The diagnosis was doing the triage

Support should not be withheld from a child while they wait for a diagnosis. That is the position set out in the executive summary of the independent review into mental health conditions, ADHD and autism, updated on 21 October 2026. It is the right call, and it quietly moves a job onto schools that most of them have no record-keeping for.

For years a diagnosis did two separate jobs. It described a child, and it triggered help. Take away the second job and something else has to trigger help instead. The review is explicit about what that something is: identified need. Which means a school's own written account of what a child cannot yet do, and what has been tried, stops being paperwork and becomes the thing that decides whether support arrives.

Most schools do that work. Very few can produce it on request.

What did the review actually say?

Recommendation 2 asks for a move from a diagnosis-dependent system to what the review calls a needs-led plus diagnosis system, supported by a shared needs assessment and prioritisation framework. The principle underneath it is stated plainly: support should be available according to a person's needs before, during and after assessment, and not be withheld while waiting for a diagnosis.

It goes further on waiting lists. No one should be taken off a waiting list without being connected to appropriate help, and alternative support should be actively arranged with the person rather than simply signposted. Signposting, in other words, is not support.

All figures below are from that executive summary, updated 21 October 2026.

Why does this shift work onto schools?

Because the waits are long enough that the school is the only service in continuous contact with the child. Open autism referrals stood at more than 254,000 people in December 2025, with 90 per cent waiting longer than thirteen weeks. Open ADHD referrals passed 560,000 by the end of 2025, seventeen times the 2019 figure, with community paediatric waiting lists at around 173,000.

The review also records that difficulties get worse while people wait. Among those waiting under two weeks, around one in four reported worsening difficulties. Among those waiting more than six months, it was more than seven in ten.

A child in that queue is in school every day. If help is to arrive before the diagnosis does, the school is where the evidence of need has to come from.

What counts as evidence of need?

Not a label, and not a professional's letter. In practice it is four things held together: what the child could not do, what the school put in place, what changed, and over what period. The last one is what makes it evidence rather than an opinion. A single observation is an anecdote. The same observation across a term, with an intervention in the middle of it, is a case.

The review notes that around one in five secondary pupils now identify as neurodivergent, well above the proportion with a formal diagnosis. Separately, SEND identified as autism rose from 1.5 per cent of pupils in 2016 to 3.3 per cent in 2024. The gap between those two numbers is the population a needs-led system is supposed to reach, and it is precisely the population with no document attached.

Why is that record so hard to produce?

Because it was never designed as a record. It accumulates as a by-product of other duties and then scatters. A provision map updated when somebody had an hour. A teaching assistant's intervention log in a notebook. Pastoral notes in one system and attendance marks in another. A conversation in a corridor that changed what a teacher did on Tuesday and was never written anywhere.

None of that is negligence. It is what happens when the people who hold the evidence are the people with the least time to file it. We wrote about the two inventories this creates in where the evidence lives, and about what happens when the deadline that used to force the filing is removed in the deadline was doing the work.

The diagnosis used to paper over the gap. A letter arrived, and nobody asked the school to show its own working. Needs-led support removes that cover.

What can a SENCO do this term?

Pick one child who is waiting, and try to assemble their case from the systems you already have. Not to file it anywhere. Just to find out how long it takes and how many places you have to look.

That exercise tells you something a policy cannot. If it takes forty minutes and three colleagues' memories for one child, then a needs-led system is not going to work on a caseload of sixty, however good the intentions behind it are. And it tells you which of the four elements is missing: usually not the need and not the intervention, but the period, because nobody recorded when things started.

What are we building because of it?

We are building the record so that it is a by-product of teaching rather than a separate job done afterwards, on the school's own hardware, so that a child's history never leaves the building to be useful. The design goal is that assembling one child's case takes a minute rather than an afternoon, because that is the difference between a needs-led duty being real and being aspirational.

It is in development. We have not measured it against a real caseload in a real school yet, and until we have, that is the honest status. When there are numbers we will publish them, including the ones that do not flatter us.