Key takeaways
- A late autism diagnosis rarely turns up difficulties you did not know about. It reassigns the cause of difficulties you had already explained some other way.
- The explanations get handed out one at a time, by families, by managers and by clinical intake, and each one is plausible on its own. That is why they never accumulate into a question.
- Competence delays recognition. While the output holds, nobody asks, including you.
- Researchers estimate that more than nine in ten autistic people over fifty in England have no diagnosis (O’Nions et al., 2023).
- The part of a late diagnosis you can actually use points forward, at the environment you are in now.
Contents
- The morning in the kitchen
- What counts as a late autism diagnosis
- What forty years of operating looked like in retrospect
- Every hard thing already had an explanation
- Can autism go undetected until adulthood?
- What the diagnosis changed, and what it didn’t
- The arithmetic is strange, because the career already happened
- The environment was the variable
- Sources
The morning in the kitchen
My wife called me into the kitchen to watch a television segment. An autistic woman who worked at a government agency was describing how she did her job. The hyperfocus. The patterns in data that nobody around her seemed to see.
“You need to see this,” my wife said. “That’s exactly how you process things.”
I was fifty-five. By then I had been a Navy surface warfare officer, a director inside two Fortune 100 companies, and the founder of two businesses of my own. Nobody had ever said the word to me. Nobody had come close.
A few days before that morning, she had told me about something painful in her life and I had sat there with nothing. It was not indifference. I felt the entire weight of it and my speech shut down. The same way it had shut down in a classroom when I was six. I could not talk in school until third grade. Five decades apart, the same mechanism. In between, I had built a complete explanation for it that had nothing to do with mechanism.
Within a few weeks I had an evaluation booked. Finding somewhere to book it was its own piece of work, because most diagnostic centers assess children. The testing ran for hours across several sessions.
What counts as a late autism diagnosis
There is no clinical cutoff that makes a diagnosis late. In practice the phrase means diagnosed as an adult, after a life has already been organized around not having the information.
It is more common than the diagnosis rates suggest. In 2023, O’Nions and colleagues modeled autism diagnosis against English primary care records. They estimated that roughly 750,000 autistic adults over twenty in England have never been diagnosed. For people over fifty, their estimate is that more than nine in ten are undiagnosed. That is an estimate with wide bounds, drawn from English data, and I would not import it directly to another country. It is still the most careful attempt anyone has made to count the people who are not counted.
I should be direct about what this essay is and is not. I am not a clinician. I am an operator with a diagnosis and thirty-seven years in leadership. What follows is my own experience set next to published research, and none of it is an assessment of you. If you are trying to work out whether any of this applies to you, that belongs with a qualified professional. The waiting lists are long. Start early.
What forty years of operating looked like in retrospect
The Navy came first. Four years as a surface warfare officer, including combat operations in the Red Sea during the Gulf War. Then more than a decade inside Dell and Tech Data, up to director level. Then two decades building and running companies, which is what I still do.
Across all of it I held retention rates near ninety-five percent. At the time I could not have told you why. I assumed I was doing something ordinary and that other managers were doing it badly.
Looking back, the pattern is not subtle. Every environment where I did my best work had explicit structure in it. The Navy hands you a written protocol for relieving the watch, spoken aloud, with no ambiguity about who holds responsibility at what moment. I did not experience that as bureaucracy. I experienced it as the only handoff I had ever seen that worked. Later, running my own companies, I wrote down things other people left implicit, because I could not run on implicit. Publishing the rules turned out to help almost everyone, which I took at the time as a lucky accident.
There is a scene from Tech Data I think about more than the rest of it. In the nineties they put every manager through a full day of assessment run by a psychologist and a group of PhD students. Role plays, analytical exercises. I came out of it exhausted. They concluded that my analysis was strong and my social skills were weak. Their response was to fly me to a three-day seminar in Atlanta, at thirty-one years old, to learn how to socialize. Correct conversational distance. How to open and close an exchange. I sat in a room with a handful of other managers who had come out of the same assessment with the same finding.
I was grateful for it then and I am still grateful. Tech Data spent real money developing somebody they could have quietly filed as a poor cultural fit. That decision is most of the reason there was a later career to write about. It also did not occur to anyone in that building that a thirty-one-year-old who needs conversation taught explicitly might be running a different operating system. Including me. The seminar was read as a gap to close. It was evidence.
Every hard thing already had an explanation
This is the part I want to be precise about. It is the mechanism, and it is the reason late diagnosis is ordinary rather than surprising.
Nothing about my life was unexplained. Every difficulty I had came with an account attached, and the account arrived early and stayed.
Going silent when someone I loved was in pain was read as not caring, including by me. Losing a full day after two days of successful meetings was poor stamina, then getting older. Restaurants were something I was bad at. Needing an agenda in advance was rigidity. Not reading as a child was settled before I was eight. My mother told people I was not a reader, in exactly the tone you would use to say somebody is left-handed.
Take any one of those on its own and it holds up. A man who goes quiet when you are upset does look like a man who is not moved. Somebody flattened after a good trip does look like somebody with no stamina. These were not stupid conclusions. They were local conclusions, drawn one at a time, and drawn one at a time they were all defensible.
The failure was never in any single reading. It was in the instrument. Four decades of measurements, every one of them individually reasonable, all of them off by the same amount in the same direction. You do not catch that by re-checking the readings. You catch it by checking what units the thing is calibrated in, and nobody checks that, because the readings look fine.
My mother had the last piece of it the whole time. When I told her about the diagnosis, her first response was about my father. He had not spoken in kindergarten either. He was sent for speech therapy. It was not new information. It had only ever been given the wrong name.
Can autism go undetected until adulthood?
It can, and the mechanism that hides it is not mysterious.
The first part is camouflaging. In 2017, Hull and colleagues interviewed autistic adults about the strategies they use to appear non-autistic. The accounts they collected describe something deliberate, effortful and draining. The finding that matters here is that the strategies work. A strategy that works produces no signal. Nobody investigates a system that is returning correct output, and the better you are at producing it, the longer the investigation is deferred.
A correction before the second part. Camouflaging is usually described as something autistic women do. In 2026, Gundeslioglu and colleagues synthesised thirty-four systematic reviews and meta-analyses on sex and gender differences in autistic characteristics. They found no conclusive evidence of difference in social communication, friendships, play behaviours or motor stereotypies. Autistic women in that synthesis were more capable at camouflaging. Autistic men camouflaged as well. Whatever hides this profile is not confined to half the population, which is part of why a man can reach fifty-five with nobody having asked the question.
The second part is that competence gets read as the absence of difficulty. If your work is good, the difficulty is assumed to be absent rather than paid for somewhere off the books. That assumption is made by managers, by families, and most durably by the person doing the paying.
The third part is that families and managers are not the only ones handing out explanations. Clinical systems hand them out too, and they keep records of it.
In 2026, Gillett and colleagues ran natural language processing across the records of 417,752 adults in UK secondary mental health care, looking at the order in which diagnoses arrive. Among people diagnosed with borderline personality disorder first, the five-year cumulative incidence of a later autism diagnosis was 3.02 percent. Among people diagnosed autistic first, later borderline personality disorder was 3.84 percent. The asymmetry sits underneath those numbers. The usual male skew in autism diagnosis was reduced in the group that came through the other diagnosis first, while the female skew held steady in the group that came the other way. Women waited longer than men for the autism diagnosis when the other one arrived first.
That is a description of what a referral system does. It is not a description of anybody reading this. A process built around one presentation does with an unfamiliar one what everybody else in this essay did. It assigns the nearest available name, and the name holds because it is plausible. The authors address their recommendation to clinicians, which is where it belongs.
The research on people diagnosed in later life keeps finding the same shape. Stagg and Belcher interviewed nine adults, aged fifty-two to fifty-four, who had been diagnosed in their fifties. What their participants had accumulated instead of a diagnosis was a stack of other explanations. Anxiety and depression among them. In several cases, a settled conviction that they were bad people. Lupindo and colleagues interviewed ten men diagnosed in adulthood and found the same sequence: indicators present early, attributed to something else, and the attribution holding for decades.
I want to be plain about where my understanding of this came from. Almost all of it came from autistic adults writing about their own recognition. Mostly online. Mostly unpaid. Mostly years ahead of the published research that now gets cited for it. The language of masking and unmasking, the double empathy work, the insistence that late recognition is a systems failure rather than a personal one. That framing is not mine. I brought thirty years from the management side of a table to a set of ideas that were already finished. The vantage point is the only part I added.
What the diagnosis changed, and what it didn’t
People expect this section to be the turn. It is not.
My capacity did not change. The difficulties are the same difficulties at the same price. Restaurants are still loud in a way that costs me something. The day after a trip is still gone. No one in my life behaved differently because a report existed, and the report did not make me better at any of it.
What changed was the attribution, not the capacity. The difficulties stopped functioning as evidence about my character and started functioning as information about a room.
That sounds like a small change and it is not. It ended a specific and expensive habit. I had been running an internal case against myself, on forty years of exhibits that were mislabeled at the point of collection.
One operational thing changed with it. I stopped trying to correct the readings and started changing the conditions that produced them. The hour after a flight is now empty on the calendar. I did not earn it. I know what it costs to schedule over it. Decisions that used to be settled in a hallway get written down. Neither of those is a coping technique. They are edits to an environment.
The arithmetic is strange, because the career already happened
Nobody warns you about the arithmetic of a late diagnosis. The information arrives after the period in which it would have been most useful. I do not get to run the thirty-seven years again with the manual open.
I also have to be accurate about why the career survived at all. I am a white man who entered the Navy in the eighties and moved up through Fortune 100 companies in the nineties and two thousands. My neurology should have closed doors that my demographics held open. Both of those things are true at once, and an essay that mentioned only the first would be more flattering and considerably less useful.
So I am not going to tell you the diagnosis redeemed anything, or that the hard decades were secretly preparation for something. They were hard decades. Some of the cost was permanent and is not recoverable by understanding it better.
What I will say is that the ledger only closes in one direction. The past does not get repriced. The next Tuesday does.
The environment was the variable
Same person, two environments, different outputs.
The Navy gave me a written watch-standing protocol, standing orders, and a fixed script for who holds responsibility at which minute. In that division I was one of the more effective officers on the ship. Later I sat in an open-plan office where people walked up to your desk mid-task, and where decisions got made in the four minutes before a meeting formally started. There I was expensive to employ. I could not have told you why.
Nothing about my processing differed between those two rooms. What differed was how much of the operating logic was written down. The Navy had made its rules explicit for reasons that had nothing to do with me. The side effect was an environment I could run in at full capacity. The office had left its rules implicit. To everyone there that felt like no rules at all. To me it was a tax I paid all day without an itemized bill.
This is the argument the disability community has been making for decades, under the heading of universal design. It did not originate with me, or with anyone in business. Build the environment for the range of people who will actually use it and you stop needing to make individual exceptions. The design usually turns out better for everybody. I encountered that argument long after I had accidentally been running a weak version of it in my own companies for twenty years.
Which is where a late diagnosis becomes useful rather than merely explanatory. It does not repair the past and it is not a verdict on you. It is an instrument, and it points at a Tuesday. It lets you look at a specific week and say which forty minutes cost more than the work inside them did. Then ask whether that cost is load-bearing, or just how the room happens to be arranged. That question was closed to me for fifty-five years. I had already answered it, and the answer was me.
When the bill comes due over years instead of days, it has a different name. I wrote about that separately, in what autistic burnout actually is.
The instrument was never broken. It was labeled in the wrong units.
Sources
- Gillett, G., Ganduri, S. R., Jewell, A., Spain, D., Happé, F., & Adamo, N. (2026). Gender Bias in Clinical Diagnosis of Borderline Personality Disorder and Autism Spectrum Disorder: A Retrospective Cohort Study. Autism in Adulthood. doi.org/10.1177/25739581261473460
- Gundeslioglu, H., Gray, K. M., Hewitt, O. M., & Langdon, P. E. (2026). Are There Sex or Gender Differences in Autistic Characteristics? An Umbrella Systematic Review. Journal of Autism and Developmental Disorders. doi.org/10.1007/s10803-026-07484-0
- Hull, L., Petrides, K. V., Allison, C., Smith, P., Baron-Cohen, S., Lai, M.-C., & Mandy, W. (2017). “Putting on My Best Normal”: Social Camouflaging in Adults with Autism Spectrum Conditions. Journal of Autism and Developmental Disorders, 47(8). doi.org/10.1007/s10803-017-3166-5
- Lupindo, B. M., Maw, A., & Shabalala, N. (2023). Late diagnosis of autism: exploring experiences of males diagnosed with autism in adulthood. Current Psychology, 42(28), 24181–24197. doi.org/10.1007/s12144-022-03514-z
- O’Nions, E., Petersen, I., Buckman, J. E. J., Charlton, R., Cooper, C., Corbett, A., Happé, F., Manthorpe, J., Richards, M., Saunders, R., Zanker, C., Mandy, W., & Stott, J. (2023). Autism in England: assessing underdiagnosis in a population-based cohort study of prospectively collected primary care data. The Lancet Regional Health – Europe, 29, 100626. doi.org/10.1016/j.lanepe.2023.100626
- Stagg, S. D., & Belcher, H. (2019). Living with autism without knowing: receiving a diagnosis in later life. Health Psychology and Behavioral Medicine, 7(1). doi.org/10.1080/21642850.2019.1684920
Peter Allen Mann. LinkedIn | Blog | SHIELD | ShieldPerception