England's ADHD Review Found a Risk of Overdiagnosis, Not Overdiagnosis
Headlines split on England's ADHD and autism review because the report says two things at once. Here is which sentences are findings, which are recommendations and which the government has promised.
On Friday, Oct. 9, two headlines about the same 599-page document pointed in opposite directions. Most said ADHD and autism in England are "at risk of over-diagnosis." A trade outlet covering children's services said in its headline that the review had found no widespread overdiagnosis. Both can point to a sentence in the final report.
The first sentence: "On the basis of evidence available to us we cannot conclude that overdiagnosis is occurring at scale, particularly given the gaps in population prevalence data and data reporting." The second, from the same paragraph of the executive summary: "However, if present trends continue we are clearly entering a phase in which the balance of risks is changing, from previous merited concerns with under recognition and underdiagnosis, towards an increased risk of misdiagnosis and overdiagnosis."
That is the whole disagreement. The review found no overdiagnosis at scale. It also warned that the odds are tilting that way. A headline that says "risk" is accurate if you read it as a rising risk and wrong if you read it as a verdict, and a headline that says "no overdiagnosis" is accurate about the verdict and silent about the warning.
The review was chaired by Prof Peter Fonagy of University College London, with Prof Sir Simon Wessely and Prof Gillian Baird as vice-chairs. The BBC reported that then-health secretary Wes Streeting ordered it after saying young people were being "over-diagnosed" and "written off." On mental health generally, the report finds the rise in distress is real: adults aged 16 to 64 meeting the criteria for a common mental disorder rose from 15.5% in 1993 to 22.6% in 2024. On ADHD and autism, its caution partly echoes him.
The numbers behind the warning, and what they cannot show
The report is careful about what "overdiagnosis" means: labeling a person with a condition they do not have, or one that "would not have caused them harm or significant difficulties if left undiscovered." Proving that needs a trustworthy yardstick for how many people truly have each condition. England does not have a recent one. The best autism estimate, about 1.2% of young people, comes from a 2017 survey. For ADHD, the same survey put prevalence at 3% of 5-to-15-year-olds in 2017, and the review's projection to today, about 3.8%, is "at best a rough estimate." No equivalent survey of adults has ever been run.
Set the recorded diagnoses beside those estimates and the pattern behind the headlines appears. The comparison is ours, not the report's, and the ages and bases do not match exactly.
| Group | Recorded in 2025 | Report's estimate | Recorded divided by estimate |
|---|---|---|---|
| Autism, ages 6 to 11 | 4.58% (about 1 in 22) | 1.2% (2017) | about 3.8 |
| Autism, ages 12 to 17 | 4.06% (about 1 in 25) | 1.2% (2017) | about 3.4 |
| ADHD, ages 12 to 17 | 3.88% (about 1 in 26) | about 3.8% (rough projection) | about 1.0 |
| ADHD, ages 6 to 11 | 2.70% (about 1 in 37) | about 3.8% (rough projection) | about 0.7 |
Autism is where the gap is wide: recorded diagnoses run at roughly three to four times the 2017 estimate. Teenage ADHD sits about level with a rough projection. ADHD in primary-school ages is still below it (about 0.9 against the measured 2017 figure of 3%). So the warning rests mostly on autism and on older children with ADHD, not on ADHD across the board.
The caveats cut both ways. Recorded figures count anyone "ever recorded" in a primary-care record, so they accumulate, yet they leave out some private assessments. The 2017 survey was "not a dedicated comprehensive autism ascertainment study." And estimates "should not be treated as fixed biological ceilings." A recorded rate above an old estimate is a reason to look harder, not proof that anyone was mislabeled.
What the report is "confident" about is narrower: rapid growth in demand, plus system failures in responding, is "creating the conditions for increasing risk of misdiagnoses." Firm warnings on thin data are common; the WHO's child obesity advice was strong on rules and thin on evidence. The risky assessments are brief, not face to face, fragmented, made without records, or focused on one condition. And it refuses the either-or: "Overdiagnosis, misdiagnosis and underdiagnosis are all system failures," and all three "can occur within the same system at the same time."
The wait is the finding, not the risk
If you want the part of the report that rests on measured harm rather than a projection, look at the queues. Monthly referrals for suspected ADHD more than tripled, from around 5,000 in 2019 to around 17,000 in 2024. By the end of 2025, open referrals for ADHD assessment exceeded 560,000. Separately, more than 254,000 people had an open autism referral by December 2025, with over 90% waiting longer than 13 weeks. Adding the two gives 814,000, our sum of two report figures; the report does not say whether anyone appears on both lists, so treat it as an upper-end sketch.
Waiting is not neutral. In a Care Quality Commission survey of community mental health service users, among people who waited under two weeks, around 1 in 4 said their difficulties worsened; after more than six months it was more than 7 in 10, though the report flags the evidence "with appropriate caution given the survey response rate of 19%." And 40% of autism clinicians surveyed reported no ongoing post-diagnostic support.
This is why the review calls the argument over "too much diagnosis or too little" a trap that "risks reinforcing the centrality of diagnosis rather than need." Fonagy put the remedy this way in the department's release:
"Diagnosis remains essential and must be protected - what has to change is that it is no longer the only door through which help can be reached."
Prof Peter Fonagy, chair of the review
The report also rejects the idea that families game the system: "seeking a diagnosis is not evidence of inappropriate demand; it is a rational response to the way support is organised." And "they are not the problem."
What a recommendation does not yet change
A child on a list, or a parent weighing a referral. Diagnosis has become "one of the principal routes" to school support, "not, however, what the legal framework intended." Under the Equality Act 2010, adjustments rest on functional need, and "there is no legal requirement for a diagnosis in order to access an EHCP" (an education, health and care plan, the legal document setting out a child's support). That is the law as the report describes it now, not a reform.
For people already waiting, the review recommends a clinically led review of the lists to offer "evidence-based support while waiting." It adds that "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, not simply signposted." That is a recommendation, and the department's release does not mention it. The report does not itself remove anyone from a list.
An adult on a list, or diagnosed privately. Independent providers now carry out more than half of NHS-funded ADHD assessments and more than a third of autism ones, and the report says it is not "a criticism of the people who turn to them after long NHS waits." Its concern is oversight, and continuity: a diagnosis obtained outside the NHS "does not necessarily provide continuity of care," leaving many people to return to a GP for prescribing without shared-care arrangements. It recommends that all assessments become a Care Quality Commission regulated activity. The government says it is "working towards" that, and Cooper's own words in the release are firmer: the department will "make all diagnostic assessments for ADHD and autism CQC-regulated." Neither comes with a date.
A person taking ADHD medication. The recommendation is to continue medication "where benefit clearly outweighs harm" and to review and safely reduce it where it does not, particularly "long-term polypharmacy in adults." The report says medication is "an important and often highly effective intervention for many people with ADHD" and that evidence "does not support the view that increased prescribing has produced widespread medication-related harm at the population level."
A Norwegian study of young people "on the margins" found few measurable long-term benefits of medication, but the report notes it was observational, covers 2009 to 2011, and that the authors "do not claim this is evidence against diagnosing or treating ADHD more generally." Questions about an individual's treatment belong with their prescriber.
Someone claiming, or considering claiming, a disability benefit. The Telegraph ran the headline "ADHD surge fuelled by benefits, report warns," citing government figures it said show claimants with autism as the main disabling condition rising from 79,395 in January 2019 to 268,965 in July 2026. Those counts are not in the report or its technical annex. What the report says is that Personal Independence Payment depends "on functional impairment rather than diagnosis," that welfare design "lies outside the remit of this Review," and that diagnosis "should not automatically be interpreted as evidence of fixed or permanent incapacity." It recommends no change to any benefit rule.
The sentence most likely to be lifted out of context warns that needs scores can "encourage people to maximise scores." That concerns how a new shared framework could go wrong. It is not a finding about why people seek diagnoses.
Recommended, promised, dated: three different things
The report asks for a cross-government mental health strategy "this year," with immediate actions over 12 to 24 months and a 10-year reform plan. The department's release says Health and Social Care Secretary Yvette Cooper will set out the response in a new Mental Health Strategy "in the coming months," later this year. From Oct. 9 to Dec. 31 is 83 days. The Telegraph reported that the recommendations have not been costed; the report itself says they "will require investment."
| Item | Review recommends | Government's reply | Date given |
|---|---|---|---|
| Clinically led waiting-list review; no one removed without help | Yes | Not mentioned in the release | None |
| All ADHD and autism assessments CQC-regulated | Yes | "Working towards"; Cooper: "make all diagnostic assessments for ADHD and autism CQC-regulated" | None |
| Ban on advertising NHS-funded assessments to the public | Yes | Secretary of State to write to the Advertising Standards Authority; no ban pledged, per the Guardian | None |
| National prices for NHS-funded services | Not a headline recommendation | NHS England will consult | Payments from April 2027 |
| Cross-government mental health strategy | This year | "Later this year" | End of 2026 |
| NICE guidance | Update guidelines | "Urgently consider" | None |
The commitments with clear edges are about money and paperwork. The government does say that, "where clinically appropriate," timely diagnosis and treatment will continue, and that people in distress will get broader help "without having to wait for a clinical diagnosis."
The Royal College of Paediatrics and Child Health welcomed the needs-led model but said it "must be accompanied by sustained investment in diagnostic and specialist services." Prof Dame Uta Frith told the Telegraph the review was "sitting on the fence" and its recommendations "utopian."
Wessely went further than the text in that paper's interview: if present trends continue, "the risk of overdiagnosis and over-medicalisation is very, very likely." Those are his words, not the report's.
For now, the only date in the government's reply that attaches directly to ADHD and autism assessment services is April 2027, and it attaches to a price list. The strategy due by the end of the year, 83 days out, is what will show whether the waiting-list review happens at all.
