Deeper Into Delusion: The Government Said Failure Was Normalised. DHCW Reported 94% Success.

For half past three in the afternoon of 30 July 2026, Digital Health and Care Wales called its Annual General Meeting — "where we'll showcase the report", its website announced — the report being its annual report for 2025-26. The report's headline, repeated across its pages: "we delivered 94% of our IMTP milestones." Its governance conclusion: "There were no significant control or governance issues identified during 2025/26."

Three weeks earlier, on 9 July, the Welsh Government had published its own account of the same organisation's same year: twenty-five itemised concerns, an independent review of DHCW's culture and leadership, and the second-most-serious intervention available to it against an NHS body. Its central finding, in the government's words:

"Repeated failure to deliver major national programmes to agreed milestones, with delivery slippage becoming normalised despite recovery actions and executive assurances."

Two official documents, published in the same month, describing the same twelve months of the same public body. One says failure is so routine it has become normal. The other says 94% success and no significant issues.

This article is about how the second document was made. Read in sequence, its sections trace a movement — each device carrying the official account one step further from the year the government described. A denominator grows; an intervention vanishes; a governance statement stops fifteen months short of the truth. The direction of travel is always the same: deeper into delusion.

Two documents, one organisation

Digital Health and Care Wales (DHCW) is the special health authority that took over the functions, programmes, staff and leadership of the NHS Wales Informatics Service in 2021. It runs the digital infrastructure of the Welsh NHS. In April 2026 the Welsh Government confirmed its escalation to Level 4 — Targeted Intervention, one step below Special Measures; the full framework documenting why was published on 9 July.

The annual report and accounts for 2025-26 — the escalation year — were approved by DHCW's board on 25 June 2026 and signed the same day by chief executive Helen Thomas, who ran the informatics service as its interim head before becoming DHCW's founding CEO. The Auditor General for Wales certified the accounts on 26 June. The report was published on DHCW's website in the week beginning 21 July, and billed for showcase at the AGM of 30 July.

That timeline matters. The board approved this account of the year eleven weeks after Level 4 was confirmed to the organisation, and the report reached the public after the government had published its twenty-five concerns. Nobody who signed it can have been unaware of the verdict it would sit beside.

How to manufacture 94%

The 94% is not invented. It is constructed — and the report itself, read closely, shows the construction.

Start with the denominator. The performance synopsis explains: "Our original Integrated Medium-Term Plan (IMTP) submitted to Welsh Government comprised 345 milestones. During the year, the plan expanded to reflect new and evolving national priorities… resulting in a revised plan of 381 milestones." Of those 381, DHCW reports 360 delivered. That yields two claims at once: 94% (360 of 381), and — because 360 is more than 345 — the assertion that DHCW "exceeded the original IMTP commitment." A plan that grows in-year lets you over-deliver against the old plan and score high against the new one, whatever happened to any individual promise.

Then the treatment of the milestones themselves. The governance statement supplies the mechanism in five words: activity was reprioritised and "milestones reprofiled where necessary." A milestone that slips and is re-dated still counts when it is eventually met. The Welsh Government's framework describes the same practice from the outside: slippage "becoming normalised despite recovery actions and executive assurances."

Then the status of the plan. Note 2.3 of the accounts records that, as a special health authority, DHCW's IMTP "did not require Ministerial approval." The 94% is scored against a target set which no one outside DHCW signed off.

And then — because annual reports are long and the caveats have to go somewhere — the report's own concessions. The performance synopsis opens: "Overall, our performance broadly met expectations but did not fully meet original delivery ambitions." The statutory well-being section, where few readers venture, goes further: outcomes "did not consistently translate into demonstrable outcomes or sustained improvements across all programmes, with some slippage and variability in delivery." Slippage — the government's word, in DHCW's report.

Most striking of all, the synopsis concedes the causal chain that the 94% is designed to obscure: diagnostic programmes including LIMS and RISP, and elements of the NHS Wales App, did not meet expectations, and "these challenges contributed to the organisation's escalation." The report knows why Level 4 happened. It says so, once, quietly — twelve pages after the 94%.

The vanishing intervention

Here is what a reader of the annual report learns about the most serious event in DHCW's history.

The phrase "Level Four" appears once in 231 pages — in the forward look, where escalation "presents us with further scope to strengthen and improve how we work across the system." Not a sanction; an opportunity.

The governance statement — the document in which an Accountable Officer is required to level with the public about control and risk — contains a section titled "Escalation and Intervention Arrangements". That section describes the March 2025 move to Level 3 (Enhanced Monitoring) and stops there. It was signed on 25 June 2026, eleven weeks after Level 4 was confirmed. The nearest the statement comes to the event itself is one oblique sentence: the decision to escalate further "indicates that, despite progress, the pace and consistency of improvement was not sufficient across all areas of delivery."

The independent review of DHCW's programme management, governance, culture and leadership — commissioned by the Welsh Government and running while the report was being finalised — is not mentioned. The Escalation Board is not mentioned. The twenty-five concerns are not mentioned. In the accounts, Note 29 — events after the reporting period, the note that exists to tell readers what happened between year-end and signature — discloses a single event: a change to pay rates for public appointees. The confirmation of Targeted Intervention on 8 April 2026 is not there.

How the board processed the escalation is recorded elsewhere in the report, in the list of its private development and briefing sessions: "Escalation Reflections" and "Escalation Outputs" — alongside a joint session with the Republic of Ireland on stakeholder engagement. The gravest intervention available short of Special Measures was handled as a reflection exercise, in sessions the public cannot see.

And the conclusion of the Annual Governance Statement, above the Accountable Officer's signature, in the year of Targeted Intervention and a twenty-one-million-pound VAT error: "There were no significant control or governance issues identified during 2025/26."

What the numbers say anyway

Accounts are harder to narrate than milestones, and DHCW's tell their own story.

The VAT furnace. For years DHCW reclaimed VAT it was not entitled to. In 2025-26 the reckoning arrived: roughly £21.8M repaid to HMRC, plus £852,447 of interest — the only entry in the year's losses and special payments register, recorded under the heading "Gross loss to the Exchequer", with a further £3.4M of interest provisioned for next year. The accounts add a detail that moves the loss from abstract to concrete: a £17.3M contingent amount that would have been "repayable to Welsh NHS organisations" had the VAT claim succeeded was extinguished when the money went back to HMRC. The ultimate losers of the failed scheme were the health boards. The furnace has a new exhibit.

The software that never arrives. DHCW's balance sheet carries £35.4M of intangible "assets under construction" — software paid for but not yet in use. In 2025-26, £8.6M of new development spend was added to that pile; £534K of it was brought into live service. Unfinished software is now 77% of DHCW's software assets by book value, and the pile has grown every year: £21.1M, £27.3M, £35.4M at the last three year-ends. This is what "delivery slippage becoming normalised" looks like in accounting form: the assets accumulate; they just never arrive.

The reward. Welsh Government cash funding to DHCW rose from £107.8M to £151.8M in the escalation year — a 41% increase, in the year the HMRC repayment fell due. Taxpayers' equity nearly doubled, from £30.9M to £58.4M. Whatever Targeted Intervention is, it is not a funding sanction.

The run-cost. The largest cost line after staff is "computer software licences and maintenance contracts": £52.4M, up £4.2M in a year — the recurring bill for keeping the existing estate alive. Set against it, the report's statutory consultancy line reads £221K; FOI disclosure puts DHCW's professional-services spending at £8.94M. Both numbers are technically defensible. That is the problem.

Capital versus rhetoric. The report promises that RISP "progresses towards a fully integrated radiology system by 2026, while LIMS2.0 progressed to the launch phase" — the same two programmes the government's framework names as exemplars of failure. The capital table shows what the promises were fed with: LIMS 2.0 received £3.0M; RISP, the fully-integrated-by-2026 programme, £631K.

The small print on the flagship claims

Each headline achievement in the report carries its own caveat, printed nearby in smaller type.

The NHS Wales App: "100% of GP practices in Wales are connected" — followed in the same sentence by "however, levels of active use and functionality vary across practices, with rollout continuing." Connection is a server-side fact; use is the claim that matters, and on the report's own figures just 30% of practices actively use digital prescribing, the app's flagship function.

GP records: after a supplier withdrew from the Welsh market in 2024, DHCW reports "156 out of 194 practices migrated to plan by March 2026." Stated the other way: two years after the withdrawal notice, 38 practices were still running on a departing supplier's system — and once migration completes, Wales will depend on a single supplier for every GP record in the country.

The unified-record mission: DHCW's own architecture census, five years in, counts "over 1,400 digital systems in use across NHS Wales… with patient information currently distributed across around 300 separate systems."

The National Data Resource: the report's risk section concedes that "without a legal gateway that is fully transparent for the use of identifiable Welsh resident information, our Programmes such as the NDR and Value in Health, will not be able to derive the full benefits" — an on-the-record admission that the legal foundation of DHCW's flagship data platform is not in place.

Digital identity, the gateway to the app: the ministerial-directions appendix reports uptake "currently running at circa 150 new patients" per week. At that rate, against a Welsh population of 3.1 million, the queue is measured in centuries.

Patient safety is reported in the same key. Thirteen clinical incidents were logged; three were escalated to the Welsh Government as Early Warning Notifications; one was subsequently escalated by the affected health board as a Nationally Reportable Incident. The report's assurance is precisely scoped: "No patient harm because of any act or omission by DHCW was identified in relation to any of these incidents." Note what that sentence does not say: that no harm occurred.

And running through all of it, a quieter problem: the numbers do not agree with each other. The report gives the Electronic Prescription Service as "over 7 million" prescriptions in one section and "more than one million items" in another, while carrying a linked headline of "fifteen million items" — three figures, none defined. It gives two different average sickness rates and three workforce counts. Its remuneration table contains a row that is arithmetically impossible: a director whose disclosed total (£80-85K) is lower than his disclosed salary (£110-115K), with no negative figure disclosed and nothing in the accompanying notes that explains it. This is the annual report of the organisation that runs the data of the Welsh NHS, certified without observation.

Staff confidence falls off a cliff

One set of numbers in the report cannot be reprofiled, because DHCW's own staff supplied them — and they fall off a cliff.

Both this annual report and last year's publish the same three staff-survey measures, in the same statutory section on whistleblowing arrangements. Like for like, on DHCW's own presentation:

These are not drifts; they are drops. Confidence that a concern would be addressed fell by more than twenty-five points in a single year. The survey's response rate rose from 62.3% to 68.5% — which the report itself welcomes as having "increased the strength of the evidence available to us." On the strengthened evidence, four in ten of DHCW's staff no longer believe the organisation would act on a concern.

In the same year, the number of formal Raising Concerns cases halved — from four to two. The report offers the two numbers in the same section without connecting them. Connected, they describe something specific: a workforce losing confidence in the channel and, in consequence, not using it. Stress and anxiety became the largest cause of sickness absence: 5,209 working days lost. The report's own culture section concedes that "morale and experience continue to be under pressure" — while elsewhere characterising bullying and harassment as "significantly low", a phrase that is an adjective, not a statistic.

The year's internal audit record follows the same grain. The one audit rated Limited assurance — on recruitment processes — was received by the Audit Committee in private session. A second Limited-assurance report, issued before year-end, is deferred to next year's reporting and is not named anywhere in this report. The Head of Internal Audit's published opinion states that "no reports were allocated a 'limited' or 'no assurance' opinion" — a sentence made true by careful scoping of which year's reports count. The advisory audit of staff culture and wellbeing carries no assurance rating at all, and its findings are unpublished.

Around this, the leadership departed. Chair Simon Jones left on 30 September 2025, having attended three of the year's board meetings; the chair's seat was still vacant at year-end. The Director of Finance was seconded out to Swansea Bay University Health Board on 9 March 2026 — three weeks before year-end, mid-escalation; her interim successor's acting-up pay escaped disclosure because details were "confirmed after the deadline to submit pension figures." Independent member Rowan Gardner left on 31 March 2026, eight days before Level 4 was confirmed. The accounts were approved by a board with no substantive chair, an acting vice-chair, and an interim finance director.

The chief executive's position is recorded in the remuneration tables. Helen Thomas's salary band rose about 3% to £170-175K — no bonus, no benefits in kind; her reported total of £275-280K adds a £105-107.5K accounting valuation of pension growth, not cash paid. The figure that best captures the era is the capital one: her pension pot's transfer value stands at £1.804M, up £170K in the year, £142K of it employer-funded — and up from £1.052M over the four years in which DHCW travelled from routine oversight to Targeted Intervention. Executive pay at DHCW is not performance-related; the report says so itself. In this year, that worked in only one direction.

Who checked this?

On 26 June 2026, the Auditor General for Wales certified DHCW's accounts with a clean opinion and a single sentence: "I have no observations to make on these financial statements." Thirteen days later, the Welsh Government published twenty-five concerns about the same organisation.

The two are formally compatible. An audit certifies that the financial statements are true and fair and that the narrative report is consistent with them — it does not verify milestone delivery, and the 94% sits in the unaudited part of the report. But that is precisely the point. The annual report is the one accountability document that every public body must publish and any citizen can read — and every formal check it passed through measured something other than whether its story was true. The board approved it. The Accountable Officer signed it, and signed the representation that all disclosable post-year-end events had been disclosed. The Auditor General had no observations. And the result is a public record in which the year of Targeted Intervention appears as 94% success with no significant issues.

The Welsh Government's own accountability conditions for DHCW's 2025-28 plan, quoted in the report, included one that reads differently now: "Meeting the de-escalation criteria relevant to your escalation status." The outcome was escalation. The framework that recorded it lists, among its twenty-five concerns, "increased FOI requests and public commentary on digital delivery." The annual report is the organisation's answer to that commentary — and it is an answer addressed to the future, because the past has already been adjudicated, in public, by the government, in its own words.

At the AGM DHCW called for 30 July, the report it promised to showcase carries 94% on its opening pages. Two Julys, two verdicts. Only one of them was independently reviewed. It was not this one.

All quotations are from the DHCW Annual Report and Accounts 2025-26 (published in the week beginning 21 July 2026) and its Accountability and Audit Report, or from the Welsh Government's "Digital Health and Care Wales oversight and escalation framework: May 2026" (published 9 July 2026). Figures are cited from the reports' named sections and notes.

CareNHS invites responses from Digital Health and Care Wales and the Welsh Government to the matters in this article. If received, we will publish them in full.