to derive two infant visual attention measures: (i) total looking time and (ii) longest look duration during the Visual Attention task. Enhanced description of OWLET procedures and validation steps, including calculation of visual angle and robustness of the procedure to device and demographic characteristics is addressed in prior work.³³ Details on manual coding of infant emotion regulation and reactivity during the Arm Restraint and Still-Face Tasks is provided in Supplementary Materials. Personnel coding the tasks were blinded to COVID status.

Infant behavior statistical analyses

Welch’s t-tests, performed in RStudio 2021.9.2,³⁴ were used to examine differences in infant outcomes based on congenital SARS-CoV-2 exposure status. Distributional characteristics of all variables were evaluated to confirm appropriateness of parametric testing prior to analyses (see Supplementary Materials). In secondary analyses, associations between illness severity and timing of infection were evaluated within the N=77 COVID+ group using linear regression and analysis of variance, respectively. Sample sizes for these analyses were N=52 with illness severity information and N=42 with trimester of infection information. Among those reporting on trimester of infection, 11 cases (21%) were infected with SARS-CoV-2 in the first trimester, 15 cases (29%) in the second trimester, and 16 cases (31%) in the third trimester.

MRI data collection

Eighty-three infant MRI studies were conducted on a Siemens 3 T scanner, using a 32-channel head coil, with following acquisition parameters: TR=2000 ms; multi-echo TE=13, 30.88, 48.16, 65.44 ms; flip-angle: 75 degrees; slice-gap: none; voxel-size: 2.8 × 2.8 × 2.8 mm³; matrix-size: 68 × 68 × 44 voxels. MRIs were acquired during natural infant sleep.

MRI statistical analyses

MRI data preprocessing is described in Supplemental Materials. The primary approach to understanding potential brain-wide differences in connectivity between congenital COVID (COVID+) and COVID− groups was to evaluate connectivity between global brain networks. The GIFT Toolbox v4.0c (http://icatb.sourceforge.net) was used to compute ICA components representing large-scale infant brain intrinsic connectivity networks (ICNs). The number of components estimated was 35, based on the image quality using the modified minimum description length (MDL) discipline.³⁵ Reliability and stability of the algorithm was ensured using ICASSO by repeating the component estimation 20 times.³⁶ Among the 35 components, 28 components were identified as meaningful infant brain ICNs, according to low-to-high frequency ratio and based on anatomical plausibility of spatial maps. COVID+ and COVID− group differences were tested using two-sample t-tests executed within GIFT’s Functional Network Connectivity (FNC) tool at a significance threshold of p < 0.01. Outliers were screened and excluded based on FNC between-network z-scores outside the range of -2.5 to 2.5. ICN pairs showing significant between-group differences were examined further using CONN³⁷ seed-based connectivity analyses (SCA), with age included as a covariate in the model. For SCA, significant FCNs were binarized and submitted as seed regions of interest. In CONN, data were band-pass filtered at 0.008 Hz < f < 0.09 Hz.

RESULTS

Sociodemographic comparisons of COVID+ and COVID – groups

Groups did not differ in sociodemographic variables, including maternal age, household income, and infant sex (p’s > 0.26). In addition, groups did not differ in infant birth outcomes, including gestational age at birth and birth weight (p’s > 0.45). Testing conditions also did not differ between groups, including infant age at time of assessment, infant remote assessment device type, infant MRI motion parameters, or number of MRI frames included in analyses (p’s > 0.4). Further, maternal self-reported psychological distress did not differ between groups (p’s > 0.2). However, significant differences in race were observed, such that there were more white infants in the COVID+ group (p < 0.02). Full description of sociodemographic, health, and testing condition comparisons between groups, and for the full sample, is provided in Table 1. Results from a secondary sensitivity analysis conducted in a propensity-matched sample, matched on infant race/ethnicity, family income, maternal education, and maternal psychological characteristics, are provided in Supplementary Material Table 4.

Observed behavioral outcomes in COVID+ infants

COVID+ and COVID− infants did not differ across behavioral domains examined using the arm restraint task, the still face task, or the visual attention task. Coded responses included measures of looking time, avoidance, orientation to mother, self-soothing, and negative affect. Results of Welch’s two-sample t-tests for each measure are provided in Fig. 1. Estimates of effect size for each comparison are visualized in Fig. 2 and reported in Table 2.

Fig. 1. Box plots depicting paired comparison of COVID+ and COVID− groups across observed infant behaviors using Welch’s t-tests.

There were no significant differences between congenital COVID exposed (COVID+), depicted in pink, and non-congenitally exposed infants (COVID−), depicted in purple, in behavioral outcomes at 6- and 12-months. Observed infant behavior was measured across domains of attention (free-viewing attention task), and emotional reactivity and regulation (Still Face and Arm Restraint Tasks). For each plot, the box represents the interquartile range and whiskers represent 1.5x the interquartile range. Measures have been coded so that favorable outcomes are in the same direction across variables for ease of interpretation.

Fig. 2. Estimated effect size for self-report and observed behavioral dimensions.

In a, the 22 unique behavioral dimensions assayed by maternal self-report are represented as circles with sizes corresponding to Cohen’s d effect size. Similarly, panel B, provides the same Cohen’s d effect size representations for the 9 unique observed behavioral dimensions. In a, b the direction of effect is denoted by the color of the circle and nominal significance (p < 0.05, uncorrected) is denoted by which column the circle resides in. Cohen’s d and p values are provided in Table 2. For contextualization, a Cohen’s d value of 0.2 represents a small effect, 0.5, a medium effect, and 0.8 a large effect. A Cohen’s d of 0.5 indicates a difference of 0.5 standard deviations between group means. SE social-emotional.

Table 2.

Paired comparison results for full sample using Welch’s t-test.

Variable	t	df	p	FDR	Cohen’s d
Behavioral measures
Longest look duration	1.61	82.90	0.11	0.50	0.30
Total looking time	1.63	94.83	0.11	0.50	0.28
Avoidance – arm restraint 6 m	0.12	80.96	0.90	0.99	0.02
Avoidance – arm restraint 12 m	0.75	44.13	0.46	0.79	0.18
Orientation to mother – still face 6 m	1.76	107.53	0.08	0.50	0.30
Self-comforting – still face 6 m	0.92	81.92	0.36	0.76	0.17
Negative affect – still face 6 m	1.18	99.23	0.24	0.68	0.20
Negative affect – arm restraint 6 m	−0.22	85.27	0.83	0.99	−0.04
Negative affect – arm restraint 12 m	−0.01	48.03	0.99	0.99	0.00
Maternal report measures
ASQ communication	−0.21	121.82	0.83	0.99	−0.03
ASQ fine motor	−0.45	117.41	0.66	0.99	−0.07
ASQ gross motor	0.41	116.23	0.68	0.99	0.07
ASQ problem solving	0.34	122.65	0.73	0.99	0.05
ASQ personal-social	−0.86	128.99	0.39	0.76	−0.13
Orienting regulation – 6m	−0.15	116.81	0.88	0.99	−0.02
Orienting regulation – 12 m	2.15	55.38	<0.04	0.50	0.47
Soothability – 6m	−1.52	119.03	0.13	0.51	−0.24
Soothability – 12 m	−0.01	73.80	0.99	0.99	0.00
Night awakenings – 6 m	2.24	154.10	<0.03	0.50	0.32
Night awakenings – 12 m	1.94	107.31	<0.05	0.50	0.32
Sleep duration – 6 m	0.96	133.34	0.67	0.99	0.11
Sleep duration – 12 m	0.22	88.67	0.82	0.99	0.04
Socioemotional competence	−1.23	84.69	0.22	0.68	−0.22
Socioemotional problems	0.76	80.61	0.45	0.79	0.14
Autism risk score	1.41	104.65	0.16	0.56	0.23
Regulatory capacity – 6m	−0.86	119.78	0.39	0.76	−0.14
Regulatory capacity – 12 m	1.63	77.44	0.11	0.50	0.30
Negative affect – 6 m	1.03	125.58	0.30	0.76	0.16
Negative affect – 12 m	0.06	92.60	0.95	0.99	0.01
Surgency – 6m	−0.93	131.03	0.35	0.76	−0.14
Surgency – 12 m	0.10	77.95	0.92	0.99	0.02

Neurodevelopmental outcomes in infants born to COVID+ and COVID− mothers. Measures in bold are statistically significant at p < 0.05, uncorrected for multiple comparisons.

Maternal report of infant behavior in COVID+ infants

COVID+ and COVID− infants did not differ in Ages and Stages (ASQ) communication, fine or gross motor, problem solving, or personal-social domains. Groups also did not differ in IBQ-R measures of soothability, regulatory capacity, negative affect, or surgency, or BITSEA measures of socioemotional problems/competence or Autism risk. Results varied with infant age for maternal report of infant duration of orienting, a temperamental measure of attention within the IBQ-R. Here, Welch’s sample t-tests revealed no difference in duration of orienting at 6 months (t(116.81) = −0.15, p = 0.88), but reduced infant duration of orienting in COVID+ infants at 12 months (t(55.38) = 2.15, p < 0.04). The other maternal report area where groups differed was in number of night awakenings; fewer night awakenings were reported for COVID+ cases at 6 (t(154) = 2.24, p < 0.03) and 12 months (t(107) = 1.94, p < 0.05). There were no group differences that remained statistically significant when False Discovery Rate (FDR) correction for multiple comparisons was applied across comparisons. As a control, we verified that there were no differences in breastfeeding rates between groups at either 6 months (p = 0.71) or 12 months (p = 0.69) that could account for nominal group differences in night awakenings. In total, significant group effects were observed for three measures, each with small to medium effect sizes (Cohen’s d 0.3–0.5).³⁸ Full report of behavioral maternal report measures is provided in Fig. 3. Estimates of effect size for each comparison are visualized in Fig. 2 and reported in Table 2. Distributions for significant effects are provided in Supplementary Fig. S3.

Fig. 3. Box plots depicting paired comparison of COVID+ and COVID− groups across maternal self-report measures of infant behavior using Welch’s t-tests.

Across 22 comparisons, three significant differences were observed: (1) night awakening frequency at 6 months, (2) night awakening frequency at 12 months, and (3) orienting regulation at 12 months. Values from infants congenitally exposed to SARS-CoV-2 (COVID+) are represented in pink, and non-congenitally exposed infants (COVID−) are depicted in purple. Questionnaires included the Ages and Stages Questionnaire (ASQ), the Brief Infant Sleep Questionnaire (BISQ), the Brief Infant-Toddler Socioemotional Assessment (BITSEA), and the Infant-Behavior-Questionnaire-Revised Very-Short-Form (IBQ). For each plot, the box represents the interquartile range and whiskers represent 1.5x the interquartile range. Measures have been coded so that favorable outcomes are in the same direction across variables for ease of interpretation.

Role of illness severity and trimester of infection

There were no significant associations between observed behavioral infant outcomes and severity of SARS-CoV-2 illness. Among outcomes measured by maternal questionnaires, maternal SARS-CoV-2 illness severity was positively related to IBQ negative affect at 12 m (n = 31, p < 0.01) (Supplemental Fig. S4); no significant association was observed for illness severity and other developmental domains, and the IBQ negative affect illness severity effect did not remain statistically significant when False Discovery Rate (FDR) correction was applied. There were no correlations between illness severity and maternal report of depression, anxiety, or PTSD symptoms (p’s > 0.12). There were no significant associations between trimester of infection and infant outcomes.

Congenital SARS-CoV-2 and infant brain ICN connectivity

Group-level independent components analysis produced 28 ICNs and 7 noise components (Supplementary Fig. S5). Overview of significant positive and negative connectivity between ICNs for each group is provided as connectograms in Supplementary Fig. S6. Visual inspection of network pairs within groups suggests there were more significant positive and negative network associations in the COVID+ group. Statistical evaluation of differences between COVID+ and COVID− groups revealed differences in between network connectivity for four ICN pairs, involving six networks (Fig. 4). Specifically, we observed positive connectivity in the COVID+ group between posterior and anterior midline regions (ICN 1 to ICN 33; p = 0.026; Cohen’s d = 0.661) and between insular and frontal regions (ICN 24 to ICN 35; p = 0.003; Cohen’s d = 0.879), as seen in Fig. 4, panels A and B. The COVID+ group also displayed reduced connectivity of both insula (ICN 21; p = 0.0002; Cohen’s d = 1.2) and lateral frontal (ICN 35; p = 0.007; Cohen’s d = 0.774) to posterior cingulate (ICN 25), as seen in Fig. 4, panels C and D. Figure 4 scatterplots show that for each significant pair, the non-exposed (COVID−) group average functional connectivity z-score is close to zero. Thus, significant between network effects reflect positive and negative associations present in the COVID+ group that are not apparent in the COVID− group. As with behavioral measures reported above, the observed neural comparisons do not remain statistically significant following multiple comparison adjustment. Supplemental Fig. S7 provides additional information about COVID+ and COVID− group differences within each age group.

Fig. 4. Examination of congenital SARS-CoV-2 exposure-related variation in infant brain functional connectivity.

Twenty-eight (non-noise) infant brain intrinsic connectivity networks (ICNs) were defined using ICA. Two-sample t-tests were performed to assess differences in between network connectivity between congenital exposure (COVID+) and non-exposed (COVID−) groups (p < 0.01). Cohen’s d effect size estimates for each observed significant pair shows that the effect sizes are all ≤ 1.2.

Age effects in significant ICN pairs

Follow-up analyses addressed whether connectivity between pairs of networks that were significant between groups shows age-related effects. This allowed us to better contextualize imaging results as representing a more or less mature developmental profile. Regression analyses revealed that the ICN21 to ICN 25 (insula to posterior cingulate regions) pair shows decreased connectivity with advancing infant age (p = 0.016; Fig. 5). The COVID+ group had significantly reduced connectivity in this pair (Fig. 4), suggesting a more advanced pattern of functional connectivity in the COVID+ group. A trend was observed for the ICN1 to ICN33 pair (posterior and anterior midline regions), which showed marginally stronger connectivity with age (p = 0.091, Fig. 5). Here, too, the COVID+ group reflected the more mature phenotype; stronger functional coupling was observed for this network pair in the COVID+ group (Fig. 4). The additional two pairs of interest did not demonstrate significant age-related change.

Fig. 5. Examination of age-related variation in infant brain functional connectivity.

Twenty-eight (non-noise) infant brain intrinsic connectivity networks (ICNs) were defined using ICA. Regression analyses were performed to assess age-related change in between network connectivity. Circles represent those who were not exposed to SARS-CoV-2, and triangles represent those who were exposed to SARS-CoV-2.

DISCUSSION

Overall, these data indicate few neurobehavioral differences exist in infants congenitally exposed to SARS-CoV-2. Where differences exist, effect sizes are modest and did not remain statistically significant following multiple comparison adjustment. These data respond to current unknowns regarding potential neurodevelopmental effects of SARS-CoV-2 infection during pregnancy. At present, there is little empirical outcome data available, and a strong corpus of findings that link other forms of maternal prenatal immune activation with both brain and behavioral differences in infants.⁴ Recent commentaries warn against the risk of negative bias in reporting results in the wake of the COVID-19 pandemic.³⁹ Thus, the approach taken here was to broadly evaluate fundamental domains of infant social, emotional, and cognitive function that develop rapidly across the first year of life. Our goal was to report all analyses and effect sizes to inform future studies aimed at replicating existing findings and/or identifying specific domains that should be followed up on or examined more closely.

Infant groups did not differ in the majority of areas in which reported and observed behavior were measured. No differences were noted in observed behavior measured during a validated visual attention task, nor in measures of infant emotional reactivity and regulation during two canonical stressor tasks (arm restraint and still face). No differences were observed in the achievement of developmental milestones assessed using the Ages and Stages Questionnaire (communication, motor (fine and gross), problem-solving, or personal-social) or in aspects of infant temperament (infant negative affect, soothability, regulatory capacity, or surgency) or socio-emotional development. Overall, in twenty-seven analyses of infant behavioral development, one area emerged as different: maternal report of infant duration of orienting at 12 months. Here, infants in the COVID− group were more likely to have longer duration of orienting. Duration of orienting characterizes temperamental differences in infants’ ability to sustain attention over extended periods of time during play or engagement with objects. Mary Rothbart and colleagues suggest that during infancy, the orienting network serves as a regulatory system, both in reducing negative affect and increasing surgency/positive affect.⁴⁰ Longitudinal studies have also linked variation in this domain with later measures of extraversion, such as approach and anticipation,⁴¹ effortful control,⁴² and risk for disorders such as autism or ADHD.⁴³ Although our findings suggest differences in orienting in infants with congenital COVID- 19 exposure, we did not observe differences in infant temperament or emotional reactivity and regulation. Thus, the present data suggest that infant orienting behavior does not likely modulate emotional expression and may not lead to sustained differences in behavior in these infants. This interpretation is consistent with our findings that COVID+ and COVID− infants did not differ in multiple other behavioral domains measured. Lack of significant differences between groups in behavioral development is aligned with and extends previous findings of no association between prenatal exposure to maternal SARS-CoV-2 infection and infant behavioral development.¹⁶ Altogether, our observed significant effect in orientation is viewed as preliminary evidence of a potential difference that warrants further examination. However, it is also important to note that these results should not be generalized beyond the specific domains and timepoints assessed in this study. For instance, these data cannot shed light onto whether there are significant effects of SARS-CoV2 infection on later developing domains (e.g., executive functions, language development). It is also possible that differences in some domains of infant attention or behavior may not emerge until later in development. Thus, continued longitudinal follow-up is important for a more complete picture.

Another area where COVID+ infants emerged as different from those without congenital SARS-CoV-2 exposure was in maternal report of infant sleep behavior. Specifically, we asked mothers about number of nighttime awakenings and sleep duration. We found that COVID+ infants had fewer nighttime awakenings at both ages, but no difference in nighttime sleep duration. Consistency of sleep patterns at 6 and 12 months fits with prior work reporting little change over this age range.⁴⁴ It is important to consider that observed differences in nighttime awakenings along with lack of difference in total nighttime sleep duration may be due in part to different sleep routines in the home. For example, household nighttime feeding practices can vary widely during the first year, and there may be cultural or co-sleeping differences in these households. The present study did not ascertain detailed information about the sleep environment, and we are thus not able to rule out the degree to which postnatal environmental factors may have contributed to the observed association between congenital COVID exposure and nighttime awakenings.

The current study additionally investigated whether SARS-CoV-2 illness severity related to variation in infant outcomes. In a subset of participants with data available on self-reported illness severity measured with the NCIPR,²⁴ there was no association between illness severity and observed behavioral outcomes. There was, however, a positive association between illness severity and maternal questionnaire report of infant negative affect at 12 months. It is unlikely this purely reflects negative reporting bias, as there were no significant correlations between illness severity and maternal report of depression, anxiety, or PTSD symptoms (p’s > 0.12). Infant negative affectivity is a key dimension of temperament and is frequently studied in early childhood given associations with later internalizing and externalizing problems in childhood. An important caveat to interpretation of this finding is that both illness severity and infant negativity were measured via maternal self-report and, therefore, may be susceptible to similar biases. Additionally, this cohort overall had temperament scores within a range considered typical, thus, the statistically significant variation in negative affect does not necessarily translate into any clinically meaningful difference. Moreover, the present sample overall may represent a lower end of the spectrum of illness severity in SARS-CoV-2 cases, and therefore may not reflect potential developmental effects among the minority of patients who suffer severe illness.

Our MRI analyses centered on functional associations between empirically derived infant neural subnetworks. This data driven approach enabled a whole-brain view of how pairs of networks may have different connectivity patterns in congenitally exposed and non-exposed groups. We observed group differences in four network pairs, involving six networks (Fig. 4). The COVID+ group showed increased positive connectivity between posterior and anterior midline regions (ICN 1 to ICN 33) and between insular and frontal regions (ICN 24 to ICN 35). The COVID+ group also showed comparatively reduced connectivity of both insula (ICN 21) and lateral frontal (ICN 35) to posterior cingulate (ICN 25). Networks labeled as ICN 21 and 35 both extend to insular regions, ICN 21 bilaterally, and ICN 35 is right lateralized and more superior. Insular regions are more connected to midline posterior cingulate (extending into the dorsal ACC) in the COVID− group, whereas insular regions are more connected to frontal regions in the COVID+ group. Interestingly, all of these subnetworks encompass regions that comprise canonical brain networks, specifically, the salience network (SN) and default mode network (DMN). The COVID− group showed more connectivity between regions that comprise the SN and DMN, a crossing of networks that will become more separable with age, seen in data presented here (Fig. 5; age effects) and in the broader literature. Thus, here, the exposed group appears to show a more mature pattern. Greater neural maturation in our COVID+ group aligns well with a recent study showing greater cortical volume and sulcal depth in congenitally SARS-CoV-2 exposed infants.²¹ Greater cortical volume and sulcal depth increase with age, and thus, findings from the independent work of Catherine Limperopoulos and colleagues also indicate that advanced maturational brain effects may accompany congenital SARS-CoV-2 exposure. The caveat here is that our MRI analyses are reported at p < 0.01, uncorrected, and should therefore be interpreted within this context.

In addition to our findings regarding SN-DMN connectivity, we found that the COVID+ group also showed a pattern that reflects more connectivity within the salience network. Specifically, the frontal regions noted as more connected to the insula in the COVID+ group correspond to the rostral PFC (area 10), which is a reliable constituent in salience network models.⁴⁵ Further, prior works⁴⁶ have suggested that the insula is important in patterning neural network development in very early human life, which supports continued follow-up of this and other cohorts to evaluate future brain and behavioral development. It is possible that variation in insular connectivity seen here may alter future brain functional organization.

An important theme that has arisen in the context of studies addressing potential neurodevelopmental differences in infants born to mothers who were infected with SARS-CoV-2 during pregnancy is the important and potentially interactive effect of maternal psychosocial stress. Initial evidence indicates it may be psychosocial stress in combination with congenital exposure, or alone, that may prove to be the significant risk factor for child developmental problems in the context of the pandemic.^15,17 Relevant to the current study is whether potential psychosocial differences in the COVID+ and COVID− groups may have contributed to our effects. Table 1 demonstrates that in this sample, the groups are well matched on prenatal psychological distress and on depression symptoms at 6 and 12 months. Thus, while not excluding all potential influences, differences between groups were not likely to have been driven by maternal psychological distress accompanying illness. Moreover, it is notable that participants overall report low levels of depression and anxiety symptoms, suggesting a relatively protected sample in contrast to elevated distress seen in prior studies. However, we note that psychosocial traits in the present sample may have offered protective effects for infant development that would obscure any potential between-groups differences that may have been visible in a less advantaged population. As such, these results may not generalize to higher-risk samples, and future work is needed to address this question.

In summary, our data indicate few neurobehavioral differences exist in infants congenitally exposed to SARS-CoV-2, and that where differences exist, effect sizes are small. It is noteworthy that our findings in infant sleep and functional connectivity denoted more mature patterns in the congenital exposure group; this finding is in line with a prior anatomical MRI study²¹ where a more mature pattern was also observed in the exposed group. It is possible that a more mature developmental profile may be seen in future continued work examining this population. Overall, replication and continued longitudinal research following these groups later into childhood will be necessary for drawing conclusions about behaviorally relevant and lasting neurodevelopmental effects of prenatal SARS-CoV-2 exposure.

Supplementary Material

Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41390-025-04409-y.

IMPACT:

• This study contributes comprehensive data on infant neurodevelopmental outcomes following prenatal SARS-CoV-2 exposure, evaluating a wide range of behavioral and neural measures to address gaps in previous research.

• Findings suggest that congenital exposure to SARS-CoV-2 does not result in significant neurodevelopmental impairments in infants, offering reassurance amidst concerns about potential long-term effects of maternal prenatal COVID-19 infection.

• Results indicate that any observed differences, such as fewer night awakenings and functional neural connectivity patterns, may reflect a more mature developmental profile in the exposed group.

• Continued longitudinal research is necessary to understand behaviorally relevant and lasting neurodevelopmental effects of prenatal SARS-CoV-2 exposure.

DATA AVAILABILITY

The investigators contributing to this manuscript are committed to the principles of open and reproducible science. Data from our fetal and infant neuroimaging studies are available on OpenNeuro and on the NIMH NDAR Data Archive. Any additional queries can be directed to the corresponding author.