To be diagnosed
with ADHD as
an adult, you
have to prove a
childhood you
can’t reliably
remember

The one question every adult assessment asks about the past

Every adult ADHD assessment, whatever else it covers, has to settle one question about a time the assessor was not there for. ADHD is defined as neurodevelopmental — it has to have begun in childhood. So an adult in their thirties, forties or fifties, sitting in front of a clinician for the first time, must establish that the difficulties they are describing now were already present at seven, or nine, or twelve.

The question is reasonable. How it gets answered is the problem. A new narrative review in Brain Sciences, from Alessandro Cuomo, Mario Pinzi and Andrea Fagiolini at the University of Siena, puts the difficulty in a single sentence: adult diagnosis “depends on establishing a childhood onset, yet the primary source of that evidence is usually the adult’s own recollection, filtered through decades of intervening experience.”

School reports are lost or never said anything useful. Parents have died, or become estranged, or remember a different child. Often, what is left is the person in the chair, reaching back thirty years for evidence about themselves.

The review is titled Childhood ADHD and Its Adult Mimics, and its stated concern is the adult whose symptoms might be better explained by something else — anxiety, depression, sleep disruption, trauma, substance use, personality disorder. That concern is legitimate and I will come to it. But inside the review’s section on retrospective recall is a body of evidence that points hard in the other direction, and it deserves to be read on its own terms, because it concerns the gate every late-diagnosed adult has to pass through.

What the evidence says memory does to the answer

The landmark finding comes from Mannuzza and colleagues, as the review reports it. They followed adults with a well-documented history of childhood ADHD and asked them, later in life, about their childhood symptoms. Many failed to recall them accurately. And the error had a direction: the general tendency was towards underreporting, not overreporting.

Sit with what that means for the gate. The people whose childhood ADHD is beyond doubt — because it was documented as it happened — are the people found to understate it when asked to remember. And there is little reason to assume memory is kinder to the undiagnosed adult trying to reconstruct the same kind of history — with no record to check it against, and no one to say it was worse than they recall.

The review adds a second finding that shows how much rests on the method. Estimates of how often childhood ADHD persists into adulthood vary substantially depending on whether they come from the person’s own report or an informant’s, and on how strictly the criteria are applied — “in some analyses by an order of magnitude.” A difference of up to tenfold, depending on who is asked and how. A measurement that can move by an order of magnitude on method is not measuring the person with much confidence.

The authors are careful here, and fairness requires reporting their caution alongside the finding. Studies comparing retrospective accounts with data collected in childhood show that recall of onset is imperfect and “cannot be assumed to err safely in one direction.” Memory does not only under-report; it can also over-report. But notice the phrasing — safely. There is no direction of error that is safe. And the one direction the review documents in people whose childhood ADHD is not in dispute is under-reporting.

Which is why the review’s own guidance on missing records matters more than it first appears: “The absence of records should not automatically exclude ADHD, but it should be documented as a source of uncertainty within a full developmental and differential assessment.” Absence of evidence, in other words, is not evidence of absence — a principle that sounds obvious, and that a gate built on recollection is structurally inclined to forget.

The late-onset cases that turned out not to be late

The most striking evidence in the review concerns a debate that ran through adult ADHD research in the mid-2010s. Three birth cohorts — Moffitt and colleagues in New Zealand’s Dunedin study, Agnew-Blais and colleagues in the UK’s E-Risk twin cohort, and Caye and colleagues in Brazil — found substantial numbers of young adults with clinically significant ADHD symptoms and impairment but no clear childhood history. For a while it looked as though ADHD might genuinely begin in adulthood for some people, which would challenge its definition as neurodevelopmental at the root.

The review notes what those studies relied on: single-timepoint self-report. Then Sibley and colleagues tested the question with a far more intensive method. Using the Multimodal Treatment Study of ADHD cohort, they applied comprehensive, repeated diagnostic assessments between the ages of 10 and 25 to every participant who initially appeared to meet criteria for late-onset ADHD.

Most of the apparent late-onset cases did not hold up. They were traced to three kinds of explanation — and first on the review’s list was an unreported childhood history. The others were a co-occurring condition, such as substance use or another psychiatric disorder, and symptom fluctuation that did not amount to a stable new disorder. Only around one in twenty of the participants initially flagged remained plausible cases of genuine late onset.

That finding will be quoted, and it will usually be quoted as evidence of doubt: most “late” ADHD is not what it seems. But read it again. Part of what looked like ADHD appearing from nowhere in adulthood was ADHD that had been there all along and simply had not been reported. The childhood history existed. The first look missed it. The closer look found it. The difference between those two results is not the presence or absence of ADHD. It is the quality of the instrument pointed at the past.

A gate built on the record the school never made

There is a real overdiagnosis argument, and the review makes it carefully. Screening questionnaires like the Adult ADHD Self-Report Scale ask about difficulty concentrating, restlessness, disorganisation and forgetfulness — none of which is specific to ADHD, and all of which anxiety, depression and poor sleep can produce. The authors are right that a high screening score should be “communicated as screening, rather than diagnostic, results,” and right that “treatment response should not be used to establish or retrospectively validate ADHD.” Trauma can carve a cognitive profile the criteria catch, and a good assessment has to be able to tell the difference.

But the review’s conclusions on inflation are more restrained than its title, and they should be quoted accurately. The drivers of rising adult diagnosis it examines “do not establish population-level diagnostic inflation.” The evidence it reviewed “does not quantify population-level false-positive diagnoses.” And the authors state plainly that “nothing in this review should support diagnostic hesitancy toward patients, particularly women and people with predominantly inattentive presentations, for whom evidence of underrecognition is better established than evidence of overdiagnosis.” I have argued elsewhere about what does drive the diagnostic surge; this review neither settles that argument nor claims to.

What the review also makes visible is who the memory gate falls hardest on. It cites the systematic evidence that women were disproportionately missed as girls, because their presentation leans inattentive and is “more easily absorbed into everyday expectations of a quiet, conscientious girl.” The quiet child, the compliant child, the child who masked — these are the children whose difficulties never became a letter home, a report card comment, a referral. Friction that was never visible was never written down. And decades later, the adult diagnosis asks for exactly that record.

So the people missed as children for structural reasons arrive at adult assessment with the thinnest paper trail, relying on the memory the evidence says tends to understate, and facing a differential that treats the anxiety, low mood and exhaustion of decades of compensation as rival explanations. This is the population arriving in perimenopause with a lifetime of compensation behind them — and among the people the system finds last, which is where the mortality gap is widest.

The review ends with a recommendation that deserves more attention than it will get: the “explicit separation of clinical conclusions from administrative or medico-legal determinations.” It is a call to stop asking a diagnosis to do two jobs at once. That tension is everywhere this week, from the autism report written as a claim form to a government review whose interim report listed “institutional incentives associated with diagnosis” among the processes that may be at work. And here it bites on the adult ADHD gate. A clinical judgement can tolerate uncertainty about a childhood nobody recorded; it can weigh it, document it, and reach a considered view. An administrative gate cannot tolerate uncertainty at all, because uncertainty is what a gate exists to refuse. When the only evidence of a childhood is a memory, and the memory tends to leave things out, a gate that demands proof is not testing whether someone had ADHD. It is testing whether anyone was watching.

Citations

Cuomo, A., Pinzi, M., Fagiolini, A. (2026) — Childhood ADHD and Its Adult Mimics: When First-Time ADHD Diagnoses in Adulthood May Reflect Other Conditions — A Narrative Review — Brain Sciences