The Safety Case DHCW Made Against Itself

On 8 April 2026, Welsh Government escalated Digital Health and Care Wales (DHCW) — the special health authority that replaced the nationally criticised NHS Wales Informatics Service (NWIS) in April 2021, under much of the same leadership — to Level 4, Targeted Intervention, citing delivery, accountability and leadership. The CEO who has led the organisation since its founding, Helen Thomas, progressed into the role from NWIS, where she was Director of Information and then interim Director and Chief Executive.

Most of the campaign's record to date has been about money and governance: programmes late, benefits restated, value for money unproven. This article is about something narrower and harder to argue with. It is about patient safety — and the point of it is that we are not the ones making the argument. DHCW's own board and committee minutes make it. The Auditor General for Wales makes it. We are quoting them.

The doom loop

The shape of it is a vicious cycle, and it runs in both directions:

A deficit-constrained system starves the digital layer — withheld funding, paused programmes, "vacancy savings" — so clinical systems stay paper-bound, obsolete, fragmented and outage-prone. That, in turn, creates avoidable cost, clinical bottlenecks and safety risk — which feeds back into the deficit and the waiting lists.

The digital layer is both a casualty of the squeeze and a contributor to the harm. You cannot dismiss it as "just the financial climate," because the record shows the failing infrastructure generating risk of its own. What follows is that record, thread by thread.

The scoreboard

Every entry on the right comes from a public meeting transcript or an Audit Wales report. The detail behind each row follows.

Critical care: a system paused because it was not safe

The clearest example of the digital build itself being judged unsafe is WICIS, the Welsh Intensive Care Information System. It failed its planned go-live at Aneurin Bevan University Health Board in late 2023. When DHCW returned to it, the problem was not merely schedule. As the Portfolio Delivery Committee was told:

"initial configuration testing revealed some significant patient safety concerns, and this led to a pause, and Welsh Government commissioned a review which outlined that … the system couldn't be implemented [as configured]." — DHCW Portfolio Delivery Committee, 30 April 2026

Welsh Government commissioned an independent patient-safety review specifically to decide whether the configured system could be deployed safely across Welsh critical-care units. It concluded that it could not — not in its then-current state. The programme sat on hold for roughly eighteen months, lacked a programme chair for part of that time, and was only restarted after a ministerial instruction in February 2026. An intensive-care unit is the last place in a hospital where the information system should be the weak link.

Cancer: a system run "well beyond end of life"

CANISC — the Cancer Network Information System Cymru — is the legacy that DHCW inherited and kept alive past the point of safety. Its FoxPro front-end has been unsupported since 2015; the Microsoft 2012 operating system it depended on reached end of life in October 2023. A DHCW director put it plainly to the Portfolio Delivery Committee:

"we have been dealing for many, many years with … the risks and then the consequences of the [CANISC] platform that we maintained well beyond end of life, to such a degree that the code base on which it was built was [barely understood]." — DHCW Portfolio Delivery Committee, 14 May 2024

This is not an abstract IT-debt problem. The Auditor General's review of cancer services tied the fragmentation of these systems directly to patients:

"the digital systems are still fragmented and they are creating potential risks for patient safety where people might disappear through the gaps between different systems." — Audit Wales, cancer services review

"Disappear through the gaps" is the regulator's phrase, not ours. It describes a cancer pathway in which the information does not reliably follow the patient.

Patient identity: 1,400 records, one clause

On 30 March 2023, DHCW's enterprise Master Patient Index — eMPI, the system that keeps each patient matched to the right record — suffered a data-integrity incident. At the 25 May 2023 board meeting it was disclosed in these terms:

"There was a reported incident [at] two o'clock on Thursday the 30th of March, and that lasted until 10:30 the following day. What happened was an overwrite of some of the elements of the demographic, particularly probabilistic matching of records … it affected just over 1,400 individual electronic patient records." — DHCW board meeting, 25 May 2023

The same board record establishes that the incident triggered harm reviews across all health boards, prompted an emergency executive meeting, and that around 200 records were still under outstanding harm review at the date of that May board. These are documented facts from DHCW's own minutes.

What did the published board minutes say about all of this? A single clause — no patient count, no harm-review detail. The full account exists in the meeting recording; the public record was reduced to a line. (This is the same sanitisation pattern documented in How DHCW Falsifies the Public Record.)

A patient-identity system whose entire purpose is to stop the wrong record reaching the wrong patient corrupted 1,400 records — and the body responsible told the public almost nothing.

Fragmentation, by design

DHCW's own digital director told the Portfolio Delivery Committee that multiplicity of systems is, in itself, a patient-safety problem:

"if we end up with … seven different systems, then you are building … the potential for patient safety risk into the system." — DHCW Portfolio Delivery Committee, 7 November 2024

That is not a warning about a hypothetical. It is happening in medicines. In February 2026, a DHCW director confirmed that NHS Wales is "making a step back" from a previously agreed single All-Wales digital drug chart — because different electronic prescribing (EPMA) suppliers across health boards will render drug charts differently:

"as everybody moves digitally, we will not have a single All-Wales digital drug chart. There [will] be different … drug chart [displays]." — DHCW Portfolio Delivery Committee, 5 February 2026

A clinician moving between Welsh hospitals will read a different drug chart in each. The body whose job is national standardisation is documenting its retreat from it.

Eye care: the harm the risk register hid

In November 2024, a board member warned that rolling all of DHCW's programme funding risks up into a single corporate risk did something dangerous: it concealed the human consequences. The example given was sight:

"the decision to roll up all of the funding risks into one risk … doesn't give full visibility to the consequences of the risk … the ophthalmology programme and the fact [that people could lose their sight]." — DHCW Portfolio Delivery Committee, 7 November 2024

The eye-care programme was subsequently paused and wound down as a national programme in 2025 — one of the failures catalogued in Nine Programmes, Zero Results. The warning was on the record before the decision was taken.

Mental health: records that do not follow the patient

The absence of a shared record is its own harm. DHCW's discovery work into mental-health services found patients being forced to repeat their history to clinician after clinician, because nothing followed them:

"people being … pushed into having to relive their … situation again and again … because there is no record … that has followed them on their journey." — DHCW Portfolio Delivery Committee, 7 November 2024

Mental-health services were described in DHCW's own baseline mapping in 2025 as "still very paper-based."

Resilience: "a never event," three years running

A patient-safety case is also a resilience case. DHCW's record shows a pattern of national outages with clinical exposure — several of them softened or omitted in the published minutes:

• July 2024 — a false fire alarm at Data Centre 1 triggered a cooling failover failure that took dozens of services down for around six hours.
• 11 June 2025 — a near-identical recurrence. The CEO told the board it "should really be a never event in terms of the level of data centres that we commission." The term — the strongest in clinical-governance vocabulary — did not make the published minutes.
• 25 March 2026 — a Public Sector Broadband Aggregator (PSBA) outage took internet and the health-and-social-care network down for all NHS Wales organisations from around 5pm until 1:20am, disrupting Office 365, electronic prescribing (EPMA), and radiology (RISP).
• A further NHS Wales network outage occurred on the morning of the 29 January 2026 Public Accountability Meeting itself.

These are the systems clinicians fall back to when everything else fails. When they fail too, the fallback is paper and phone calls — which is exactly where "missed handoffs" live. (On the wider pattern, see Celebrating the 5-Year Crash Against Reality.)

Warned, then it happened

The throughline is that the harm was predicted on the record before it materialised — which is what turns a list of incidents into an accountability case:

The counterfactual that proves it

The strongest evidence that the digital layer matters for safety is what happens when it works. DHCW's Executive Medical Director told the board:

"electronic test requesting for radiology and pathology … shortens waiting time straight to test on cancer referral pathways, [it] takes at least 10 to 14 days off that pathway." — DHCW board meeting, 29 September 2025

Ten to fourteen days off a cancer pathway, from one working digital service. That is the measure of what the paused, obsolete and fragmented systems above are costing — in the same currency, on the same pathways, in the same organisation.

The money, said carefully

It is tempting to convert all of this into a share of NHS Wales's deficit. We will not, because no source supports a figure, and the campaign's credibility depends on not inventing one. The honest version is opportunity cost: across these programmes DHCW absorbed very large sums while the record shows withheld funding letters, one-year funding runways, and paused and descoped programmes — and, on the Cabinet Secretary's own assessment, no demonstrated return on investment. The money did not buy the efficiency it promised, and it did not buy the safety either. The digital layer did not cause the deficit; it failed to relieve it, and in the ways set out above it added risk of its own.

What the record asks

None of the quotations above came from a leak, an insider, or a tribunal. Every one was spoken in a public meeting or written in a public audit, by the people responsible for the systems. The patient-safety case against NHS Wales's digital layer is not an external accusation. It is an internal admission — repeated, dated, and minuted — that has simply never been read in one place.

That is the question for the Welsh Government's Level 4 intervention: not whether DHCW's programmes are late, but whether the body charged with the safe digital backbone of NHS Wales has been allowed to run cancer, critical-care and patient-identity systems past the point its own board called unsafe.

This is the patient-safety dimension of a wider pattern. For the overview, see The Trajectory of Failure; for the delivery record programme by programme, Nine Programmes, Zero Results.

CareNHS invites responses from DHCW and from anyone named or referenced in this article. If a response is received, we will publish it in full.

Correspondence: Rose Davies, CareNHS.