Fonagy review proposes national standards for ADHD and autism assessment

The Fonagy review (the independent review into mental health conditions, ADHD and autism, October 2026) wants every ADHD and autism assessment held to the same national standards. The standards have not been written yet. But the review says where misdiagnosis is most likely, and each of those risks is something an assessment record can show was handled. So the practical question for a clinic is whether your records show how you reached each conclusion, not only what you concluded.

What this means for you

• Every ADHD and autism diagnostic assessment you do is likely to become CQC-regulated, with national standards to meet.
• The review links misdiagnosis to assessments that are brief, remote, fragmented, made without informants or records, or focused on one condition. Your records should answer each of those.
• That comes down to seven areas of the record. Most are already in NICE guidance, AQAS and CAAQAS, so you can check ten recent assessments against them now.

1. National assessment standards are proposed, but none are written yet#

Recommendation 5 of the final report, chaired by Professor Peter Fonagy and published by DHSC on 9 October 2026, says all ADHD and autism diagnostic assessments should be CQC-regulated, with the same quality standards for NHS, independent, voluntary and digital providers (p444). CQC confirmed the same day that DHSC has asked it to regulate them once new regulations are in place.

The review says the standards should require:

• assessment of co-occurring conditions and functioning
• clear guidance on when remote and in-person assessment are the right choice
• the same training standards for all assessors

For autism, it also wants more consistent records of what was assessed, the conclusions reached and the support recommended (p179). Its clinician roundtables and provider survey found variation across NHS and independent pathways in assessment length, multidisciplinary input, formulation, attention to co-occurring conditions and how impairment is judged, which it says supports common standards for all providers rather than assumptions about either sector (p165).

The government says it is working towards CQC regulation. No standards, timetable or consultation have been published, and its full response is due in a Mental Health Strategy later this year. My guess, and it is only a guess, is that the standards will build on NICE, AQAS and CAAQAS. Page numbers refer to the report PDF.

2. Misdiagnosis is likelier in five conditions, and your records can answer each#

The review says misdiagnosis is more likely where assessment "is brief, not face to face, fragmented, made without access to informants or records, and is focused on a single condition" (p14). It cannot conclude that overdiagnosis is happening at scale, but says that if present trends continue, the balance of risk is shifting towards misdiagnosis and overdiagnosis (p15).

A record can answer each of those five. Which other conditions were considered? Who was asked, and which records were read? Why was it remote or in person? How was the conclusion reached, and how long did it take? Who did each part, and what training do they have? The sections below take these one at a time, and the self-audit at the end puts them together.

3. Co-occurring conditions: show which you considered, and why you assessed for both#

By 2024, among 16 to 24-year-olds with a common mental disorder such as anxiety or depression, 57% also screened positive for ADHD and 61% for autistic traits (pp12-13). A positive screen is not a diagnosis, but few people arrive with a single question. What gets missed when an assessment looks at one condition:

• In one city, children with both autism and ADHD had the second condition recorded four to five years after the first, on average, which the review says may reflect diagnostic overshadowing (p299).
• In CPRD analyses of children diagnosed with autism, diagnosis was delayed by around three years following an earlier anxiety diagnosis, and five years following depression (p204).
• Children with both autism and ADHD respond to stimulants less often than those with ADHD alone: around 50% against 70% to 80%, though stimulants remain an effective and recommended treatment for many of them (p209).

So a template built around one condition is now a quality risk. The record should show which other conditions you considered and why each was ruled in or out.

Assessing for both needs its own reason. The review warns that current optional tariffs may encourage combined ADHD and autism assessments because they pay more, whether or not both are needed (p329). Where both are needed, Recommendation 5 wants them done separately in a planned order, with shared clinical oversight (p444). Record the clinical reason, in case a commissioner asks why both were billed.

4. Informants and records: show who you asked and what you read before the appointment#

An assessment "made without access to informants or records" is one of the review's five risks. The record should show who was asked (parents, partners, teachers), which questionnaires came back, and which earlier records were read.

Much of this is settled before the first appointment. Annex A, the review's illustrative framework for assessing need, supports gathering structured information before appointments, and calls putting the least experienced staff at the point of entry a false economy (p588). One trust put most of its shorter waits down to senior, experienced staff at the front door (p326). So treat intake as part of the assessment: collect the history and questionnaires early, and have an experienced clinician review them.

5. Remote or in person: record why, and know your diagnosis rate#

The review asks for guidance on when to assess remotely but does not write it. In our experience, adult ADHD interviews, developmental histories and informant questionnaires move to video most easily. Autism observation is harder. The review says autism assessment needs developmental history, direct observation, self-report and careful assessment of current functioning (p178), and NICE CG128 and CG142 both expect direct observation, so a remote-first autism pathway needs to show how it meets that. Record, for each patient, why the assessment was remote or in person and when you switched. More in remote ADHD and autism assessments.

Expect to be asked about outcomes too. Independent and Right to Choose providers completed more than one-third of NHS-funded autism assessments in 2025-26. Responding providers reported around 90% remote delivery, and diagnostic conversion rates from 66% to 93% across providers, with averages of 79% to 86% (p165). The review attributes these figures to the Independent Healthcare Providers Network (IHPN). The review gives no NHS figures to compare, does not link remote delivery to those rates, and does not say what a right diagnosis rate would be. Even so, I expect commissioners and CQC to ask for your diagnosis rate and case mix. Know both before they ask.

6. Formulation and functioning: show your reasoning, whatever the outcome#

The review wants formulation and clinical understanding to come before a diagnosis (p263). For autism, people should get a formulation and summary of needs whether or not autism is diagnosed, so someone told "not autism" still leaves with an explanation and a route to support (p179).

This matters for complaints too. Clinicians told the review that after a long wait, a "no" can be devastating for the person, and some people complain after not being diagnosed (p162, p181). A written account of their needs and where to get support gives them something to take away whatever the outcome.

Functioning needs the same care. Tools used in ADHD and autism services cover only 8% to 41% of the relevant areas in the WHO's functioning framework, the ICF Core Sets, though the review notes this measures coverage rather than accuracy (p394). Annex A, which the review offers as an illustrative framework rather than a recommendation (p583), names WHODAS 2.0 as one validated measure for adults, and says that where the person's account and the clinician's view differ, both should be recorded (p588).

7. Assessor training: keep a file for each assessor#

Recommendation 5 asks for the same training standards for all assessors (p444). Recommendation 6 asks for a nationally agreed competency framework setting out the knowledge and skills needed at each level of the system (p446), though it does not say whether that will cover diagnostic assessors. NICE NG87 already says ADHD should be diagnosed only by a specialist psychiatrist, paediatrician or other appropriately qualified healthcare professional with training and expertise in diagnosing ADHD. Keep a file for each assessor with their training on each tool (DIVA-5, ADOS-2 and ADI-R), supervision and case counts.

8. Check ten recent records against all seven areas#

The areas above, plus a post-diagnostic plan, make seven. Four are named for the proposed standards and three are recommended elsewhere in the review.

What each assessment record should showFour areas the review says the standards should cover, and three more it recommends elsewhere. For each, pick what your recent records show. Nothing is saved or sent.AreaEvery recordSometimesNot yet1Co-occurring conditionsWhich other conditions were considered and why each was ruled in or out. If ADHD and autism were both assessed, the clinical reason for each.Review, Rec 5 · NICE NG87, CG128, CG142 · AQAS, CAAQASEvery recordSometimesNot yet2Day-to-day functioningImpairment in named areas of life, with real examples. For ADHD, at least moderate impairment in two or more settings.Review, Rec 5 · NICE NG87 · AQAS, CAAQASEvery recordSometimesNot yet3More than one informantWho was asked, when they were chased and what came back, and what you tried when no one was available.Review, p14 and p181 · NICE NG87, CG142 · AQAS, CAAQASEvery recordSometimesNot yet4Formulation for every outcomeThe person's needs explained before criteria are applied, and a summary of needs even when the answer is not ADHD or not autism.Review, p179 and p263 · NICE CG128Every recordSometimesNot yet5Remote or in personWhy this patient was seen remotely or in person. For autism, how direct observation was done.Review, Rec 5 · NICE CG128, CG142 · AQAS, CAAQASEvery recordSometimesNot yet6Assessor trainingA file for each assessor with training on each tool, supervision and case counts.Review, Recs 5 and 6 · NICE NG87, CG142 · AQAS, CAAQASEvery recordSometimesNot yet7Post-diagnostic planA clear explanation, a profile of strengths and needs, co-occurring conditions, advice on adjustments and routes back into care.Review, pp165-166 · NICE CG128 · AQASEvery recordSometimesNot yet0of 7 in
every recordYour result0 of 7 markedAnswer all seven to see where your records fall short and what to fix first.Source and notes

For audit checklists, see AQAS for adults and CAAQAS for children. The post-diagnostic plan is covered in the post-diagnostic offer, and the rest of the review in our series overview and CQC regulation article.

9. Six steps you can take before the standards are written#

1. Take your ten most recent completed assessments and mark each area in the self-audit above.
2. Where an area only sometimes appears, make it a required field in your report template.
3. Write down when you assess remotely, when you switch a patient to face to face, and record the reason for each patient.
4. Work out your diagnosis rate and case mix, by condition and by who pays.
5. Only combine ADHD and autism assessments where there is a clinical reason for both, and record it.
6. Look at who reviews new referrals, and how much they know before the first appointment.

10. Where Cerebric fits#

Much of this evidence comes from admin. Cerebric collects the patient's history and questionnaires before the first appointment, and sends and chases informant questionnaires, so the record shows who was asked and when. It records the assessment and drafts the report in your template, with every sentence linked to its source. The clinician reviews the draft, writes the formulation, makes the diagnosis and signs. At clinics using Cerebric, report writing time fell by 75%.¹

11. Footnotes#

1. Measured at UK Right to Choose and private neurodevelopmental providers using Cerebric, 2025–26. ↩