News · Brain & Mental Health
Genetic risk fell as autism and ADHD diagnoses rose in Denmark
A JAMA Psychiatry study of 37,000 Danes found people diagnosed more recently carried less genetic liability, pointing at broader criteria rather than a new environmental cause.
Based on a peer-reviewed national case-cohort study in JAMA Psychiatry
- Researchers used the Danish iPSYCH2015 case-cohort to study 17,071 people first diagnosed with autism and 20,111 with ADHD between 1994 and 2016, publishing in JAMA Psychiatry.
- They tracked whether the average genetic liability of newly diagnosed people changed over those 22 years.
- It fell. A more recent ADHD diagnosis came with lower genetic risk for ADHD (-0.06 SD per 10 years; 95% CI, -0.09 to -0.03).
- The same held for autism (-0.07 SD per 10 years; 95% CI, -0.10 to -0.04), and for related traits including bipolar disorder and schizophrenia.
- Empirical trends were compared against simulations of what different explanations would predict.
- The pattern matched broadening diagnostic criteria, not the arrival of a new risk factor.
- Only 24.6% of the autism cases and 29.6% of the ADHD cases were female, so the trend is dominated by males.
- Polygenic scores describe groups. They say nothing about any individual, and a lower average score does not mean a diagnosis is less real.
Autism and ADHD diagnoses have climbed steeply for three decades, and the argument about why has largely been conducted without evidence that could settle it. A study in JAMA Psychiatry found a way to test the competing explanations against each other, using the genomes of the people being diagnosed.
The authors state the problem directly: the incidences of attention-deficit/hyperactivity disorder (ADHD) and autism spectrum disorder (ASD) have increased markedly over recent decades, raising concerns about the emergence of new risk factors. Existing explanations point at diagnostic practice, stigma and awareness, but critically few studies have explored changes in underlying risk factors.
The idea
Genes do not change over 20 years. Diagnostic criteria do.
So if you measure the average genetic liability among people newly diagnosed each year, you have a way of distinguishing explanations. A genuinely new environmental cause would pull in people regardless of their genetics, leaving the average roughly flat. Loosening the threshold for diagnosis should pull in people with less liability, dragging the average down.
The researchers set out to assess changes in the genetic risk profile of individuals diagnosed with ASD or ADHD by year of incident diagnosis and explore results in comparison with those expected under simulated scenarios.
The data
This needs a country that keeps complete records and has genotyped a large slice of its population. Denmark is one of very few.
The study used the Lundbeck Foundation Initiative for Integrative Psychiatric Research (iPSYCH2015) study, a population-based case-cohort in Denmark among individuals with incident diagnoses for ASD and ADHD made from 1994 to 2016.
The numbers are substantial: 17,071 individuals with incident diagnoses for ASD and 20,111 with diagnoses for ADHD. Mean age at diagnosis was 9.7 years for autism and 11.0 for ADHD.
Both samples are heavily male - 24.6% female for autism, 29.6% for ADHD - which reflects who was being diagnosed in Denmark across that window and is itself part of the story other research is now telling.
What happened to genetic liability
It went down, consistently, across both conditions.
A more recent ADHD diagnosis was associated with decreased genetic risk for ADHD, falling 0.06 standard deviations per decade. Autism did the same: a more recent ASD diagnosis was associated with decreased genetic risk for ASD, at 0.07 standard deviations per decade.
The decline was not confined to the condition being diagnosed. Genetic liability for other disorders, including ASD, bipolar, and schizophrenia, also fell among people newly diagnosed with ADHD. The same broad drift appeared for autism, extending to bipolar disorder, schizophrenia, and educational attainment.
That breadth matters. A drop confined to one score could be a quirk of one measurement. A drop across a whole family of related scores looks like a change in who is crossing the threshold.
Why simulation was needed
A trend on its own does not identify a cause. The researchers’ second move was to work out what each competing explanation should look like, then check.
By analyzing multiple polygenic scores together with simulated expectations, this study could disambiguate among competing hypotheses about scenarios that may be associated with increased rates of ADHD and ASD diagnoses.
The verdict: findings support broadening diagnostic criteria as an explanation for increasing rates.
What this does not say
It does not say recent diagnoses are wrong, or milder, or less deserving of support. Genetic liability is one input among many, and a person with a lower polygenic score can be as impaired as one with a higher score. The criteria were widened on purpose, because the narrow versions were missing people.
Nor does it rule out every environmental contribution. The design detects explanations that shift the genetic composition of the diagnosed group. Something affecting people uniformly, regardless of genotype, would be harder to see, though it would also have to account for the observed decline rather than merely coexist with it.
The Danish setting is both the strength and the boundary. Complete registers and universal access make the analysis possible and make diagnosis unusually evenly distributed. In systems where getting assessed depends on money or persistence, the mix of who gets diagnosed will follow different rules.
And the window closes in 2016. What has happened since, including the sharp rise in adult and female diagnoses reported elsewhere, is outside this data.
Why it matters
The rise in these diagnoses gets treated in public argument as evidence that something is happening to children. This study is the closest thing yet to a direct test of that claim, and it points somewhere less alarming and more mundane: the definition moved.
People also ask
What is a polygenic score?
A single number summarizing thousands of common genetic variants, each contributing a small amount of liability to a trait or condition. It is a statistical aggregate, not a test. Scores overlap heavily between people who have a condition and people who do not, which is why they are useful for comparing groups of thousands and close to useless for telling any individual anything.
Why would falling genetic risk explain rising diagnoses?
Think of a threshold. If the bar for diagnosis is very high, only people with substantial liability cross it, and the average genetic score among diagnosed people is high. Lower the bar and you additionally pick up people with less liability, so numbers go up and the average genetic score goes down. That is precisely the pattern found. A new environmental cause would not produce it, because an external trigger does not change anyone's genome.
Does this mean recent diagnoses are wrong or less valid?
No, and this is the most important thing to get right. Genetic liability is only one contributor to either condition, and a lower polygenic score says nothing about whether someone genuinely meets criteria or genuinely needs support. Broader criteria were adopted deliberately, because narrow ones were missing people who struggled. The finding describes who is now being identified, not whether they should be.
How do simulations strengthen the argument?
Because several explanations for rising rates predict different genetic trends. More awareness, more stigma reduction, a new environmental exposure, broader criteria: each implies a different pattern in the average polygenic score over time. The researchers modeled what each scenario should produce and compared those to what the data actually showed. That turns a single correlation into a way of ruling explanations in and out.
Does this apply outside Denmark?
Cautiously. Denmark has near-complete national registers and universal healthcare, which is why this analysis is possible there and hard elsewhere, and it also means diagnostic access is unusually even. Countries where diagnosis depends on insurance or private assessment may see different dynamics. The genetic samples are also of European ancestry, and polygenic scores transfer poorly across ancestries.
References
- LaBianca S, Lousdal ML, Krebs MD, et al. Changes in Genetic Contributions to ASD and ADHD by Year of Diagnosis. JAMA Psychiatry (2026).
- Centers for Disease Control and Prevention. Data and Statistics on Autism Spectrum Disorder.
- National Institute of Mental Health. Attention-Deficit/Hyperactivity Disorder.