Overshadowed: does ADHD explain late-diagnosed autism?

More and more people are being diagnosed autistic as adults. A lot of them are ADHD too - about two in five autistic people, by the best pooled estimate. ¹ So I keep wondering whether the ADHD is part of why the autism diagnosis came so late. If a clinician has already found ADHD, do they stop looking?

I know this one from the inside. The psychiatrist who diagnosed my ADHD told me I couldn’t be autistic, because I looked him in the eye. It took me another year of insisting before I got to see someone else - and that time, I came away with an autism diagnosis too.

After that, I went looking for research on whether this happens to other adults… and couldn’t find any. ²

1 But why were we missed?

I wrote about late autism diagnosis in my book Loving Variant Minds, and about how often ADHD comes along with it. Then I moved on, without asking the obvious next question: why is it so often these people who get missed?

Some of the usual answers are well known. Older diagnostic criteria missed anyone who didn’t look obviously impaired. Autistic people learn to camouflage. And clinicians were trained on a picture of autism that looked like a young boy. All of that is true, but none of it is new.

What gets talked about less is that second condition sitting right next to the first. A meta-analysis of 63 studies put the rate of ADHD in autistic people at 38.5 per cent at the time of the study, and 40.2 per cent over a lifetime. ¹ The studies behind that average varied a lot, depending on age, IQ, where people were recruited and which criteria were used, so the true figure depends on where you look. Either way, this is a big group of people.

That number doesn’t answer my question, though. It tells us how often the two conditions turn up together. It says nothing about when each one gets diagnosed, and timing is the whole question. If AuDHD adults got their autism diagnosis at the same age as everyone else, the overlap could be as high as you like and it wouldn’t explain any of the lateness. What I’d want to see is a simple comparison: age at autism diagnosis for adults who already had an ADHD diagnosis, against adults who didn’t. I haven’t found it anywhere.

There’s another catch with the overlap figure, and the meta-analysis points it out itself: where you recruit changes the answer. ¹ An ADHD clinic will find autism at whatever rate it happens to check for it, and an autism clinic will do the reverse. So part of a high overlap figure can come from who gets asked the second question, rather than from the conditions themselves.

2 In children, the delay is real

Here the evidence is solid. Kentrou and colleagues found that children and teenagers who already had an ADHD diagnosis had their autism recognised 1.5 years later on average, and 2.6 years later for girls. ³ Sainsbury and colleagues came at it from the other side. Children who ended up with both diagnoses got their ADHD diagnosis earlier than children with ADHD alone (a median of 79 months against 96), and their autism diagnosis much later (about 120 months against 81). Their parents also remembered less obviously unusual language and social development in the early years. ⁴

A US study of nearly 100,000 children, based on what parents reported, found the same thing on a bigger scale. Among autistic children who also had ADHD, 45 per cent were diagnosed with ADHD first, and that group waited about three years longer for their autism diagnosis. ⁵ In over 13,000 children on Missouri Medicaid, the more diagnoses a child had before autism, the older they were when autism was finally diagnosed. ADHD was the most common of those earlier diagnoses, although it didn’t carry more of the effect than any other. ⁶ And a 2025 study found that having three or more other psychiatric conditions delayed an autism diagnosis by four to six years, with ADHD and depression among them. ⁷

So the direction is clear, and I’ve no argument with it. My problem is the scale. All of these delays are measured in months, or a few years, inside childhood. What I’m asking about is a diagnosis that turns up three or four decades late, in someone who was never assessed for autism until adulthood. Nobody has shown that whatever pushes a diagnosis from age six to age nine is the same thing that keeps it away until 45. It might be. It might also be something else that happens to share the name. Stretching a two-year childhood delay across forty years of adult life is a big assumption, and nobody has tested it.

3 Genes and brains say AuDHD is its own thing

I didn’t expect to find this, and it’s the strongest evidence I’ve got.

Zhang and colleagues looked at genetic and developmental data across four birth cohorts and found two different paths linked to the age of autism diagnosis. ⁸ One goes with earlier diagnosis, lower social and communication ability in early childhood, and only a moderate genetic overlap with ADHD. The other goes with later diagnosis; social, emotional and behavioural difficulties that show up more in adolescence than early childhood; and a moderate-to-high genetic overlap with ADHD and other mental health conditions. That’s a very different kind of evidence from the clinic studies, because it has nothing to do with who got referred to whom. It suggests late-diagnosed autism and ADHD-linked autism may share their genetic and developmental roots.

Two brain-imaging studies point the same way. Watanabe and Watanabe studied resting brain activity in 338 children. In children with both autism and ADHD, the brain pattern behind the social and communication differences was the same as in autism alone. The ADHD-like traits, though, came from a different pattern than in ADHD alone. ⁹ A much larger study by Norman and colleagues, pooling data from 12,732 children and teenagers, found something similar: autism traits and ADHD traits went with different patterns of brain connectivity, even when they occurred together, although the effects were small. ¹⁰

Put those side by side and they change how I see the problem. If autism in AuDHD were simply fainter, its brain pattern would look weaker in children who have both. It doesn’t - Watanabe’s children had the same autism pattern as children with autism alone. What’s different is the ADHD half. It looks enough like ADHD to be diagnosed as ADHD, quite correctly, while working differently underneath. So a clinician can get the ADHD diagnosis completely right and still never think to ask about autism. Nobody has to make a mistake. The ADHD explains enough of what’s in front of them that the autism question never comes up.

Both studies are in children, so the same problem of scale applies. But that’s two separate methods, genetic and imaging, both suggesting AuDHD is more than autism and ADHD added together.

4 In adults, we have the numbers but not the timing

Turn to adults and it flips. There’s plenty of data on how often the two conditions turn up together, and almost nothing on when each one was diagnosed.

A large Medicaid study found ADHD was far more common in autistic adults than in the general population, and that having both went with worse health. ¹¹ A Greek study of newly diagnosed adults found at least one other psychiatric condition in half of those with autism alone, and in nearly three-quarters of those with ADHD or both. ¹² A companion study found the combined group differed from the ADHD-only group in imagination, empathy and attention to detail, and from the autism-only group in childhood hyperactivity and current impulsivity. ¹³ And the Swedish ABIS study found young adults with both conditions had higher unemployment than those with either one alone. ¹⁴

Not one of these asks when the autism diagnosis arrived compared with the ADHD diagnosis, or whether that gap is different for people with autism alone. They show that AuDHD adults exist in large numbers and do worse. They don’t show that we wait longer for the autism half. And nobody has checked whether Watanabe’s and Norman’s brain findings still hold in adults.

5 It runs the other way too, for a different reason

People in the AuDHD community often say it both ways: ADHD hides autism, and autism hides ADHD. The first half is what this piece is about. The second half is true too, but for a different reason.

From 1994 to 2013, the DSM-IV ruled out an ADHD diagnosis in anyone who already had an autism-spectrum diagnosis. ¹⁵ This was a rule written into the manual. DSM-5 dropped the rule in 2013, because the two occur together so often. But a whole generation diagnosed autistic as children before then couldn’t have had ADHD written down, however obvious it was, and many have never been reassessed. Even now, the standard rating scales are poor at telling the inattentive type of ADHD apart from autism-related attention problems - much worse than they are for the hyperactive type. ¹⁶ So the quieter form of ADHD is still the one most likely to go unnoticed inside an autism diagnosis.

That explains why ADHD arrived late, or never, for people already diagnosed autistic. It’s the mirror image of my question, and the two need to be kept apart. Mix them together, and a delay caused by the old DSM rule could easily be mistaken for evidence that ADHD hides autism.

It’s also worth knowing that DSM-5’s change only worked in one direction. No rule ever stopped autism being diagnosed in someone who already had ADHD, so there was nothing to remove. But being allowed to diagnose both doesn’t mean clinicians check for both. The delays in children diagnosed well after 2013, like the Missouri group, suggest the habit hasn’t caught up. ⁶

6 Three explanations, and why it matters which one

“ADHD delays autism diagnosis” could mean at least three different things.

It could be the clinician. An ADHD diagnosis is already on file, the remaining traits get put down to it, and nobody looks further. After the brain studies above, I don’t think that even needs a careless clinician - just one who got the ADHD right and had no reason to ask more.

Sometimes the clinician does ask, and still gets it wrong. Mine ruled autism out on eye contact. Differences in eye contact are only one example the diagnostic criteria give of how autism can show up in non-verbal communication; they aren’t a requirement. ¹⁷ And forcing eye contact is one of the most common ways autistic adults camouflage. ¹⁸ So the very thing he took as proof I wasn’t autistic can be part of how autism gets hidden. I only got a second opinion because I kept pushing. Plenty of people wouldn’t, and they’d still be carrying just the ADHD label.

It could be camouflaging. Something about having both conditions, like more practice at fitting in, makes the autism harder for anyone to spot. Milner and colleagues found that camouflaging predicts age at autism diagnosis, and more strongly in women than men. ¹⁹

Or it could be a different course of development altogether - Zhang’s later path, where the difficulties don’t show up in early childhood, so there’s nothing early to miss. ⁸

Those three point to very different fixes. If it’s the clinician, the fix costs almost nothing: an ADHD diagnosis shouldn’t close the question of autism. If it’s camouflaging, we need better tools, ones that can see autism underneath good compensation instead of counting the compensation as evidence against it. If it’s a different path of development, neither of those helps, because there was nothing early to catch. Then late-emerging, ADHD-linked autism needs to be recognised as a presentation in its own right, instead of being measured against a picture built from three-year-olds.

Gender makes it messier. ²⁰ Gesi and colleagues found that misdiagnosed autistic men were most often labelled ADHD, while misdiagnosed autistic women were more often labelled with a personality disorder - although that rested on only seven men and ten women. ²¹ Girls and women with ADHD are more often the inattentive type, and Hinshaw and Nguyen argue that this quieter presentation is part of why ADHD in girls gets missed. ²² That’s the same type the rating scales are worst at separating from autism. ¹⁶ Add camouflaging, which predicts later diagnosis more strongly in women, ¹⁹ and women with AuDHD may be hit from several sides at once.

That doesn’t mean women’s ADHD is quieter by nature. When symptoms are measured by clinical interview rather than rating scales, there’s no real gender difference in how severe inattention or hyperactivity are; the gap only shows up in rating scales and community samples, which may say more about who gets referred. ²³ What seems to differ is which box people get sorted into, and how visible that box is to the people doing the sorting. The autism gender ratio has the same problem. The old figure of four boys to every girl hasn’t held up. The most careful meta-analysis puts it closer to three to one, ²⁴ and recent birth-cohort data show it narrowing towards even in the youngest children. ²⁵

And then nobody agrees on what “late” means. A 2025 review found studies using cut-offs anywhere from age two to age 55. ²⁶ When a children’s study calls a diagnosis late at nine, and an adult study calls it late at 45, they’re using the same word for two different things.

What I think

My guess is that all three are going on, in different amounts for men and women. Some clinicians stop looking once ADHD explains why someone was referred, simply because the question never comes up. Camouflaging probably accounts for more of it in women than in men. And I suspect some of it will turn out to be Zhang’s later path: autism that just didn’t look like autism in early childhood, and that no amount of clinician training would have caught.

I could be wrong on all three.

But I keep thinking about the adults this affects. Someone who spent thirty years with an ADHD diagnosis and no autism diagnosis has usually been told, one way or another, that the ADHD was the whole story. If a clinician stopped looking, that’s worth knowing, and it’s easy to fix. If their autism simply didn’t look like the version anyone was checking for, that’s worth knowing too. Until somebody measures it, nobody can tell them which one it was.

⁂

---