Evidence and statistics

Is autism increasing? What UK data can—and cannot—tell us

Why recorded autism numbers rise, what different UK datasets measure, and what they cannot establish about underlying prevalence.

Written by Lee Thomas Smith MSc. Factual review by ZenEmu editorial team. How ZenEmu writes and reviews guides.

Autism is in the news again, and you may be wondering what to make of reports that it is increasing. Are more people autistic, or are we getting better at recognising people who were previously missed? UK research shows a clear rise in recorded diagnoses, but that is not the same as showing an equivalent rise in how common autism is in the population.[9] Greater awareness, changes in diagnostic criteria and the recognition of previously undiagnosed adults may all help explain why the figures have changed.[9][23][30]

So what can the numbers actually tell us? This guide looks at what different UK datasets count, where the gaps are, and why diagnoses, assessment waiting lists and estimates of the autistic population answer different questions. The aim is to help you make sense of the headlines, with a clear distinction between what the evidence shows and what remains uncertain.

For: Anyone trying to make sense of claims that autism is “increasing”.

Last reviewed: 8 October 2026.

In brief

English adult household surveys found around 1 in 100 adults, with an estimate of 1.0% in 2007.[11] The corresponding 2014 estimate was 0.7%; the two estimates were not significantly different.[3][25]

Meanwhile, UK primary-care records show sharply rising recorded new diagnoses, especially among adults and women.[9]

Those statements can both be true. They answer different questions.

Key numbers, and what each counts

The rows are deliberately not added together. They describe different groups and different systems.

The 787% figure is about recorded new diagnoses

Russell and colleagues examined diagnoses newly recorded in UK primary-care records between 1998 and 2018. They reported a 787% increase in recorded incidence across the period.[9]

CPRD means the Clinical Practice Research Datalink, a database of primary-care records from participating practices.[9]

The study used an index for the all-age trend: it set the 1998 incidence to 100 and showed the relative change in later years.[9] The paper does not give all-age rates, so we have not calculated one.[9]

The authors did give an adult example: one adult per 100,000 was newly diagnosed in 1998, compared with 20 per 100,000 in 2018.[9]

The authors said that an increase in underlying prevalence was possible, but increased reporting and application of diagnosis was more likely. They noted especially large increases among adults, women and people recorded with diagnoses not associated with intellectual disability.[9]

A referral, a school record and a diagnosis are not the same thing

NHS England's autism statistics count referral-pathway activity for suspected autism, including open referrals and the time from referral to a first relevant care contact.[19] They do not count completed diagnoses or estimate how many autistic people live in England.[19]

In June 2026, the series recorded 294,792 people with an open suspected-autism referral in England.[19] The series also has reporting and provider-coverage changes, which matter when interpreting movement over time.[19]

England's school census recorded 289,646 pupils with autistic spectrum disorder as a primary special educational need in January 2026.[20] The Department for Education says that primary-need information does not show the prevalence of a medical condition or diagnosis in the population: a pupil may be autistic while a different need is recorded as primary.[20]

Northern Ireland's annual school-age figure is different again: 18,472 children, or 6.2%, were recorded as diagnosed with autism in 2025/26.[27] It is a school-age administrative record, so it cannot answer an adult or whole-population question.[27]

If you are looking for the latest figures rather than an explanation of what they mean, see UK autism data.

Why Northern Ireland's 6.2% is not an answer to the adult-survey question

The Northern Ireland figure is a record of school-age children with a recorded diagnosis. The APMS figures are assessment-defined estimates for adults in English private households.[3][25][27]

They describe different generations, different countries and different routes into the data. The adult APMS rounds took place in 2007 and 2014; the Northern Ireland figure is for the 2025/26 school year.[3][25][27]

The adult survey used assessment rather than a register of diagnoses.[3][25] More recent English primary-care research also found evidence of substantial adult underdiagnosis, particularly in older age groups.[30]

That makes it unsurprising that a current school-age diagnosis record can sit above an older adult assessment estimate. It does not show that autism is more common among children in Northern Ireland than among adults in England, that prevalence has increased from about 1% to 6.2%, or that one number is the correct replacement for the other.

What the England population studies can tell us

The Adult Psychiatric Morbidity Survey used a two-stage approach in private households in England, with a diagnostic assessment in a selected phase-two group.[3][25] Its estimate was 1.0% in 2007 and 0.7% in 2014.[25]

The corresponding 95% confidence intervals were 0.5% to 2.0% in 2007 and 0.3% to 1.3% in 2014.[25] The report says there was no statistically significant difference between the two years.[3]

Combining the household samples produced an estimate of 0.8% (95% CI 0.5 to 1.3), based on 31 identified participants.[3] That pooled result is not a 2014 result.

Brugha and colleagues produced a different 1.1% estimate across adult ages and ability levels.[12] It combined the APMS material with a case-register survey of 290 adults with intellectual disability; that case-register fieldwork was in 2010.[12]

The 2014 APMS landing page records that a May 2026 investigation found no data issue in the autism chapter, but led to a methodology clarification and a new version of Appendix B.[28]

The 2023/24 APMS autism chapter is a separate release. NHS England withdrew it after identifying a processing error that affected the quoted values.[17]

Why there is no honest single UK total

England, Wales, Scotland and Northern Ireland do not publish one shared, comparable series for autism diagnoses, referrals or population prevalence.[19][20][21][22][26][27]

Wales' available neurodevelopmental waiting-time information combines ADHD and autism. The Welsh Government says it is management information rather than official statistics and should not be treated as a reliable measure for Wales as a whole or for individual health boards.[21]

Scotland's pupil census collects reasons why pupils require additional support to access education, not the number diagnosed with specific conditions.[26]

Northern Ireland's quarterly referral and diagnosis publication is paused while the Encompass patient-record rollout is quality-assured, and the available series has missing trust data that limits comparison across quarters.[22] Its annual school-age diagnosis series is a different measure.[27]

A single UK line chart would conceal those differences.

Why recorded numbers can change

DSM-5 brought several previously separate diagnoses, including Asperger's disorder, into one autism-spectrum diagnosis; ICD-11 has its own autism-spectrum category and qualifiers.[18][23]

The UK primary-care study found the largest increases among adults and women.[9]

Neither point establishes one cause of the UK trend. Diagnosis, recognition, service access, coding and data collection can overlap in the same recorded series.

For classification context, see What happened to Asperger's syndrome?.

A brief note on claims about causes

The NHS says research has shown no link between the MMR vaccine and autism.[16] The MHRA says there is no evidence that taking paracetamol during pregnancy causes autism in children, and that recent studies do not show a causal association.[15]

Those are separate questions from the interpretation of diagnosis or referral trends.

A short checklist for a careful headline

Before repeating a number, ask:

1. What does it count?
2. Where and when was it measured?
3. Who is included, and who is left out?
4. What changed in the series itself?
5. What conclusion does the number not establish?

Related guides

• UK autism data for current figures, dates and sources.
• Why has the number of autistic adults receiving PIP increased? for the narrower question of benefit caseloads and condition coding.
• What happened to Asperger's syndrome? for classification history.
• Why does autism seem to run in my family? for a separate discussion of family pattern and population genetics.