Prenatal Antidepressants and Maternal Depression Linked to Small Rise in Learning Needs

One of the largest studies ever conducted on the question of whether antidepressants taken during pregnancy affect a child’s developing brain has delivered a nuanced and reassuring-but-cautionary answer. Researchers in Scotland and Wales followed more than 167,000 children born in Wales between 2009 and 2016, tracking them through linked primary care, maternity and education records until 2022, when the children were on average nearly six years old. The analysis, published in PLOS Medicine, examined whether prenatal exposure to maternal depression, antidepressant medication, or both was associated with an increased likelihood of being identified with special educational needs, a broad and objective marker of neurodevelopmental difficulty that captures autism, attention deficit hyperactivity disorder, communication problems, learning difficulties, sensory impairment and emotional or behavioral conditions.

The scale of the dataset is what gives the study its power. Of the 167,447 children included, 12,630, or 7.6 percent, were exposed to untreated maternal depression during pregnancy, meaning their mothers had a recorded depression diagnosis in the two years before delivery but no antidepressant prescriptions. A further 7,005 children, or 4.2 percent, were exposed to treated depression, with mothers who had both a diagnosis and at least one antidepressant prescription during pregnancy or in the month before conception. Another 5,996 children, or 3.6 percent, were exposed to antidepressant medication without any recorded depression diagnosis, a group that typically reflects women continuing medication prescribed for earlier episodes of illness. This four-way structure, exposed to depression alone, medication alone, both, or neither, allowed the team to begin disentangling two effects that are usually hopelessly entangled in observational data.

The headline finding is that both exposures were associated with higher odds of special educational need, but neither effect is enormous in absolute terms. Children whose mothers had depression had a 25 percent higher odds of special educational need than unexposed children, with an adjusted odds ratio of 1.25. Children exposed to antidepressants had a 46 percent higher odds, an adjusted odds ratio of 1.46. When the researchers converted these relative measures into predicted probabilities, the picture became more tangible. Among unexposed children, 20.0 per 100 were identified with special educational need. That figure rose to 23.6 per 100 for children exposed to maternal depression alone and 26.3 per 100 for children exposed to antidepressants without a recorded depression diagnosis. Children exposed to both depression and medication had the highest predicted probability, at 26.5 per 100.

Those absolute numbers are the study’s most important contribution, because relative risks of this size are common in observational epidemiology and often mean far less than they appear to. A 6.3 percent absolute increase in risk, the figure associated with antidepressant exposure among mothers without a depression diagnosis, means that for roughly every sixteen children exposed, one additional child would be identified with special educational need compared with the background rate. Among mothers with depression, the absolute increase linked to medication was smaller still, at 2.9 percent. The researchers attribute this attenuation to a statistical interaction between the two exposures: once depression itself is accounted for, the additional contribution of the medication shrinks, suggesting that a substantial share of the apparent drug effect may in fact reflect the underlying illness or its severity.

The cause-specific results add texture. Antidepressant exposure was associated with autism spectrum disorder at an adjusted odds ratio of 1.59, and with attention deficit hyperactivity disorder at an adjusted hazard ratio of 1.89 in survival models that tracked time to diagnosis. Associations were also observed across the other special educational need categories, including sensory impairment, communication difficulties, learning difficulties, physical and medical difficulties, and behavioral, emotional and social difficulties. The breadth of these associations is itself informative. A medication acting on fetal neurodevelopment would not necessarily be expected to raise risk uniformly across such heterogeneous outcomes, and a diffuse pattern can equally arise from residual confounding, the statistical ghost of unmeasured differences between women who take medication and those who do not.

That confounding problem is the study’s central limitation, and the authors are explicit about it. Women who are prescribed antidepressants during pregnancy are, on average, likely to have more severe or more persistent depression than women whose depression goes untreated, whether because of symptom severity, treatment access, or clinical decision-making. If depression severity itself drives part of the neurodevelopmental risk, then some of the excess risk assigned to medication actually belongs to the illness. The researchers used generalized estimating equation models with an interaction term between depression and medication exposure, adjusted for a wide set of covariates including child sex, area-level deprivation, ethnicity, maternal age, smoking, parity, multiple birth and maternal epilepsy, but no adjustment can fully erase confounding by indication. The team also fitted separate models for individual antidepressant classes and combinations, and used Cox proportional hazards regression alongside flexible parametric survival models for the timing of ADHD diagnoses.

One striking result is the consistency across drug classes. No single antidepressant class showed a markedly different association with special educational need risk, which cuts against a simple pharmacological mechanism, since selective serotonin reuptake inhibitors, serotonin and norepinephrine reuptake inhibitors, tricyclic antidepressants and other agents differ substantially in their pharmacology and placental transfer. The exceptions were tricyclic antidepressants and serotonin and norepinephrine reuptake inhibitors, which showed significant differences in absolute risk of special educational need when treated and untreated depression were compared, at 8.8 percent for each class. With relatively small numbers of children exposed to these less commonly prescribed drugs, the confidence intervals were wide, spanning from roughly 3 to 14 percentage points, so these class-specific estimates should be read with considerable caution.

For clinicians and pregnant women weighing treatment decisions, the practical message is one of balance rather than alarm. Untreated depression during pregnancy carries its own documented risks, including poorer prenatal care, increased preterm birth, and impaired postpartum bonding and caregiving, and the present study confirms that children of mothers with untreated depression face an elevated likelihood of neurodevelopmental difficulty. The additional absolute risk associated with medication appears small, and the authors conclude that the higher risk seen among offspring of mothers with depression cannot be attributed solely to antidepressant exposure. The decision to continue, adjust or discontinue medication therefore remains an individualized clinical judgment, one that this study suggests should weigh a modest possible additional risk against the well-established harms of leaving maternal illness untreated.

What the study cannot yet settle is causation. The authors state plainly that prenatal exposure to antidepressants may confer a small additional absolute risk, but that the findings may partly represent confounding by depression severity, and that further research is needed. Future work will likely need richer measures of depression severity, symptom trajectories and genetic risk, since parental psychiatric illness carries heritable components that no prescription record can capture. Until then, this population-wide analysis stands as one of the most rigorous attempts to separate the effects of an illness from the effects of its treatment, and its central lesson is that the question is not simply whether antidepressants are safe in pregnancy, but how to weigh a small and uncertain medication effect against the demonstrable risks of maternal depression left untreated.

Subject of Research: Prenatal exposure to maternal depression and antidepressants and child neurodevelopmental outcomes

Article Title: Prenatal exposure to maternal depression and antidepressants and neurodevelopmental outcomes: A population cohort study

Article References: McIlvride, S., Rao, N., Singh, S., Nelson, S. M., Pell, J. P., & Fleming, M. (2026). Prenatal exposure to maternal depression and antidepressants and neurodevelopmental outcomes: A population cohort study. PLOS Medicine, 23(9), e1004658. https://doi.org/10.1371/journal.pmed.1004658

Image Credits: AI Generated

DOI: 10.1371/journal.pmed.1004658

Keywords: maternal depression, antidepressants, pregnancy, special educational needs, autism spectrum disorder, ADHD, neurodevelopment, population cohort study, confounding by indication, child health, PLOS Medicine, epidemiology