Quick Answer
An assessment at 45 is not the same procedure as an assessment at 25, run later. It is a different procedure, run on worse evidence, and the reasons are structural rather than personal.
The diagnostic criteria require evidence that symptoms were present before the age of 12. In an adult, the method for producing that evidence is memory, yours or a relative’s. That method has been measured against documented childhood records, and at age 22 it was 55.4 percent accurate. Sensitivity was 32.8 percent.
Nobody has repeated that measurement at 45.
You were asked what you were like in Year 4 and the room went quiet
You said you thought you were fine. You said you got good reports. Your mother, if she is still around to ask, remembers you as a dreamy child who read a lot, which is either the whole answer or none of it.
What you could not say, because it does not fit in the box on the form, is that you have no idea. You were seven. There is nobody left who watched you closely at seven, and the person who did the watching had nothing to compare you to.
The evidence problem, measured
Breda and colleagues used a birth cohort of 3,810 people. Symptoms had been documented at age 11. At 22, the same people were asked to recall their childhood.
Recall of childhood ADHD symptoms was 55.4 percent accurate. Sensitivity was 32.8 percent, which means roughly two thirds of the people who had documented symptoms at 11 did not report them at 22. Positive predictive value was 40.7 percent.
The authors called retrospective recall an unreliable method for characterising the developmental trajectory.
That study was run on 22-year-olds recalling eleven years back. An assessment at 45 asks for a recall across thirty-three years, usually with no documented childhood record to check it against, and often with the people who would have been asked no longer available.
The criterion has not changed. The instrument for meeting it has quietly become something else.
The record that was never created
The second problem is upstream of memory. For a large number of women now in their forties and fifties, there is no childhood evidence because nobody generated any.
The expert consensus statement on females with ADHD describes an experiment: change a child’s name and pronouns from male to female, keep everything else identical, and the child becomes less likely to be referred for support and less likely to be considered suitable for treatment. The statement also notes that parents underestimate impairment in girls while over-rating the same behaviours in boys.
So the absence of a school record is not neutral information about whether something was there. It is a record of who was being looked at.
The thing that helped then counts against you now
The same consensus statement describes compensatory behaviour in girls, socially adaptive behaviour, compliance, coping strategies that mask, and notes these contribute to perceptions that prevent referral in the first place. It also describes women more often receiving a primary diagnosis of an internalising or personality disorder, which delays the right one.
Thirty years later, the compensating shows up in an assessment room as a functional history. Degree, job, children raised. Presented as evidence against, when it is the record of what the compensating cost.
Four differences, side by side
| Element | At 8, or at 25 | At 45 |
|---|---|---|
| Childhood evidence | Current or recent. Teachers and parents available and observing now. | Recalled across decades, by people who may be gone, about a child nobody was watching for this. |
| Presentation | Relatively unmasked. Fewer years of built systems on top. | Thirty years of compensations, some of which now read as evidence of not needing help. |
| What else is in the picture | Usually one thing to explain. | Accumulated anxiety, depression, trauma, and in midlife a hormonal transition, each of which can be offered as the whole answer. |
| Route in | Often a school or a paediatric pathway, at no personal cost. | Self-referred, and adults diagnosed in adulthood report cost and lack of insurance as significantly higher barriers. |
The last row is from a 2025 survey of 129 autistic adults in the US. Adults diagnosed in adulthood reported significantly higher barriers around cost and insurance than those diagnosed as children. What they reported lower was difficulty recognising themselves in descriptions of autism, which is its own finding.
What this is not, before you use it
None of this is an argument that a late assessment is not worth doing. The measured difference in outcome runs the other way, and it sits on a separate page.
The Breda study is on ADHD, in a Brazilian birth cohort, measuring recall at 22. It establishes that retrospective recall is a weak instrument. It does not tell you the size of the error at 45, because nobody has measured that.
The referral-bias experiment concerns children and how adults respond to them. It supports the claim that fewer girls were referred. It does not quantify how many.
None of it says clinicians are acting in bad faith. A clinician following the criteria carefully will still ask for evidence from before age 12, because that is what the criteria require. The gap is in the instrument, not in anyone’s intent.
What can be brought instead
The assessment asks for childhood evidence. Where a school record does not exist, other things stand in, and they carry more weight when they arrive already assembled.
- Siblings, not only parents. A sister who shared a bedroom saw the unfinished projects, the lost things and the night-time reading. A parent saw a well-behaved child.
- School reports, even good ones. “Capable but does not apply herself,” “chatty,” “careless errors,” “would do better if she concentrated.” Those phrases were the observation. They were just filed under character.
- The adult record as a proxy. Job changes, house moves, unfinished qualifications, the pattern of starting well. A CV read sideways is a developmental history.
- What it cost to look fine. Hours spent on tasks other people finished quickly. Systems built and rebuilt. The list of things dropped the moment the compensating stopped working.
- A written timeline. One page, dates on the left. A timeline is harder to set aside than a description.
One sentence for the gap in the record
For when the question about childhood arrives and the answer is that there is nothing on file.
That last distinction is the one worth making out loud. Missing information and a negative finding look identical on a form and mean opposite things.
Sources
- Breda V, Rohde LA, Menezes AMB, et al. (2020). Revisiting ADHD age-of-onset in adults: to what extent should we rely on the recall of childhood symptoms? Psychological Medicine 50(5):857.
- Young S, Adamo N, Ásgeirsdóttir BB, et al. (2020). Females with ADHD: an expert consensus statement. BMC Psychiatry 20:404.
- Abu-Ramadan TM, Tassone AU, Andrzejewski TM, et al. (2025). Diagnostic experiences and barriers to diagnosis among autistic adults in the United States: associations with diagnostic timing and gender. Journal of Autism and Developmental Disorders.
- Holden E, Kobayashi-Wood H (2025). Adverse experiences of women with undiagnosed ADHD and the invaluable role of diagnosis. Scientific Reports 15:20945.
- Attoe DE, Climie EA (2023). Miss. Diagnosis: a systematic review of ADHD in adult women. Journal of Attention Disorders 27(7):645.
The bottom line
A late assessment is run on an evidence base that was never built, using a recall method measured at 55 percent accuracy in people half your age. Finding it harder is not a sign that there is nothing there.
You are not broken. You are not too much. You are smart and capable, and you can do anything with the right tools.
What not knowing cost you, measured. The number on the other side of the assessment.
When they say you are too young and too capable. The two refusals, with a script and a letter.
The 5-year roadmap after late diagnosis. What comes after the assessment.
Perimenopause and ADHD. For when the assessment and the transition arrive together.