Maternal psychiatric admissions and Child developmental delays: A nationwide cohort study

Abstract

Purpose

Children of mothers experiencing mental illness are known to exhibit elevated rates of developmental delays. However, there is little evidence-based research on the impact of perinatal psychiatric admissions. Here we aim to quantify the association between maternal mental illness that leads to admission, and the likelihood of child developmental delay. In addition, we aim to understand the association between the timing of admission, the maternal diagnosis and the developmental domain in which delays manifest.

Methods

Retrospective cohort study of data from well-child visits of 634,918 term-born children assessed between 2016 and 2022 at the ages of 1 to 36 months. Likelihood of failure in milestone attainment was modelled based on maternal psychiatric information available through the national registry, and adjusted for demographic and birth data from the Israeli national program of Maternal Child Health Clinics. Secondary analyses examined stratification by maternal diagnosis, timing of admission and developmental domain, as well as comparison to the association between the likelihood of child developmental delay and maternal high scores on the Edinburgh Postnatal Depression Scale.

Results

Maternal mental illness that leads to psychiatric admission was associated with an increased odds ratio for child developmental delay (OR ~ 1.7). Higher odds ratio was associated with psychotic disorders diagnoses, multiple admissions and admissions between birth and child assessment. Nonetheless, increased odds were suggested also when first admission occurred after child assessment. In addition, increased odds were more evident when children were assessed at an older age.

Conclusions

Children of mothers experiencing mental illness exhibit higher rates of developmental delays, especially when mothers are admitted. Identifying and supporting high-risk dyads during the peripartum period is therefore critical for both mother and child.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00737-025-01613-x.

Article highlights

• Maternal mental illness is known to be negatively associated with child development.

• Here we examine the effect of maternal psychiatric admissions, showing it increases the odds for child developmental delays by 70%.

• Higher likelihood of delay is associated with admissions due to psychosis, multiple admissions, and admissions in the prepartum period.

• Likelihood of delay is elevated even when admission occurs after developmental assessment.

Supplementary Information

The online version contains supplementary material available at 10.1007/s00737-025-01613-x.

Introduction

In January 2023 the TRIUMPH for New Moms Act was approved by the U.S. congress and signed into law (IHPL 2023), funding the development of a national strategy to improve maternal mental health outcomes and expand mental health resources for new mothers.1 This exemplifies the growing recognition of the importance of maternal mental health. Indeed, maternal mental health outcomes, especially with regard to child development, have been studied for at least three decades (Beck 1998; Power et al. 2021).

Maternal psychiatric disorders can have significant and lasting effects on children’s development. Research indicates that maternal mental illness, particularly when severe enough to require hospitalization, is associated with increased rates of developmental delays, emotional and behavioral difficulties, and both somatic and mental health disorders in children (Bell et al. 2023; Milgrom et al. 2016; Watkeys et al. 2023). As early childhood is recognized as a critical period for brain development, characterized by both high adaptability and vulnerability to external influences (Benitez 2024; Chelini et al. 2022), disruptions in maternal care due to psychiatric admissions during this time may have particularly profound impacts.

Many of the studies on maternal psychiatric admissions have focused on the peripartum period, and in particular on peripartum depression (Cummings & Davies 1994; Bernard-Bonnin et al. 2004; Evans et al. 2012; Kaplan et al. 2014; Stein et al. 2014, 2018; Liu et al. 2017; Netsi et al. 2018; Murray et al. 2018; Rogers et al. 2020; Lubotzky-Gete et al. 2021), identifying a significant association between peripartum symptoms of depression and poor developmental outcomes of the child that extend from infancy through childhood and into adolescence (Rogers et al. 2020). Importantly, such studies tend to rely on self-reported (or screening) assessments of mental health, rather than on clinical diagnoses (Rogers et al. 2020). For example, a recent study of over 100,000 Israeli children found that the mother’s score on the Edinburgh Postnatal Depression Scale (EPDS) questionnaire is associated with an increased likelihood for failing to attain certain well-child visit milestones during the child’s first two years of life (Lubotzky-Gete et al. 2021).

Studies that analyzed the effect of maternal schizophrenia on child outcomes (Bennedsen 1998; Niemi et al. 2003; Lin et al. 2010; Judd et al. 2014; Hameed & Lewis 2016; Vigod et al. 2020) mostly examined obstetric complications, such as low birth weight, and health outcomes at later ages. Notably, several small-scale longitudinal studies (Niemi et al. 2003; Ranning et al. 2015) have shown that maternal schizophrenia is associated with higher rates of motor, cognitive, and social deficits in children.

Other studies have assessed the impact of maternal mental illness across multiple diagnoses, or without specific evaluation of maternal mental distress (Bell et al. 2023; Bennett et al. 2016; Burger et al. 2022; Gold & Marcus 2008; Gül et al. 2020; Hope et al. 2021; Kahn et al. 2004; Kingston & Tough 2014; McDonald et al. 2016). For example, a study of small cohorts in South Africa evaluated children of mentally ill mothers using the Bayley Scales of Infant and Toddler Development. Detrimental effects were identified, including language and cognitive delays in children of mothers with paranoid ideation, and gross motor delays in children of mothers exhibiting symptoms of hostility and anxiety. (Burger et al. 2022). On the other hand, in a small Turkish cohort of subclinical mothers (Gül et al. 2020), differential effects were identified, such as language-cognitive delays in children of mothers with paranoid ideation, and gross motor delays in children of mothers with symptoms of hostility and anxiety.

Most pertinent to our work is a study of two large cohorts, from Canada and Australia, examining the impact of maternal psychiatric admissions on child developmental vulnerability (Bell et al. 2023). Developmental vulnerability (or school readiness) was assessed by kindergarten and pre-school teachers at school entry using the Early Development Instrument (Janus & Offord 2007). This questionnaire compares child development in five domains (Physical Health & Well-being, Social Competence, Emotional Maturity, Language and Cognitive Development, Communication Skills & General Knowledge) to standardized norms, and considers children scoring in the lower decile as developmentally vulnerable. Maternal psychiatric admission was suggested to increase odds by a factor of 1.5–2.1, depending on the domain and cohort. Psychiatric admissions were identified by the diagnosis at discharge, by either being in the 290–319 range for ICD-9, or F00-F99 and O99.3 for ICD-10. Interestingly, the child’s age at maternal admission had little bearing on these odds.

In this work we describe the association between maternal psychiatric admissions and developmental milestone attainment in a large nationwide Israeli cohort. Israel is an appealing choice for such research for two main reasons: (1) The Israeli Ministry of Health maintains a comprehensive National Psychiatric Hospitalization Registry (NPHR) since 1950 (Lichtenberg et al. 1999), systematically documenting all psychiatric hospital admissions across the country. (2) It has a long-running national well-child visits program, which is free of charge and has a very high compliance rate, especially during early ages. Well-child visits are regularly scheduled clinic appointments where a child’s development and growth are monitored, and guidance and referrals are given as needed (alongside immunizations and other preventive care). During these visits, child development is evaluated using age-appropriate milestones whose attainment is recorded in structured electronic health records (EHRs). A description of these milestones is available online (Sadaka 2022), alongside the developmental domains they relate to, the ages at which they are assessed, and an explanation of how to evaluate them.

Our research hypotheses regarding the association between maternal psychiatric admissions and child developmental delay were as follows: (1) that the strongest impact would be on the development of children whose mothers were admitted between birth and well-child visit, since maternal absence may lead to under stimulation and subsequent developmental delay; (2) that children of mothers admitted with psychotic disorders would be most adversely affected since, as also suggested by previous work, such disorders adversely affect inter-personal and organizational capabilities, hampering mother–child interaction; (3) that the strongest impact would be in the social and linguistic development of the child, due to disadvantageous parental interaction and maternal absence, and that this effect will become more apparent as the child grows older.

Accordingly, our analysis examined the following factors: (1) the time of admission, relative to the birth and the child’s assessment; (2) the diagnoses of the admitted mothers; (3) the domain of the child’s assessment; and (4) the age of the child’s assessment. In addition, since mothers in Israel are routinely screened for symptoms of postpartum depression, we compare the strength of the association between these symptoms and developmental delays, to that suggested for maternal admissions.

Methods

This study follows the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guidelines for cohort studies.

Study design

This is a retrospective cohort study where, in the main analysis, the intervention is maternal mental illness that leads to psychiatric admission, and the outcome is child developmental delay between ages 12 and 24 months. Notably, admissions occurring after the outcome were included, as they were considered proxies for underlying mental distress present prior to the outcome but not yet resulting in hospitalization. Indeed, the group of mothers who were admitted only after a developmental delay was assessed constitutes an important comparison group vis-à-vis those admitted beforehand.

Several secondary analyses were also performed, to provide a more exhaustive and refined view of the association between maternal admission and child developmental delay:

1. Varying the age brackets at which development is assessed.

2. Restricting the outcome to delays in a specific developmental domain.

3. Considering different features of admissions – when they occurred, how many times, and what were the diagnoses.

4. Including the result of postpartum depression screening (using the Edinburgh Postnatal Depression Scale) as an explanatory variable, and comparing its contribution to the likelihood of child developmental delay with that of maternal admission.

Inclusion and exclusion criteria

This work considers mothers who were:

1. Listed in the national psychiatric admissions registry as having been admitted between 1996 and 2023 (inclusive) for any reason.

2. Were 18–41 years old at the time of admission,

3. Gave birth on term between 2016 and 2022 and had their child assessed at a participating clinic during this time period.

Study cohorts

The study included full-term children who were born and assessed at a participating maternal and child health clinic (MCHC) between 2016 and 2022 (inclusive). Children for whom gestational age is missing or invalid were excluded. Children who were not assessed when they were in the age range of interest (for the relevant milestones) were also excluded.

The children were divided into two groups: those whose mothers were listed in the NPHR (admitted cohort) and those who were not (unadmitted cohort). The admitted cohort was further divided into three subgroups based on admission timing: prebirth admission (mothers admitted before birth but not between birth and MCHC assessment), birth-assessment admission (mothers admitted between birth and MCHC assessment, possibly at other times), and post-assessment admission (mothers admitted only after the child's MCHC assessment). This partition is sketched in Fig. 1.

Fig. 1.

Schematic depiction of the three sub-cohorts of children whose mothers were admitted. Arrows point to the time windows where mothers in the cohort were admitted. In addition, mothers in the prebirth cohort may also have been admitted after assessment, and mothers in the birth-assessment cohort may also have been admitted before birth or after assessment. Children are typically assessed twice between ages 12 and 24 months – once at age 12 months, and once at age 18 months.

Data collection: Child developmental surveillance

Developmental surveillance in Israel is routinely conducted from birth until 6 years of age, according to national guidelines. Certified public health nurses perform these assessments in approximately 1,000 MCHCs. The standardized protocol includes the evaluation of 59 developmental milestones across four domains: personal-social, language, fine motor, and gross motor skills. Parents are recommended to attend the MCHCs following hospital discharge at prescribed times thereafter. Specifically, well-child visits are recommended at ages 1, 3, 6, 9,12,18, 24, 36, 48 and 60 months. Attendance rates are high, particularly during the first two years of life, when these visits usually coincide with the national vaccination schedule.

During each visit, attainment of developmental milestones appropriate to the child’s age ("age step", defined as the intervals between recommended visit ages) is assessed and recorded based on the expected norms reflected in the Tipat Halav Israeli Screening (THIS) scale (Sudry et al. 2022). Children who fail to attain milestones beyond the age threshold corresponding to the 95th percentile of the national normative range are referred to a pediatrician for further evaluation to rule out developmental delay.

Approximately 65% of the children born in Israel have their MCHC-collected data documented within a unified electronic health record (EHR) system. This includes sociodemographic information and birth-related details, as well as visit-specific data on the attainment of developmental milestones. Additionally, most mothers are screened for postpartum depression (PPD) using the EPDS questionnaire, 4–8 weeks postpartum.

The demographic data of the mothers was based on self-reported information, with two exceptions: socioeconomic status (SES) and being Haredi (a strict orthodox branch of Judaisim). These were deduced from the mother’s home address: The Israeli Central Bureau of Statistics partitions the country into geostatistical areas. Each such area is assigned to one of ten clusters based on aggregated socioeconomic characteristics of its residents. In addition, the rate of Haredi population is in each area estimated. Herein, SES was labeled as “low” for clusters 1–3, “medium” for clusters 4–7, and “high” for clusters 8–10.

Parity of birth was not explicitly recorded in the dataset, and was estimated herein from the number of older siblings for whom records exist in the dataset.

Data collection: Maternal psychiatric admissions

The Israeli Ministry of Health has maintained a comprehensive National Psychiatric Hospitalization Registry (NPHR) since 1950 (Lichtenberg et al. 1999), systematically documenting all psychiatric hospital admissions across the country. This registry includes demographic and clinical data for each admission (up to six ICD-10 diagnostic codes assigned at discharge, or, if unavailable, at admission), as well as admission and discharge dates.

The present study focuses on specific diagnostic groups within this registry – schizophrenia (ICD-10 code F20), other psychotic disorders (F21–F28), bipolar disorder (F30, F31), depressive disorders (F32, F33), eating disorders (F50), and personality disorders (F60). An admission is associated with a diagnostic group if one of the recorded diagnostic codes belongs to this group (ignoring subcategories, if listed).

Outcome definition

In line with national guidelines for developmental surveillance, milestone failure is considered herein as “severe”, if it is attained by 95% or more of children at the same age, and it is used as an indicator for developmental delay. This is assessed for any milestone, as well as for milestones from a specific domain – language, social or motor (encompassing both fine motor and gross motor milestones). These outcomes are denoted any-failure, language-failure, social-failure and motor-failure. The main outcome of interest is any-failure, indicating some developmental delay.

Additionally, we used the THIS scale to calculate a Developmental Delay Score (Bilu et al. 2023) (DDS) for each child, based on milestone attainment within a specified age range. DDS enables sensitivity analysis by providing a granular assessment of developmental delay. Instead of categorizing milestone failures as either “severe” or “non-severe”, each failure is assigned a score reflecting its frequency among the same age population. The greater the proportion of children who attain the milestone at this age – the higher the corresponding developmental delay score. To compute the DDS for a specific visit (or set of visits), a score is computed for each failed milestone, and then averaged over all examined milestones.

In the main analysis, an outcome of developmental delay was defined by severe milestone attainment failure between the ages of 12 and 24 months. In secondary analyses we considered other age ranges as well: 1–6 months, 6–12 months, 12–18 months, 18–24 months and 24–36 months.

Developmental delay likelihood modelling and statistical analysis

The likelihood for severe milestone failure was modelled by a multivariable logistic regression. Covariates in the model included mother demographics – Age, employment status, marital status, being Jewish, being Israeli born, being Haredi and SES cluster; and neonatal information – Sex, birth weight, birth type, gestational age, indication for multiple gestation, and parity (see supplemental for how they were encoded). To reduce intra-cluster correlations of siblings data from the same mothers, Generalized Estimate Equations (GEE) (Hardin 2003) was used.

In addition, in the primary analysis, a binary variable indicated whether the mother was ever admitted (between the ages of 18 and 41).

In three secondary analyses we aimed to look at admissions at a finer granularity. This was done using three different variations on the primary model.

In the admission time modelling, the binary indicator for admission was replaced by three binary variables, indicating whether the mother belongs to the prebirth, birth-assessment or post-assessment cohort.

In the admission diagnosis modelling, we instead replaced the indicator for admission with binary variables indicating the diagnostic group associated with admissions. Such an indicator takes a value of 1 if any of a mother’s admissions include a diagnosis from the indicated diagnostic group. Hence, a mother may be associated with more than one diagnostic group.

Finally, in the admission count modelling, instead of using a binary variable for admission we used a categorical variable indicating for each child the total number of maternal admissions in the dataset – zero, one, two, or more than two.

An additional secondary analysis aimed to compare the impact of postpartum depressive symptoms to that of maternal psychiatric admissions. To do so, we added the binary result of EPDS screening as a covariate. A positive EPDS was defined as a total score above 10, or a non-zero answer to the question about self-harm (Lubotzky-Gete et al. 2021). This was done on the subset of children whose mothers were screened.

The significance of differences between cohort covariates was estimated using Fisher exact test for binary variables, chi-squared test for categorical variables, and Welch t-test for continuous ones.

Results

Population characteristics

Out of 634,918 children who met the inclusion criterion for the main analysis, 3,339 (0.5%) were born to mothers with a recorded psychiatric admission (Table 1). Among them, 2,025 (61%) were in the prebirth cohort, 430 (13%) in the birth-assessment cohort and 884 (26%) in the post-assessment cohort.

Table 1.

Demographic and birth-related characteristics of children and mothers in the study cohorts. Numbers in parentheses are percentages from the relevant cohort, except last five lines. which show interquartile range (IQR)

	Cohort
	Not admitted	Admitted	Prebirth	Birth-Assessment	Post-Assessment
N	631,579	3,339	2,025	430	884
Sex: Male	322,958 (51.14%)	1,694 (50.73%)	1,015 (50.12%)	209 (48.60%)	470 (53.17%)
Sex: Female	308,621 (48.86%)	1,645 (49.27%)	1,010 (49.88%)	221 (51.40%)	414 (46.83%)
SES: Low	260,432 (41.24%)	1,158 (34.68%)	608 (30.02%)	171 (39.77%)	379 (42.87%)
SES: Medium	241,366 (38.22%)	1,542 (46.18%)	977 (48.25%)	194 (45.12%)	371 (41.97%)
SES: High	66,501 (10.53%)	308 (9.22%)	231 (11.41%)	27 (6.28%)	50 (5.66%)
SES: Missing	63,280 (10.02%)	331 (9.91%)	209 (10.32%)	38 (8.84%)	84 (9.50%)
Ethnicity: Jewish	389,589 (61.68%)	2,338 (70.02%)	1,483 (73.23%)	296 (68.84%)	559 (63.24%)
Ethnicity: Muslim Arab	138,118 (21.87%)	495 (14.82%)	256 (12.64%)	72 (16.74%)	167 (18.89%)
Ethnicity: Missing	57,750 (9.14%)	339 (10.15%)	197 (9.73%)	46 (10.70%)	96 (10.86%)
Ethnicity: Muslim Bedouin	12,683 (2.01%)	13 (0.39%)	7 (0.35%)	1 (0.23%)	5 (0.57%)
Ethnicity: Other	12,569 (1.99%)	97 (2.91%)	52 (2.57%)	9 (2.09%)	36 (4.07%)
Ethnicity: Druse	12,176 (1.93%)	36 (1.08%)	16 (0.79%)	5 (1.16%)	15 (1.70%)
Ethnicity: Christian Arab	8,694 (1.38%)	21 (0.63%)	14 (0.69%)	1 (0.23%)	6 (0.68%)
Marital status: Married	560,669 (88.77%)	2,581 (77.30%)	1,523 (75.21%)	334 (77.67%)	724 (81.90%)
Marital status: Missing	35,853 (5.68%)	200 (5.99%)	120 (5.93%)	24 (5.58%)	56 (6.33%)
Marital status: Single	21,173 (3.35%)	381 (11.41%)	253 (12.49%)	51 (11.86%)	77 (8.71%)
Marital status: Other	7,499 (1.19%)	72 (2.16%)	52 (2.57%)	9 (2.09%)	11 (1.24%)
Marital status: Divorced	6,385 (1.01%)	105 (3.14%)	77 (3.80%)	12 (2.79%)	16 (1.81%)
Employment status: Working	308,920 (48.91%)	1,292 (38.69%)	823 (40.64%)	142 (33.02%)	327 (36.99%)
Employment status: Missing	165,829 (26.26%)	883 (26.45%)	542 (26.77%)	123 (28.60%)	218 (24.66%)
Employment status: Not Working	128,308 (20.32%)	1,059 (31.72%)	611 (30.17%)	146 (33.95%)	302 (34.16%)
Employment status: Student	28,522 (4.52%)	105 (3.14%)	49 (2.42%)	19 (4.42%)	37 (4.19%)
Education level: Academic	200,992 (31.82%)	720 (21.56%)	460 (22.72%)	84 (19.53%)	176 (19.91%)
Education level: Missing	180,049 (28.51%)	1,007 (30.16%)	608 (30.02%)	136 (31.63%)	263 (29.75%)
Education level: High School	167,923 (26.59%)	1,188 (35.58%)	705 (34.81%)	146 (33.95%)	337 (38.12%)
Education level: Tertiary Education	69,747 (11.04%)	317 (9.49%)	175 (8.64%)	55 (12.79%)	87 (9.84%)
Education level: Elementary	12,868 (2.04%)	107 (3.20%)	77 (3.80%)	9 (2.09%)	21 (2.38%)
Birth type: Spontaneous	490,904 (77.73%)	2,407 (72.09%)	1,421 (70.17%)	315 (73.26%)	671 (75.90%)
Birth type: Caesarean section	103,021 (16.31%)	718 (21.50%)	469 (23.16%)	93 (21.63%)	156 (17.65%)
Birth type: Instrumental	35,573 (5.63%)	204 (6.11%)	130 (6.42%)	20 (4.65%)	54 (6.11%)
Birth type: Missing	2,081 (0.33%)	10 (0.30%)	5 (0.25%)	2 (0.47%)	3 (0.34%)
Haredi	121,374 (19.22%)	539 (16.14%)	275 (13.58%)	84 (19.53%)	180 (20.36%)
Israeli-born	557,078 (88.20%)	2,732 (81.82%)	1,627 (80.35%)	346 (80.47%)	759 (85.86%)
Multiple gestation	10,777 (1.71%)	28 (0.84%)	20 (0.99%)	4 (0.93%)	4 (0.45%)
EPDS screened	429,187 (67.95%)	2,222 (66.55%)	1,346 (66.47%)	289 (67.21%)	587 (66.40%)
Days between birth and EPDS median (IQR):	37 (32–62)	38 (32–63)	38 (32–63)	39 (32–64)	38 (32–63)
Parity median (IQR):	1 (1–2)	1 (1–2)	1 (1–2)	1 (1–2)	1 (1–1)
Gestational age median (IQR):	39 (38–40)	39 (38–40)	39 (38–40)	39 (38–40)	39 (38–40)
Birth weight median (IQR):	3 (3–4)	3 (3–4)	3 (3–4)	3 (3–4)	3 (3–4)
Mother age median (IQR):	30 (26–34)	31 (27–35)	32 (29–36)	29 (26–34)	28 (25–32)

Children in the admitted cohort were more often Jewish compared to the unadmitted cohort (70% vs 62%, p < 0.001), and their families were less often from low SES clusters (35% vs 41%, p < 0.001). Their mothers were less often married (77% vs 89%, p < 0.001), employed (39% vs 49%, p < 0.001), Israeli-born (82% vs 88%, p < 0.001) or had an academic degree (22% vs 32%, p < 0.001). The three sub-cohorts based on the timing of admission, were generally similar in covariate distribution, with the exception of the post-assessment cohort having a higher rate of children to low SES mothers (43% vs 35%, p < 0.001), and younger mothers (median age at giving birth 28 vs 31 years, p < 0.001).

Among mothers of children in the admitted cohort, the frequency of each diagnostic group was roughly 20% (range 18%−23%), with the exception of the less common diagnosis of eating disorder (10%; Table 2). In the birth-assessment cohort diagnoses of psychotic disorders (other than schizophrenia) were more frequent compared to the entire admitted cohort (28% vs 18%, p < 0.001), while eating disorders (1.4% vs 10%, p < 0.001) and personality disorders (12% vs 21%, p < 0.001) were less so. In the post-assessment cohort diagnoses of depression were relatively more frequent (31% vs 23%, p < 0.001), as were eating disorders in the prebirth cohort (14% vs 10%; p < 0.001).

Table 2.

Distribution of diagnostic groups associated with admissions in the study cohorts. Numbers in parentheses are percentages from the relevant cohort. Note that discharge notes may contain more than one diagnoses

	Admitted	Prebirth	Birth-Assessment	Post-Assessment
N	3,339	2,025	430	884
Admission: Schizophrenia	597 (17.88%)	352 (17.38%)	78 (18.14%)	115 (13.01%)
Admission: Other Psychosis	610 (18.27%)	309 (15.26%)	121 (28.14%)	130 (14.71%)
Admission: Bipolar	644 (19.29%)	323 (15.95%)	92 (21.40%)	180 (20.36%)
Admission: Depression	761 (22.79%)	355 (17.53%)	93 (21.63%)	275 (31.11%)
Admission: Eating dis	333 (9.97%)	287 (14.17%)	6 (1.40%)	34 (3.85%)
Admission: Personality dis	696 (20.84%)	419 (20.69%)	50 (11.63%)	183 (20.70%)
Admission: other	714 (21.38%)	0 (0.00%)	84 (19.53%)	0 (0.00%)

Prevalence of developmental delay

Among 631,579 children in the unadmitted cohort who were assessed at ages 12–24 months, 8% severely failed to attain a developmental milestone in any domain. Stratified by domains, 2%, 3% and 4% of the children failed to attain a social, language or a motor milestone, respectively. By contrast, 14% of children in the admitted cohort severely failed attaining a milestone at this age range (p < 0.001), with an even higher rate (17%) in the birth-assessment sub-cohort. Relative failure rate was high in the social and language domains, being more than twofold higher in the admitted cohort than in the unadmitted (p < 0.001), and threefold higher in the birth-assessment cohort. However, failure rates in the birth-assessment cohort were not found to be significantly different from those in the prebirth cohort or the post-assessment cohort.

Similarly, the mean DDS score in the unadmitted cohort was significantly lower than in the admitted cohort (0.15 vs 0.23, p < 0.001), and this difference was most prominent in the social domain. Interestingly, this score was similar among the three sub-cohorts defined by admission time (Table 3).

Table 3.

Number of children assessed (lines with bold title) by developmental domain, and rates of failure among them alongside mean Developmental Delay Score (DDS), in each of the cohorts. Numbers in parentheses are standard errors.

	Not Admitted	Admitted	Prebirth	Birth-Assessment	Post-Assessment
Assessed for Any	631,465	3,339	2,025	430	884
Failure rate in Any (SE)	0.08 (0.000)	0.14 (0.006)	0.14 (0.008)	0.17 (0.018)	0.14 (0.012)
Mean DDS in Any (SE)	0.15 (0.000)	0.23 (0.006)	0.23 (0.008)	0.24 (0.016)	0.22 (0.011)
Assessed for Social	631,081	3,336	2,024	428	884
Failure rate in Social (SE)	0.02 (0.000)	0.05 (0.004)	0.05 (0.005)	0.06 (0.011)	0.05 (0.007)
Mean DDS in Social (SE)	0.08 (0.000)	0.15 (0.007)	0.14 (0.009)	0.15 (0.019)	0.15 (0.013)
Assessed for Language	631,346	3,339	2,025	430	884
Failure rate in Language (SE)	0.03 (0.000)	0.07 (0.005)	0.08 (0.006)	0.09 (0.014)	0.06 (0.008)
Mean DDS in Language (SE)	0.18 (0.000)	0.27 (0.007)	0.28 (0.009)	0.28 (0.019)	0.26 (0.013)
Assessed for Motor	630,515	3,331	2,019	429	883
Failure rate in Motor (SE)	0.04 (0.000)	0.07 (0.004)	0.07 (0.006)	0.09 (0.014)	0.06 (0.008)
Mean DDS in Motor (SE)	0.12 (0.000)	0.19 (0.008)	0.20 (0.011)	0.19 (0.023)	0.18 (0.015)

Taken together, these results suggest that maternal admission is associated with a higher likelihood for child developmental delay. In the next subsection we analyze this association when adjusting for the effect of additional covariates.

Association between psychiatric admissions and developmental delay

To quantify how the association between maternal psychiatric admissions and child developmental delay relates to other risk factors for developmental delay, we modeled the likelihood of severe milestone failure at ages 12–24 months using a multivariable logistic regression model, with the covariates listed in Table 1. Figure 2 (top) depicts the odds ratios of the any-failure outcome for each of the covariates (relative to the mode value, where relevant).

Fig. 2.

Odds ratios for milestone failure in any domain at age 12–24 months. Ratios for categorical covariates are relative to their mode value. Top: basic modelling results. Bottom: admission time modelling results

Increased likelihood of any-failure was observed for birth weight below 2.5 kg (OR = 1.47, 95% CI 1.40–1.54), early gestational age (week 37–38: OR = 1.23, 95% CI 1.20–1.28; week 39: OR = 1.07, 95% CI 1.04–1.09), birth by cesarean section (OR = 1.18, 95% CI 1.15–1.21), as well as for males (OR = 1.35, 95% CI 1.32–1.37) and for Jewish children (OR = 1.41, 95% CI 1.37–1.44). Reduced likelihood was observed for children of mothers who are Israeli born (OR = 0.80, 95% CI 0.78–0.82), reside in high SES areas (OR = 0.72, 95% CI 0.70–0.75), are employed (OR = 0.96, 95% CI 0.94–0.98), married (OR = 0.87, 95% CI 0.84–0.89) and have academic education (OR = 0.76, 95% CI 0.74–0.78). Importantly, the covariate associated with the highest likelihood of any-failure outcome was admission of the mother at some point in her life (OR 1.71, 95% CI 1.54–1.89). This association was even stronger for the language-failure (OR = 1.90, 95% CI 1.66–2.18) and social-failure (OR = 1.82, 95% CI 1.55–2.14) outcomes (Supplementary Figures S1-S3).

These estimates suggest that, although multiple factors contribute to the likelihood of child developmental delay, maternal admission remains a prominent predictor even after adjusting for these factors.

Impact of different aspects of admission

The results of the admission time modelling are depicted in Fig. 2 (bottom). Among the three admission time indicators, the highest likelihood was associated with the birth-assessment admission time (OR = 2.07, 95% CI 1.60–2.67), yet this likelihood was not significantly different from those associated with the other two admission times (OR = 1.64 and 1.70 for prebirth and post-assessment admissions, respectively). Similar results were observed when considering domain-specific failures (Supplementary Figures S4-S6).

The admission count modelling suggested that multiple maternal admissions were associated with higher likelihood for child developmental delay. Specifically, children of mothers who were admitted once were found to be at lower odds (OR = 1.47, 95% CI 1.28–1.70) compared to those of mothers with two admissions (OR = 2.00, 95% CI 1.60–2.51), three admissions (OR = 2.02, 95% CI 1.47–2.70) or more than three admissions (OR = 2.11, 95% CI 1.65–2.70).

Finally, Fig. 3 depicts the odds ratios of the any-failure outcome in the admission diagnosis modelling, suggesting that the likelihood is highest for mothers with psychotic disorders (OR = 1.48, 95% CI 1.11–1.97), and in particular schizophrenia (OR = 1.84, 95% CI 1.44–2.36). Depression (OR = 1.34, 95% CI 1.06–1.69) and bipolar disorder (OR = 1.29, 95% CI 1.00–1.67, p = 0.05) were associated with a lower likelihood, which is nonetheless statistically significant. Diagnoses of personality disorders and eating disorders were not found to be associated with a significantly increased likelihood.

Fig. 3.

Odds ratio for severe milestone failure at ages 12–24 months according to the mothers' discharge diagnosis.

Interestingly, a somewhat different association was observed when considering domain-specific outcomes (Supplementary Figures S7-S9). The association between bipolar disorder and attainment of social and language milestones did not appear to be significant, nor was the association between depression and attainment of motor milestones. Schizophrenia, while associated with the highest likelihood of failure in language and motor milestones, did not appear to pose significant likelihood of failure in social milestones.

In summary, the association between maternal admission and child developmental delay is suggested to be nuanced, influenced by the timing and frequency of the admissions, as well as the nature of the underlying mental illness. Moreover, the effect may vary across different developmental domains.

Association at different age steps

So far, our analysis focused on milestone failures at ages 12–24 months. Next, we examined the association between maternal admission and milestone failure at different age brackets. Likelihood was modelled as before (a single binary variable for admission) but five different age ranges were considered as the observation periods for this outcome. Figure 4 depicts the resultant odds ratios, suggesting that likelihood significantly increases as the child grows older (p < 0.001; Mann–Kendall test (Yue & Wang 2004)), from 1.20 (95% CI 1.10–1.31) in the first six months of life, to 1.81 (95% CI 1.56–2.11) at ages 2–3 years. This is further supported by an interaction analysis described in the supplemental material.

Fig. 4.

Odds ratios of any-failure outcome for children at different age brackets whose mothers were admitted, compared to unadmitted. The y-axis denotes the age bracket in months at which milestone attainment was assessed.

When considering domain-specific failures results are largely similar (Supplementary Figures S10-S12), though monotonicity is not completely maintained in the language and motor domains.

Comparison to the risk associated with a positive EPDS

Finally, we compared the known association between maternal high EPDS scores and developmental delay (Lubotzky-Gete et al. 2021), to the association between maternal psychiatric admission and developmental delay. For this, we analyzed the 431,409 children whose mothers were screened with EPDS, adding a binary variable to the regression model indicating whether the earliest postpartum EPDS was positive.

Figure 5 depicts the odds ratios for the any-failure outcome at ages 12–24 months suggested by this model. Children whose mothers were EPDS-positive had higher odds (OR = 1.20, 95% CI 1.13–1.27), but these odds were significantly lower than those observed for children whose mothers were admitted (OR = 1.71, 95% CI 1.51–1.93). Similar results were obtained when considering domain-specific failures (Supplementary Figures S13-S15).

Fig. 5.

Odds ratios for any-failure outcome at age 12–24 months in a cohort of children whose mothers were screened with EPDS

Discussion

The results of this large longitudinal cohort study align with previous work suggesting a significant association between maternal mental illness and children’s developmental delay (Bell et al. 2023; Milgrom et al. 2016; Watkeys et al. 2023). Indeed, we find that severe failure in milestone attainment at ages 12–24 months is about twice as common among children of mothers who were admitted to a psychiatric hospital, compared to those whose mothers were not admitted.

Importantly, there are several maternal sociodemographic confounding factors, which are associated with both maternal admissions and developmental delay. For example - lower levels of education, being unmarried or unemployed, and delivery by cesarean section. However, even when controlling for these factors, we find that children of admitted mothers are at a significantly higher likelihood for a milestone failure than their counterparts. Moreover, this detrimental relation is more clearly evident in social and language milestones than in motor milestones.

Contrary to our initial hypothesis, and consistent with Bell et al. (2023), we do not find that the relative time of admission has a significant impact on the likelihood of developmental delay. While children whose mothers were admitted between birth and developmental assessment were found to have the highest odds ratio (OR = 2.07), it was not significantly higher than for children whose mothers’ first admission occurred after the assessment (OR = 1.70), perhaps because admission is a rare event, and a larger sample size would be required to achieve statistical significance. This suggests that the mother's mental illness, often beginning prior to admission, may have a greater impact on the child's development than the admission itself and the mother’s absence. In other words, timely intervention is crucial for healthy child development, even when admission does not coincide with the examined developmental period, and probably even when mental illness is not severe enough to warrant admission.

Additionally, our analysis suggests a higher developmental delay likelihood for children of mothers with multiple admissions. This may be because multiple admissions are associated with a more severe, or chronic, illness (Swett 1995), and because such mothers are more often absent from their child’s life, leading to under stimulation and hampering, among other things, their ability to form a healthy relationship. Future research might examine lifetime admissions (beyond age 41) to investigate this more fully.

As expected, the highest likelihood was associated with mothers who experienced psychotic disorders (OR = 1.48), and in particular schizophrenia (OR = 1.84). Similar to other works (Gül et al. 2020), we find that there are different association strengths for different developmental domains, with schizophrenia, for example, having a much stronger association with failure in the language domain (OR = 2.19), than in the other domains (social OR = 1.32, motor OR = 1.58). Parental schizophrenia is associated with a high likelihood for child developmental delay compared to other parental psychiatric disorders due to genetic, biological, and environmental factors. Children of parents with schizophrenia inherit an elevated genetic risk for neurodevelopmental disorders (Schmitt et al. 2023), while prenatal stress, inflammation, and perinatal complications can further impact brain development (Ge et al. 2025). Postnatally, suboptimal caregiving (Davidsen et al. 2015) and environmental stressors (Wan et al. 2008) can further increase vulnerability to cognitive, emotional, and behavioral difficulties. Therefore, these children require comprehensive early intervention and extensive support to mitigate long-term risks.

We also find that the association between psychiatric admission and milestone failures becomes stronger as the assessed child is older. There can be several contributing factors to this observation. First, when maternal mental illness leads to suboptimal parenting, these interactions may accumulate over time, exacerbating the developmental gap as the child grows older (Kerna et al. 2024). Second, some developmental delays take longer to manifest, and may become more readily apparent as the assessment becomes more refined and elaborate (Antolovich & Cooper 2025). Finally, we have seen that admissions between birth and assessment are associated with a higher likelihood of failure. While in our data this was not found to be significantly higher than the likelihood at other admission times, this may nonetheless be a contributing factor, since as time passes some mothers will be readmitted and missing from their child’s life.

Finally, our data allows us to compare the likelihood associated with a positive EPDS screen to that posed by psychiatric admissions. In this, our work extends that of (Lubotzky-Gete et al. 2021) who looked at a subset of the cohort herein (children born in 2014–2015), and used a somewhat different definition of milestone failure. In particular, their work suggests an increased likelihood of failure in milestone attainment at age 12–24 months for children whose mothers had a positive EPDS, ranging from OR = 1.10 for failure at “walking” (motor milestone), to OR = 1.47 for failure to “points at specific objects” (language milestone). The results herein suggest a similar association with EPDS-positivity (OR = 1.20), yet a stronger association with psychiatric admission (OR = 1.71) even when controlling for EPDS-positivity. This suggests that children may benefit from expanded maternal screening, complementing the use of the EPDS with screening for symptoms of mental distress other than depression, anxiety and self-harm ideation. Since the strongest association was found to be with psychotic disorders, future work might investigate the efficacy of adapting questions from screens for such conditions, such as those in the PQ-B (Loewy et al. 2011), to the peripartum setting.

Limitations

The analysis of the three admission-defined cohorts is limited by the availability of data to admissions at ages 18–41 years. While admission of minors in Israel is usually “a last resort”, this nonetheless may wrongly identify mothers with prebirth admission as being first admitted only after assessment. However, the likelihood associated with these two cohorts is similar, so the impact of this bias is likely minor. In addition, in the pre-birth cohort, we did not differentiate between pregnancy and pre-pregnancy admissions due to the small number of the former, and future work might investigate whether a higher likelihood is associated with it.

A more concerning source of bias is that mothers experiencing more extreme mental distress may be less likely to visit the MCHC. Hence, our analysis may miss the more extreme impact of admission on child development and underestimate the associated likelihood.

Another potential source of bias, when interpreting these findings in a global context, is that Israel is somewhat unique among western countries in its ongoing involvement in violent conflicts, which may shape its mental health landscape. We note, however, that the prevalence of mental illness in Israel is broadly similar to that in other countries (Kodesh et al. 2012; Levinson et al. 2007).

Finally, despite modeling sociodemographic and neonatal covariates, there are likely additional hidden confounders. We do not have information about medical care in the community, in particular birth-related complications and use of psychotropic drugs, nor any information about the fathers or family history of developmental delay. These may not only affect the likelihood of developmental delay, but may prove important in identifying children at risk.

Moreover, the mechanisms underlying the adverse effects of parental mental illness on child development are complex, interrelated, and may exert both direct and indirect influences. These mechanisms encompass both prenatal factors—such as genetic transmission, intrauterine hormonal influences, and exposure to psychoactive medications or substances—and postnatal factors, including parent-infant separation, deprivation, maltreatment, impaired parent-infant interactions, and attachment disruptions (Sutter-Dallay 2021). Hence, the association between admission (or even mental illness) and developmental delay should not be fully interpreted as causal, and addressing maternal mental illness to mitigate the risk of child developmental delay should be accompanied by direct interventions to advance the child.

Conclusions

Our findings underscore the significant impact of maternal psychiatric illness, particularly psychiatric admissions, on children's developmental outcomes. Children of mothers with psychiatric admissions, especially those with multiple admissions or psychotic disorders, have higher odds for developmental delays, with social and language milestones being most affected. While the mechanisms underlying this relationship are likely complex and multifactorial, the data suggests that the timing of maternal admission may be less influential than the underlying chronicity and severity of maternal illness.

These findings highlight the importance of early identification and intervention, not only for addressing maternal mental health but also for directly supporting child development. It is essential to provide close care and support to women with severe mental illness during pregnancy, as well as postpartum, to both them and their children. Future research should continue to explore the long-term effects of maternal mental illness and evaluate intervention strategies that address both maternal and child needs simultaneously.

Authors contributions

YB, GA, VB and MW conceived and designed this work. LA and RY acquired the data. YB analyzed the data. YB, GA, NGS, VB and MW interpreted the results. YB, GA and NGS wrote the paper. All authors contributed to the revision of the manuscripts, reviewed and approved it.

Funding

The authors have no funding to disclose.

Data availability

Data may not be shared due to Israeli Ministry of Health privacy restrictions.

Declarations

Ethical approval

All analyses were carried out in accordance with relevant guidelines and regulations. The study protocol was approved by the Israeli Ministry of Health ethical committee (Ref. 20189308) and by the Sheeba Medical Center ethical committee (Ref. 4438–17-SMC). The need for informed consent was waived owing to the use of de-identified data.

Competing interests

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