Maternal sleep disorders linked to child developmental disorders in nearly 2 million Korean births

Bad nights are part of pregnancy for many women. A pooled analysis of 24 studies put the overall prevalence of insomnia symptoms during pregnancy at 38.2%. Breathing problems in sleep are common too: a review of 37 studies on sleep-disordered breathing during pregnancy put them at nearly 20%. Whether any of this reaches the baby’s developing brain is a question many expectant parents ask, and one that research has struggled to answer.

A study of every live birth in South Korea over five years now offers a far bigger look than earlier research. Tak Kyu Oh and In-Ae Song linked mothers’ and children’s national health insurance records. Children born to mothers with a diagnosed sleep disorder were later diagnosed with a neurodevelopmental condition more often than similar children whose mothers had none. By the end of follow-up, 13.3% of the first group and 9.9% of the second had such a diagnosis. The study was published on October 2 in the Journal of Clinical Sleep Medicine, alongside an editorial that calls the link a real signal and warns about what it cannot show.

The study was observational, which means it can show a link but not that a mother’s sleep caused her child’s diagnosis.

How the Korean study linked mothers and children

The researchers took all live births in South Korea between January 2008 and December 2012, almost 2 million of them, and looked back through each mother’s medical history. A mother counted as having a sleep disorder if a doctor had recorded one in the year before her last menstrual period or during the pregnancy itself. The diagnoses included insomnia, sleep apnea and restless legs syndrome, an urge to move the legs that tends to strike at night.

That turned out to be rare on paper: 33,229 children (1.7%) were born to mothers with a recorded sleep disorder. Each of these children was matched with up to five children of mothers without one, chosen to resemble them on recorded characteristics, a technique called propensity-score matching. The matched cohort comprised 33,183 exposed children and 160,131 unexposed ones.

The researchers then followed the children until the end of 2023, giving up to 15 years of follow-up, and counted neurodevelopmental disorders. These are conditions of brain development that usually show up in childhood: intellectual disability, delays in speech, learning or movement, autism, ADHD, conduct and emotional disorders that begin in childhood, and tic disorders. Here they were counted from diagnosis codes in the same insurance records.

Most earlier work on this question was far smaller. According to the editorial, by Tsou and Tapia, previous studies were constrained by small samples, short follow-up and questionnaires, and the national studies so far had addressed single conditions, such as maternal sleep apnea. A link between mothers’ sleep and children’s development had been long suggested by smaller studies, they write, but none was large enough to confirm it.

How often children received a developmental diagnosis

After matching, the rate of any neurodevelopmental diagnosis was about 36% higher in children whose mothers had a sleep disorder. The absolute figures appear in the editorial, which reports a cumulative incidence of 13.3% versus 9.9%. That is the share of children in each group with a diagnosis by the end of follow-up. For every 100 children, about three or four more in the exposed group had one.

The link showed up for each of the three named disorders, insomnia, restless legs syndrome and sleep apnea, with the rate between 28% and 48% higher depending on which one the mother had. Insomnia made up roughly two-thirds of the diagnosed group, so it drives the overall figure. Only 388 mothers had a recorded sleep apnea diagnosis and 453 had restless legs, and the sleep apnea result in particular was imprecise. The rest of the group carried other sleep disorder codes, which the editorial does not break down, beyond noting that rarer types such as parasomnias were too few to interpret.

Tsou and Tapia add that the association was consistent across intellectual, developmental, and behavioral diagnoses, and that it persisted in singleton and nulliparous births, meaning single babies and first-born children. It also held whether or not mothers had recorded mental health conditions or pregnancy complications, which the editorial calls maternal psychiatric and obstetric comorbidity.

One detail cuts against the simplest explanation. The link was stronger when the disorder was diagnosed before than during pregnancy. The paper reports these timing results as separate estimates, not as a split of the headline 36%. The rate of diagnoses in children was about 31% higher when the mother’s sleep disorder was recorded in the year before the pregnancy began. It was about 18% higher when the disorder was first recorded during the pregnancy. If disrupted sleep during pregnancy were harming the fetus directly, a diagnosis made during the pregnancy might be expected to carry the larger risk. A diagnosis made before pregnancy may mark a longer-lasting problem, or a mother who differs in other ways from those whose sleep troubles began with the pregnancy.

Is a mother’s sleep the cause, or something that travels with it?

There are plausible biological routes. The editorial lists intermittent hypoxia, glucocorticoid signaling, and systemic inflammation, that is, repeated drops in oxygen, stress hormones and inflammation, as ways disrupted sleep could reach the fetal brain. In one rat experiment, exposing pregnant rats to repeated oxygen drops, a hallmark feature of sleep apnea, led to offspring with altered brain connections and behavior that the researchers described as autism-relevant. Rats are not people, and the human study measured nothing of this kind.

The three disorders also arise in different ways, which makes a single explanation unlikely. Pregnancy drives these disorders through distinct pathways, Tsou and Tapia note: weight gain and changes in the airway feed sleep apnea, falling iron stores feed restless legs, and a state of constant alertness feeds insomnia. The three gave broadly similar results. A diagnosis code says nothing about oxygen levels, iron or stress hormones, so the study cannot tell which, if any, of those routes matters for the child.

Other explanations fit the data as well. Sleep disorders cluster with anxiety and depression, and those conditions have their own links to children’s development. In an Italian birth cohort of 3,634 single-born children, maternal anxiety, depression and sleep disorders were each tied to more ADHD symptoms at age 4. Genes that shape a mother’s sleep and mood may also shape her child’s development, and a shared home environment can do the same.

The usual way to test for such family factors is to compare brothers and sisters, one born after a pregnancy with the exposure and one without. This study used matching, which balances recorded characteristics but cannot reach genes or family life, and the editorial lists sibling and cousin comparisons among the next steps. Two other Korean studies, both about sleeping pills and sedatives taken in pregnancy, show how much such a comparison can change. In a BMJ study of more than 3.8 million children, exposed children at first had more psychiatric diagnoses, but the link was gone after familial factors were accounted for. In a 2025 study by Oh and Song, 13.9% of exposed children had a neurodevelopmental diagnosis against 11.4% of unexposed children. Here the link held when siblings were compared, and the authors concluded that neither familial nor genetic confounding fully explains it.

Limits of diagnosis codes in the sleep disorders study

The biggest limit is how the exposure was measured. As the editorial puts it, maternal sleep disorder was defined by a diagnostic code, not a phenotype, meaning a billing record and not a measurement of how badly a woman slept. A code requires a mother to report the problem, seek care and receive a diagnosis, and disrupted sleep is so often normalized in pregnancy that most affected women never get one. That is why only 1.7% were coded when far more pregnant women report symptoms.

The editorial notes that this misclassification can bias estimates in either direction. Undiagnosed mothers in the comparison group would shrink the difference. Mothers who see doctors often for their own sleep may also have children who are more often assessed and diagnosed, which would inflate it.

The children’s diagnoses came from the same insurance data, so they depend on who was taken to a doctor and assessed. A code says nothing about how severe a mother’s sleep problem was or how much sleep she lost, which the editorial writers call the exposure that matters biologically. The results come from one country’s health system and may not carry over elsewhere.

What the results mean for pregnant women with sleep disorders

The study’s authors write that their results “highlight the critical need for early identification and management of sleep disturbances in women of childbearing age to potentially mitigate neurodevelopmental risks in the next generation”. The editorial writers agree that recognition comes first but are more cautious about treatment. Options in pregnancy are narrow, they note. CPAP, a bedside machine that keeps the airway open through a mask, remains among the few clearly acceptable options for sleep apnea. And in their words, “the evidence that treating maternal sleep disorders benefits the fetus remains unfortunately inconclusive”. Drug choices are limited as well: dopamine-acting drugs and opioids for restless legs are avoided in pregnancy, and several sleeping pills are discouraged. Tsou and Tapia call for studies that compare siblings or cousins, that measure sleep directly with recordings, and that track whether treated and untreated mothers’ children fare differently.

How sleep problems in a particular pregnancy are managed is assessed by the woman’s doctor.

Most children of mothers with a sleep disorder in this study, about 87 in 100, had no developmental diagnosis at all. The study points to a link worth chasing, in a minority of mothers whose sleep problems reached a medical record, and leaves the cause unsettled.