Depression during pregnancy creates a difficult clinical trade-off. Untreated illness can affect maternal health during pregnancy and after birth, while patients and clinicians also need reliable evidence about the longer-term outcomes associated with antidepressant treatment. A large population study from Wales now adds unusually broad evidence to that discussion by linking maternal health records, prescribing data and children’s education records.
The study, published in PLOS Medicine, followed 167,447 children born in Wales between 2009 and 2016 through 2022. The researchers found that children whose mothers had depression before or during pregnancy were more likely to have special educational needs later in childhood, whether or not the mothers had received antidepressants. Antidepressant exposure was associated with a further increase, but the size of that additional difference was modest among children whose mothers had depression and cannot be assumed to have been caused by the medication itself.
A population-wide view of pregnancy and childhood development
Saraid McIlvride and colleagues at the University of Glasgow used linked routinely collected health, maternity and education records. Children in the cohort had a mean follow-up age of 5.68 years. Rather than relying on parental recall, the study identified maternal depression diagnoses and antidepressant prescriptions from healthcare records and connected these exposures with later recorded educational and neurodevelopmental outcomes.
Antidepressant exposure was defined as one or more prescriptions during pregnancy or in the month before conception. Maternal depression was identified from diagnoses recorded during the two years before delivery. This allowed the researchers to distinguish several clinically important groups, including children whose mothers had neither recorded depression nor antidepressant exposure, children exposed to maternal depression without antidepressant treatment, and children whose mothers had both depression and antidepressant treatment.
The main outcome was special educational needs, a broad educational measure that can include support for learning, behavioural, emotional, social or developmental difficulties. The researchers also examined specific neurodevelopmental outcomes, including attention deficit hyperactivity disorder and autism spectrum disorder. Their statistical models accounted for a range of measured characteristics, including child sex, deprivation, ethnicity, maternal age, smoking, parity, multiple birth and maternal epilepsy.
The largest difference was associated with maternal depression itself
The absolute numbers help put the findings into perspective. Among children who were not exposed to either recorded maternal depression or antidepressants, about 20 in every 100 had special educational needs. Among children whose mothers had depression but were not treated with antidepressants, this rose to about 24 in every 100.
For children whose mothers had depression and received antidepressant treatment, the figure was about 27 in every 100. In other words, within the group already exposed to maternal depression, antidepressant treatment was associated with roughly three additional cases of special educational needs per 100 children compared with untreated maternal depression.
The modelling produced a similar message. When the researchers compared exposure patterns, antidepressant exposure was associated with an estimated 2.9 additional cases of special educational needs per 100 children among those exposed to maternal depression. Among children without a recorded maternal depression exposure, antidepressant exposure was associated with an estimated 6.3 additional cases per 100 children.
These results matter because they argue against a simple interpretation in which antidepressant medication alone explains the developmental differences. Children whose mothers had depression already showed higher rates of educational support needs even without antidepressant exposure.
Why treatment cannot be separated cleanly from illness severity
The central methodological problem is confounding by indication. Women with more severe, persistent or recurrent depression may be more likely to receive antidepressants during pregnancy. If severity itself is associated with later outcomes in children, some of the apparent medication association can arise because the treated and untreated groups were different before treatment was considered.
The linked-record design allowed the researchers to adjust for several important measured characteristics, but administrative data cannot fully capture depression severity, symptom history, treatment adherence, psychosocial stress, genetics, family environment and every other factor that could influence child development. A prescription record also shows that medication was prescribed, not necessarily exactly how consistently it was taken.
For that reason, the study identifies associations rather than proving that antidepressants caused the additional educational needs. The authors explicitly caution against interpreting the findings as a reason for women to stop medication automatically. Untreated depression also carries risks, including effects on maternal functioning and a higher risk of postnatal depression.
A potentially useful signal for schools and healthcare teams
One practical implication is that prenatal mental-health history could help identify children who may benefit from closer developmental or educational observation. The researchers suggest that antidepressant exposure during pregnancy may function as a marker for additional support needs at school, even if medication itself is not the cause of those needs.
This distinction is important. A risk marker can be useful without being a causal mechanism. If maternal depression and treatment history identify a group with a somewhat higher probability of later support needs, health and education systems could potentially use that information to improve early recognition while avoiding the unsupported conclusion that treatment caused the outcome.
The findings also illustrate why absolute risks are valuable in communicating observational health research. Saying that an exposure is associated with increased risk can sound dramatic without context. Here, the difference between untreated and treated maternal depression was approximately three children with special educational needs per 100, while both groups had higher rates than the approximately 20 per 100 observed in the reference group.
What the study can and cannot establish
The study’s major strength is scale. More than 167,000 children were studied using population-level records that linked maternal health and prescribing information to outcomes observed years later in education and healthcare systems. This reduces some of the recall and selection problems that affect smaller questionnaire studies.
Its limitations are equally important for interpretation. The research was observational, so it cannot establish causality. Residual confounding by depression severity is a particularly important concern. Prescription records do not guarantee medication use, and routinely collected administrative records can miss diagnoses or circumstances that were never recorded. The cohort was also drawn from Wales, meaning healthcare, prescribing and educational-support systems may differ from those elsewhere.
The average follow-up age was still relatively young, so some neurodevelopmental or educational difficulties that become clearer later in childhood may not yet have been identified for every participant. Longer follow-up could therefore change estimates for particular outcomes.
Overall, the study supports a more nuanced interpretation than either reassurance that antidepressant exposure has no association with later development or a claim that treatment is directly harmful. Maternal depression itself was associated with greater educational and neurodevelopmental needs, while antidepressant exposure was associated with an additional difference that was comparatively small among mothers with depression and remains vulnerable to confounding. Decisions about treatment during pregnancy therefore still require individual assessment of both the risks of illness and the uncertain risks associated with medication.
Source Information
Study: Prenatal exposure to maternal depression and antidepressants and neurodevelopmental outcomes: A population cohort study
Authors: Saraid McIlvride, Neha Rao, Sarjit Singh, Scott M. Nelson, Jill P. Pell and Michael Fleming
Journal: PLOS Medicine, Volume 23, Issue 9, e1004658
Published: 29 September 2026
DOI: 10.1371/journal.pmed.1004658
