Pregnancy Diet and Child ADHD: Genetic Confounding

There are observational studies showing that an unhealthy diet during pregnancy is linked to the likelihood of ADHD in the child. It is perfectly natural for something to stir inside a mother or mother-to-be who reads this sentence. A new study that uses the maternal, paternal and child genetic data of three separate cohorts asks this: could part of this association be an illusion coming from the genes that mother and child share? The abstract's answer points to the association possibly being inflated.

Who the study comes from, and where

This post is a Turkish summary of the article published open access in the journal Molecular Psychiatry. The article was published under the Creative Commons Attribution 4.0 (CC BY 4.0) license; the findings belong to the authors of the article, and the account and appraisal here are mine.

The article is in volume 31, issue 10 (2026) of the journal, on pages 5619-5628. Among the authors there are seventeen names, including Kristina Aagaard, Casper-Emil T. Pedersen, Jens Richardt M. Jepsen, Alexandra Havdahl, Laurie J. Hannigan and Klaus Bønnelykke. The abstract does not write the institutions.

The study is an observational genetic analysis: nobody was given a diet, nobody's diet was changed. Three separate cohorts were used. The first is COPSAC2010, 437 mother-father-child trios. The second is MoBa, 41,580 trios. The third is ALSPAC, 1,211 trios. ADHD diagnosis and ADHD trait scores were assessed in the children, and the abstract says this was done at age 10 (the sentence comes up while describing COPSAC2010; in the other two cohorts the abstract does not write how and at what age the measurement was made). The abstract does not give the full names of the cohorts. As far as I know COPSAC is of Danish, MoBa of Norwegian and ALSPAC of British origin (this information is not in the abstract, it is my general knowledge).

What genetic confounding means concretely

The abstract's first step is this: in observational studies an association is seen between unhealthy diet in pregnancy and the likelihood of ADHD in the child. The question is what the cause of this association is.

Say some of a mother's genes slightly affect her food preferences. Let the same genes, or genes linked to them, also have a connection with ADHD predisposition. The mother passes these genes to her child. In the end ADHD is seen more often in the children of mothers who eat irregularly, but the reason may not be the food itself. The child carries the predisposition to ADHD in the genes, not on the mother's plate. There is a real association, but this association cannot be read as "she ate it, so it happened". This is what is called genetic confounding. This example is my tool for explaining; the abstract does not say which genes are shared. That the heritability of ADHD is high is established knowledge, with more detail in the article on whether ADHD is hereditary.

How the trio model works

The authors used a polygenic score (PGS). This is a calculation that adds up the many small differences in a person's genetic makeup into a single number (this definition is general knowledge). Here it was calculated for "genetic predisposition to a healthy dietary pattern" and derived separately for mother, father and child.

I understand the logic of the trio model like this (this explanation is my interpretation, the abstract does not elaborate on it). The child receives half of the mother's genes. If the mother's score is linked to the child's ADHD, there are two possibilities. Either because the child carries these genes in their own body (a direct genetic effect, that is, confounding), or because the environment the mother creates through these genes, for example her diet, affects the child (an indirect genetic effect). When the child's and the father's score are taken into account, the part passed directly to the child is filtered out and what remains is the mother's effect through the environment. The abstract says that it looked at whether the mother's score is still associated after the child's and father's scores are taken into account.

What they found

In the COPSAC2010 trios (437), the mother's healthy diet score, even after the child's and father's scores were taken into account, was associated with a lower ADHD trait score in the child. The authors interpret this as indirect genetic effects consistent with a causal effect from maternal diet. The abstract does not give the size of the effect or its statistical values.

Then came the replication attempt. In MoBa (41,580 trios) and in ALSPAC (1,211 trios) this finding was not replicated. The authors write that in these two cohorts the direct genetic effect estimates point to genetic confounding playing an important role. I read this as follows: in these two cohorts the signal appeared in the part coming from the genes the child carries rather than from the mother's environment (my interpretation).

What a mixed result means

A finding that came out in one cohort and did not in two is not a situation to be taken lightly in science, but not one to be exaggerated either. In the smallest cohort there is a signal, and in the two large cohorts there is, instead of it, a confounding signal. As general knowledge, as the sample grows the estimates are expected to be more precise. So a result that does not appear in 41,580 trios is less easily put down to chance than a result that appears in 437 trios. Even so, the abstract does not explain why the result appeared in the small cohort, and I will not speculate.

The authors' overall interpretation is this: these genetic results show that genetic confounding is a way in which the association between maternal diet in pregnancy and ADHD in the child can be inflated, and they do not give solid evidence consistent with a causal pathway. "No solid evidence" does not mean "certainly no effect". The difference between these two sentences is important.

Why it cannot be used to blame the mother

The most common mistake when reading an observational association is to count something seen next to another thing as its cause. This study shows exactly how that mistake can arise. If part of the association comes from genes, then for that part neither has the mother done anything wrong, nor has the child got ADHD because of diet. The same genes may carry both the preference and the predisposition, and nobody chooses this.

I covered the possible causes of ADHD separately in the article on what causes ADHD. I summarised the Swedish study that looks at the health outcomes of children of parents with ADHD in a separate article, and the same principle applies there: occurring together does not show the cause. If the mother or father has ADHD themselves, being a parent with ADHD is tiring enough already, and there is no need for the blame of "it happened because of your diet" on top. Stigma is very often fed by exactly this kind of simple causal sentence.

What the abstract does not say

The authors themselves state the limits of the study. First, the predictive power of the polygenic score is limited: it explains only a small part of the differences in diet during pregnancy. I read this as limiting the method's power to discriminate (my interpretation). Second, diet is not a single thing, it is an exposure with many components. The score captures a general "healthy dietary pattern" but says nothing about a particular food or deficiency.

The abstract also does not say: which foods or which period of diet might matter, the effect sizes, the confidence intervals, who made up the cohorts, how in each cohort ADHD was assessed, whether only as a diagnosis or as a trait score. That a balanced diet in pregnancy matters for other health outcomes is general knowledge; this study says nothing about them.

What this finding means for you

If you are pregnant or your child has ADHD and you are looking back and questioning yourself, this study does not say to you "you are not guilty", because it is not a study that would make such a claim. But it shows this: part of the association seen between diet in pregnancy and ADHD may have a second explanation, such as shared genes. This is information that should lighten the guilt you carry.

For the adult reader thinking about their own ADHD, this also applies: this is not a verdict on your own childhood. Cohort averages do not describe individual people.

What I take from this

I do not read this study as "diet is unimportant". What I read is this: for the association between diet in pregnancy and ADHD in the child, we do not have a solid enough basis to turn it into the simple cause-and-effect story it is often told as. Moreover this story often tells the association in one direction and puts it on the mother.

I think the most valuable side of this study is that it tested a result from one cohort against two other cohorts and wrote openly that it did not replicate. This is what science should do. The practical result for someone who is pregnant is simple: arrange your diet for your own health, within your own means. Do not see it as an exam for your child's ADHD.