JAMA has
mapped what
“neurodivergent”
denotes in the
research —
and suggests
reserving the
word for identity,
not diagnosis

A field finally audits its own foundational word

For years now, researchers have recruited “neurodivergent” participants, measured “neurodivergent” outcomes, and published findings about “neurodivergent people” — without the field systematically checking what the word actually gathers when a study reaches for it. The assumption was that everybody knew. The word did so much work, so fluently, that auditing it felt unnecessary.

But somebody has now done the audit. A team led from the University of Southampton — Alessio Bellato as first author, Samuele Cortese as last, twenty-two authors in all, including experts with lived experience and representatives of ADHD-Europe and Autism-Europe — has published a scoping review in JAMA Psychiatry asking a question so basic it is remarkable the field ran this long without an answer: when the empirical literature says neurodivergent, which clinical diagnoses does it mean?

The method was orthodox and thorough: JBI scoping review methodology, the PRISMA extension for scoping reviews, four databases searched to August 2025, and an inclusion rule targeting studies that recruited participants defined as neurodivergent or neurodiverse. What came back was 134 studies and 151,134 participants.

The numbers: 134 studies, six diagnoses, and a fifth that named none

Here is what the terms were found to hold. Across the 134 studies, the diagnosis most commonly associated with neurodivergence and neurodiversity was autism, appearing in 92 studies (68.7%). Then ADHD in 86 (64.2%), dyslexia in 44 (32.8%), dyspraxia in 21 (15.7%), anxiety disorder in 20 (14.9%), and dyscalculia in 17 (12.7%).

Read that list again and stop at the fifth entry. Anxiety disorder is not a neurodevelopmental condition under any diagnostic manual in use. It is one of the most common mental health conditions in the population — and it is now travelling, in roughly one study in seven, under the “neurodivergence” banner. Nobody voted on that. No nosology committee sat. The word simply absorbed it, the way the word absorbs things.

The distribution above the fold tells its own story, too. The term was coined broad — a banner for every departure from the cognitive norm — yet in research practice it has collapsed onto the big two. Autism and ADHD appear in two-thirds of studies each; dyslexia in a third; dyspraxia and dyscalculia, conditions the umbrella was supposed to shelter on equal terms, trail at 15.7% and 12.7%. Whatever the word’s founding intention, its working meaning in the literature is autistic and/or ADHD, approximately — with everything else present mostly as garnish.

Then there is how the participants entered the studies at all. In 56 of the 134 (41.8%), participants were identified by self-reported diagnosis — the researchers took their word that a clinician, at some point, said so. In 50 studies (37.3%), participants were identified by self-identification — the identity itself the basis for inclusion.

And 29 studies — 21.6% of the literature reviewed — did not specify any clinical diagnosis at all. A fifth of the studies reviewed never said which, if any, clinical diagnoses their participants had.

Heterogeneity is a polite word for a category that will not hold

None of this is an argument that self-identification is illegitimate, and it is worth being precise about that before the harder point. People self-identify for reasons this corpus has examined at length, assessment routes are rationed and expensive, and a person’s account of themselves is data of a kind. The problem is not that these studies exist. The problem is what happens when they aggregate.

A study whose every participant carries a clinician-confirmed autism diagnosis and a study whose participants ticked a box marked neurodivergent are, in the literature as it now stands, the same kind of study. Both are “research on neurodivergent people”. Both flow into the same review articles, the same workplace guidance, the same commissioning decks, the same training slides. When a consultancy tells an employer that “neurodivergent employees report X”, or a service is designed around what “the neurodivergent community needs”, the evidence underneath is a pooled object in which a fifth of the pool never specified a diagnosis — and no one downstream can un-pool it.

And the problem is not confined to the fifth of studies that specified nothing. The largest single group — the 41.8% that identified participants by self-reported diagnosis — rests on participants’ own reports of an assessment rather than a confirmed record of one. That may be entirely reasonable in context. But it is a different evidential object from a confirmed diagnosis, and the pooled literature treats the two as if they were the same.

None of this is unfixable, which is what makes the current state a choice rather than a fate. A study can state its ascertainment in one sentence: diagnosis confirmed, diagnosis self-reported, identity self-declared. A review can stratify by that sentence. The moment the literature does either, the pooled object separates back into populations and the findings become interpretable again. The fix costs a line of text per paper. The fact that a JAMA Psychiatry review was needed to point this out tells you how little demand there has been for the line.

Research runs on defined populations. That is not pedantry; it is the entire mechanism by which a finding about a sample becomes knowledge about the world. A category that ranges silently across confirmed diagnosis, reported diagnosis, adopted identity and unspecified is not a population. It is a word. I have argued before that ADHD is better understood as a construct than a natural kind — this review supplies the measurement for the umbrella term above it. The category’s failure to hold is no longer only an argument. It is a finding in JAMA Psychiatry.

The conclusion the review draws — and the split it suggests

The authors’ key-points summary states the finding this way:

These findings suggest that while neurodivergence and neurodiversity are valuable terms for capturing lived experience, their lack of specificity may introduce substantial heterogeneity in research focused on clinical categories.

The full abstract then takes the step that matters most. The authors suggest “prioritizing precise diagnostic descriptors in clinical practice and research involving clinically defined conditions, while reserving neurodivergent for studies and contexts focused on self-identification and identity.”

Read that suggestion slowly, because it splits one word into two registers. Neurodivergent for identity and self-identification — the word answering who someone is, where it is, in the authors’ own phrase, valuable for capturing lived experience. And precise diagnostic language wherever the question is classification — how someone is to be counted, compared and treated. One word has been doing two jobs, in two registers that do not convert, and a team of the field’s own senior figures, publishing in one of psychiatry’s leading journals with lived-experience co-authors, has now suggested separating them.

That is the fault line this publication keeps returning to, reached from the other side. The difference between the question of who someone is and the question of how they are to be sorted sits at the centre of The Neurodiversity Book. Here it arrives not as philosophy but as a research suggestion, with data underneath it.

What happens next is the familiar part. Suggestions in abstracts do not rewrite methods sections. Commissioners will keep buying research “on neurodivergent people”; the incentives that made the identity worth claiming have not moved; and nothing obliges a researcher to follow the advice.

But something has shifted, and it is worth stating plainly. Until 16 September, a researcher recruiting “neurodivergent” participants without specifying them could claim the field’s shared understanding as cover. That cover is now gone. The lack of specificity is published, quantified and peer-reviewed in JAMA Psychiatry, alongside an explicit suggestion for what to do about it. From here, using the word without saying what it means is not a convention. It is a choice — and the literature that results will be measuring the choice, not the people.

Citations

Bellato, A., Long, M.R.P., Barilà, M., Frigeri, G., Rizzo, L. et al. (2026) — Mapping Clinical Diagnoses Associated With Neurodiversity and Neurodivergence in Mental Health Research: A Scoping Review — JAMA Psychiatry