Link between flu in pregnancy and children's ADHD and other diagnoses was not seen when siblings were compared

For decades, researchers have asked whether an infection during pregnancy can shape a child’s later mental health and development. Influenza has been at the center of that question, because it is common, it comes around every year, and laboratory work suggests a mother’s immune response could affect the developing brain.

A study of more than 2.7 million children in South Korea and Japan, published on October 8 in JAMA Network Open, offers a careful answer. Children whose mothers had flu while pregnant were slightly more likely to be diagnosed with a neuropsychiatric condition, chiefly attention-deficit/hyperactivity disorder (ADHD). But when the researchers compared those children with their own brothers and sisters, the difference was not seen. That points toward something families share, more than the infection itself, as the explanation, though the sibling comparison was small.

Why flu in pregnancy has been studied for effects on children

The concern has a biological basis. The authors write that maternal immune activation during pregnancy has been implicated in altered fetal brain development. Immune activation means the body’s defensive response to an infection. In animals the evidence is strong. Biological plausibility for this association is well established in animal models, the paper says.

People are harder to study. The authors point out that human research cannot directly assess fetal brain tissue, so it relies on comparing children whose mothers were and were not infected.

Those comparisons have not agreed with one another. A systematic review in 2022 screened more than 12,000 records and included 42 studies of flu in pregnancy and children’s neuropsychiatric outcomes. The studies differed too much to pool. Seven of them concerned autism, and none found a clear link. The reviewers’ verdict was blunt. No conclusions could be drawn about the association between exposure to maternal influenza and neuropsychiatric outcomes due to the limited quantity and quality of available research. They called for large observational studies with long-term follow-up.

There was a second problem, beyond size. Mothers who catch flu while pregnant may differ from those who do not, in ways that also bear on their children. When earlier studies of other infections compared siblings, the paper notes, associations observed in full-cohort analyses often attenuate to the null, raising the possibility of residual familial confounding. Attenuate means weaken, and the null means no difference. Familial confounding is the possibility that something running in families, and not the infection, accounts for a link.

How the study followed children after flu in pregnancy

The study has Jaehyun Kong as first author and Dong Keon Yon as senior author, and its ethical approval came from Kyung Hee University in South Korea. It used insurance records from two countries.

The main cohort came from South Korea, whose database rests on a mandatory single-payer health insurance system covering approximately 98% of the population. The researchers took women aged 13 to 58 years who had a live birth between January 1, 2010, and December 31, 2017, and their linked offspring, and followed the children to the end of 2023. The main analysis was limited to mothers aged 20 or older. Children with certain serious conditions present from birth, such as chromosomal abnormalities, were left out.

That gave 2,564,435 mother-child pairs. In 35,973 of them, or 1.4%, the mother had flu during the pregnancy. A second cohort, from a Japanese insurance database, was used to check the results. It had 181,681 pairs, and 11,198 mothers, or 6.2%, had flu.

Flu was counted from medical records. Exposure was defined as a first recorded diagnosis of influenza infection during pregnancy. Mothers without a documented influenza diagnosis during pregnancy served as the control group. The gap between 1.4% and 6.2% is a reminder that the two systems record flu differently. Influenza testing and diagnostic practices also differed between the 2 countries, the authors write.

The primary outcome was the first postnatal diagnosis of any neuropsychiatric disorder in offspring. Postnatal means after birth, and neuropsychiatric here covers conditions of mental health and brain development. Five groups were counted: ADHD, autism spectrum disorder (ASD), obsessive-compulsive disorder (OCD), intellectual disability and a group of severe conditions. Severe neuropsychiatric disorders included schizophrenia spectrum and other psychotic disorders, bipolar affective disorders, and severe depressive disorders.

Two kinds of comparison followed. The first compared exposed children with all other children, using statistical weighting to balance the two groups on the mother’s age, income, region, number of previous births, pregnancy complications and prior use of health care.

The second looked inside families. Sibling-comparison analyses were performed using exposure-discordant sibling pairs identified through a family-level identifier. Exposure-discordant means the mother had flu in one pregnancy and not in another. Each exposed child is compared with a brother or sister who was not exposed.

What the population comparison found in children

Across the whole population, there was a small difference. In South Korea, children born to mothers with influenza infection had an increased risk of any neuropsychiatric disorder compared with those born to noninfected mothers. The rate was 9% higher, with a range from 4% to 14%. Japan gave almost the same figure, 10% higher, with a range from 2% to 19%.

Two conditions accounted for it. In South Korea the rate of ADHD was 13% higher after flu in pregnancy, and the rate of OCD was 37% higher. In Japan the rate of ADHD was 26% higher. For the rest, the paper reports that no clear associations were observed for severe neuropsychiatric disorders, ASD, or intellectual disability.

Timing seemed to matter. The paper says maternal influenza infection during the first and second trimesters was associated with increased risks of any neuropsychiatric disorder. A trimester is one of the three roughly three-month stages of pregnancy. In South Korea the rate was 18% higher after flu in the first and 9% higher after flu in the second.

On their own, these results look like support for a real, if modest, effect. The overall link appeared in two countries, and in South Korea it was strongest for flu early in pregnancy.

What changed when siblings were compared

Then came the comparison within families. It could use far fewer children, because it needs mothers who had flu in one pregnancy and not in another. There were 33,191 such mother-child pairs in South Korea, 15,698 of them exposed, and 7,842 in Japan, 3,678 of them exposed.

Among them no difference was seen. In South Korea, children exposed to flu before birth had a rate of neuropsychiatric diagnoses 2% lower than their unexposed siblings, with a range from 11% lower to 9% higher. In Japan the rate was 4% higher, with a range from 12% lower to 22% higher. Both ranges sit across zero. The links with ADHD and OCD were likewise no longer distinguishable from no difference.

The authors’ summary is that the association seen in the population was real as a pattern but, in their words, it was not evident in within-family comparisons, suggesting that the population-level estimate may have been influenced by shared familial factors.

They spell out what such factors might be. Characteristics shared within families may be related to both maternal influenza and offspring neuropsychiatric risk, thereby generating a population-level association without necessarily reflecting a direct intrauterine effect. Intrauterine means in the womb. Such characteristics may include shared genetic background and stable patterns of health care use, the latter potentially affecting ascertainment of both exposure and outcome. Put plainly, a family that sees doctors often is more likely to have a mother’s flu written down and also more likely to have a child’s ADHD diagnosed.

One result ran the other way. In the Korean sibling analysis, intellectual disability was 31% more common in exposed children, with a range from 1% to 70%. In Japan, though, this association was not observed in the replication cohort, and the authors advise caution about it.

How this fits with earlier pregnancy infection studies

The pattern, a link in the population that fades among siblings, has been seen before with other infections.

In Sweden, a study of 1,066,956 people born between 1992 and 2002 looked at infections serious enough to put a pregnant woman in hospital. In the whole population the rate of ADHD in the children was more than twice as high. Among siblings it was 3% higher, with a range from 24% lower to 41% higher. The authors concluded that the association between maternal IDP and offspring ADHD is largely due to unmeasured familial confounding, IDP being their shorthand for infection during pregnancy.

A Stockholm study of autism and intellectual disability reached a matching conclusion for autism. The association with autism does not appear to reflect a causal relationship, but is more likely to be explained by factors shared between family members such as genetic variation or aspects of the shared environment. For intellectual disability its authors were less sure, writing that a causal role of maternal infections cannot be excluded. That echoes the unreplicated Korean result.

A Danish study widened the lens to every diagnosis a pregnant woman might receive. It covered all children born in Denmark between 1998 and 2015 and tested 236 maternal conditions against autism. Thirty showed a link. But family-based analyses provided strong evidence for familial confounding in most of the observed associations.

On flu and autism specifically, a California study of children born from 2000 to 2010 had already reported a null result. There was no association between maternal influenza infection anytime during pregnancy and increased ASD risk.

Not every study points the same way. A Swedish cohort followed for up to 41 years examined infections treated in hospital, without a sibling comparison. Its authors reported that fetal exposure to any maternal infection increased the risk of an inpatient diagnosis in the child of autism and of depression. Those were more severe infections than most flu, and the diagnoses counted were ones made in hospital.

What the flu study cannot show

Proof that family factors explain everything. The sibling analysis was much smaller than the main one, so its estimates are less precise. The authors say so directly. This likely reflects limited statistical power in sibling analyses, which are restricted to exposure-discordant siblings, resulting in a small sample size. Statistical power is a study’s ability to detect a real difference. They add that the imprecision of the estimates precludes concluding that familial factors fully account for the population-level association.

Timing within families. The trimester results came only from the population comparison. Too few sibling pairs were available to test them within families, so the first-trimester signal has not been checked against family background.

Which infections were counted. Flu was identified from diagnosis codes, not laboratory tests. The paper accepts that some influenza infections may have been misclassified and that mild or asymptomatic infections may have been missed. Asymptomatic means without symptoms.

Later life. The children were young. The paper says follow-up was limited to childhood and early adolescence, with a maximum offspring age of approximately 13 years. As a result, later-onset disorders, including schizophrenia, bipolar disorder, and OCD, may not have been fully captured. The results, the authors write, primarily apply to diagnoses made during childhood and early adolescence rather than to lifetime neuropsychiatric risk.

Other countries. The authors note that the study population was derived from Asian countries, where health systems and diagnostic habits differ from those elsewhere.

Other harms of flu. The study concerns one group of outcomes. It does not suggest flu in pregnancy is harmless. The same paper says influenza infection warrants attention, given its substantial maternal and fetal risks, and cites earlier work that found an association of gestational influenza with several adverse offspring outcomes, including seizures and late pregnancy loss. Gestational means during pregnancy.

How flu is prevented or treated in a particular pregnancy is assessed by their doctor.

For parents who had flu while pregnant, the study’s message is that a child’s later ADHD or similar diagnosis may owe more to what families share than to that infection, with the caveat that the within-family evidence is still thin.