DWP’s ADHD denials triple normal rate
DWP's own PIP data shows ADHD and autistic claimants failed at three times the average rate, even as most win on appeal.
Applicants who came to the Department for Work and Pensions with an ADHD diagnosis described anticipatory anxiety that started weeks before their assessment, palpitations, sleepless nights and a sense of being disbelieved before they had said a word. That is not a claimant’s account leaked to a newspaper. It is the DWP’s own commissioned research, published last October, into how people with anxiety and neurodivergent conditions experience claiming Personal Independence Payment. Now DWP’s own newer figures show why the dread was justified: this is what dwp adhd benefits decisions look like from the inside.
Analysis of the department’s latest PIP change-of-circumstances data, published on 13 July by welfare specialists Benefits and Work, shows ADHD and autistic claimants are failed at roughly three times the rate of everyone else on the system. Across the 50 most-reviewed conditions, 5.5% of reassessments end in a failed assessment. For ADHD, it is 16.3%. For autism, 17%.
The DWP ADHD benefits gap, in the department’s own numbers
The data covers people already receiving PIP who reported a change in their condition and were reassessed, 1,783 ADHD cases and 2,058 autism cases in the same clearance window. Cancers, strokes, heart disease and multiple sclerosis all saw above-average rates of award increases when reassessed. ADHD, autism, anxiety, learning disabilities and depression all sat below average. The pattern is consistent: conditions with visible, physical evidence are believed. Conditions that live in executive function, planning a journey, managing money, remembering medication, are not.
That is the mechanism, not a coincidence. PIP assessors are told to measure functional impact rather than diagnosis, which sounds reasonable until you ask what a 45-minute interview can actually see of someone whose impairment is inconsistency itself. A clinical letter confirming ADHD does not describe what happens on the days a claimant cannot make a phone call or leave the house. The assessment format is not built to capture it, and the department’s figures now show that.
The clearest evidence the department is getting this wrong first time is what happens when claimants fight back. On PIP appeals generally, roughly seven in ten are overturned in the claimant’s favour at tribunal. A department that gets it wrong seven times out of ten when challenged is not measuring function. It is guessing, and relying on the claimant not having the strength left to appeal.
What the caseload story leaves out
Kemi Badenoch and parts of the corporate press have turned the rising ADHD caseload, up from 71,528 claims in July 2024 to 100,207 by April 2026, into a story about a system too soft on young claimants. Mindy Ptolomey, a disability studies lecturer at the University of Leeds, has called that framing “disingenuous”, pointing out it is now “practically impossible for some young people in the UK right now to access an ADHD diagnosis” in the first place. The rise in claims does not make the claims illegitimate. The department’s own failure rate suggests the opposite problem: too many genuine claimants being refused, not too many false ones being waved through.
It is worth being precise about what the figures do and don’t show. ADHD claimants reassessed under change of circumstances were still more likely to have their award increased than decreased, 28.7% against 5.5%. The scandal is not that the system fails everyone. It is that, against people whose disability is harder to photograph, it fails three times as often as it should.
Disability Minister Stephen Timms is running a review of the PIP assessment, due to report in autumn 2026, examining whether the criteria capture fluctuating and hidden conditions at all. The department has the evidence already. It commissioned the research that described the anxiety its own process causes, and it holds the figures that show who that process gets wrong.
