“How did you go bankrupt?”
“Two ways. Gradually, then suddenly.”
Ernest Hemingway, The Sun Also Rises
I've spent 30 years in digital transformation. Government, defence, financial services, healthcare. I've watched organisations throw money at technology and seen it work, and I've seen it fail. What's happening in the NHS right now is something different. It's not failure. It's something worse. It's investment that actively makes the problem harder to solve.
It almost brings tears to my eyes.
The NHS's fundamental problem is flow.
Not data. Not apps. Not platforms. Flow. The flow of patients across organisational boundaries. The flow of clinical responsibility from one team to another. The flow of consent, of intent, of observations, of outcomes across the gaps between trusts, between primary and secondary care, between the NHS and the organisations it partners with.
And the NHS cannot see it. It is a system that is structurally unable to observe itself at the points that matter most.
This isn't a new observation. The research is overwhelming and consistent.
The Joint Commission has found that communication failures are the root cause of most sentinel events in healthcare, and that inadequate handoffs are a factor in around 80% of all serious adverse events [1]. That includes wrong-site surgery, medication errors, treatment delays, and patient deaths. Studies in teaching hospitals have documented up to 4,000 handoffs per day. Each one is a potential failure point.
The Institute of Medicine put it starkly: “it is in inadequate handoffs that safety often fails first.”
In the NHS specifically, the picture is devastating. England experiences an estimated 1.8 million undetected medication errors at care transitions every year [2], resulting in harm across approximately 31,600 patient episodes. Fifty-two percent occur at hospital admission: the precise moment a patient crosses an organisational boundary [3]. A 2025 study using Freedom of Information requests found that more than 10,000 digital health technologies currently deployed across the NHS lack documented assurance against mandatory clinical safety standards [4]. In a typical NHS trust, three out of four digital tools influencing patient care do not demonstrate compliance with minimum legal safety requirements.
The NHS reports 2.4 million patient safety incidents a year. Around 30% result in some level of harm. Approximately 12,000 lead to severe harm or death. Clinical negligence liabilities have reached £60 billion, quadrupling in two decades, with annual settlement costs of £3.6 billion in 2024/25 [5]. The National Audit Office confirmed that clinical negligence is now the second largest government liability after nuclear decommissioning [6]. The OECD estimates that safety failures could account for 13% of total healthcare spend: a potential cost of £25 billion a year in England alone [7].
And the system that is supposed to learn from these failures? The Public Accounts Committee concluded that the evidence points to “a system overwhelmed by safety recommendations that it cannot action.”
These aren't just statistics. They are the measurable consequences of a system that cannot observe its own flow.
The Darzi Review & The 10-Year Health Plan
Lord Darzi's 2024 independent investigation [8], the diagnosis that preceded the Ten Year Plan, confirmed that the NHS “remains in the foothills of digital transformation.” He found that many IT initiatives “have brought little benefit while adding to the workload of clinicians.” The NHS, he noted, struggles with data sharing, digital maturity is low across much of the service, and crucially, “many of the productivity problems in the NHS are caused by interaction between different parts of the system.”
That last finding is the one that matters most. The problem isn't inside any one organisation. It's between them. It's in the flow.
The Darzi review also found a £37 billion capital spending shortfall over the preceding decade. And yet in the 2016 NHS Digital Maturity assessment, only 56 out of 249 trusts could share care records digitally with non-primary care providers. The money that was spent went into local systems that deepened the silos rather than connecting them.
The government's Ten Year Health Plan, “Fit for the Future,” promises three transformational shifts: from hospital to community, from analogue to digital, and from treatment to prevention.
Every single one of these shifts depends on flow working across organisational boundaries. And none of them can be achieved without it.
But the plan is already in trouble. The pattern of failure is instructive.
The BMA held an extraordinary Special Representative Meeting in September 2025, the first in over a decade, and voted to formally oppose the plan [9]. Its council chair warned that “the whole nation knows the health service is in deep crisis” and cautioned that “in ten years' time, we won't be discussing another plan. We will be lamenting the demise of what was another Labour Health Secretary's hard-won creation.” A BMA survey of nearly 3,000 doctors found that 89% of GPs believe the plan will decrease their autonomy, 77% said it would reduce continuity of patient care, and over a quarter cited inadequate IT and estates as a major concern [10]. Since October 2025, GP practices across England have been working under protest against imposed contract changes.
The BMJ was blunt: neighbourhood health plans “look a lot like previous initiatives to integrate health and care services” and evidence suggests “politicians hoping for cost savings or reductions in hospital use will be disappointed.” The King's Fund noted that the three shifts “have had widespread support from those working in health policy” but “have proved stubbornly difficult to implement. Progress on all three has been glacial, and in some cases is going in the opposite direction” [11]. The NHS Confederation described the neighbourhood health concept as “a conceptual muddle.” The plan was published without an implementation plan. No new funding was announced beyond existing spending review commitments. And the first financial planning round for 2025/26 revealed a projected deficit of £6.6 billion across integrated care systems and trusts.
This is the pattern: ambitious top-down vision, insufficient engagement with the people who have to deliver it, no foundational infrastructure to make it work. The plan promises to “deliver neighbourhood health at pace.” But pace without infrastructure is just chaos distributed faster.
The shift to community means more care delivered in more places, by more organisations. A “neighbourhood NHS” doesn't mean care stays in the neighbourhood. It means the neighbourhood becomes another node in a patient's pathway, another boundary to cross, another handoff to govern. The plan promises that two-thirds of outpatient appointments will move to digital alternatives. That means clinical observations being made in apps, remote monitoring tools, and virtual consultations. Observations that need to flow back into the rest of the pathway, governed and traceable. It promises a single patient record accessible through the NHS App by 2028, while the system is simultaneously procuring dozens of apps, each with its own data store, each maintaining its own version of reality about the same patient.
The plan promises to shift to prevention. But prevention depends on longitudinal visibility across a patient's entire journey, across every organisation they touch. It's the ultimate flow problem.
Here's the uncomfortable truth: the Ten Year Plan's objectives are essentially unachievable unless the NHS solves the flow problem. Not just at neighbourhood level, not just within an ICB footprint, but across the entire system. If one app affects the patient, every other app in the pathway must know about it. That isn't a feature request. It's a prerequisite for everything the plan is trying to do.
Without flow infrastructure, the Ten Year Plan is a strategy built on foundations that don't exist. And the professionals who are being asked to deliver it already know.
Every few weeks, another digital health app is procured. Another platform. Another “solution.” Each one solves a real problem for a specific team, a specific pathway, a specific trust. And each one comes with its own data store, its own version of the patient, its own understanding of what's happening and who's responsible.
Nobody asks the flow question at procurement. Nobody asks: How does this system participate in the governed transfer of clinical responsibility? How does it make its observations visible to the rest of the pathway?
Nobody asks because there's no infrastructure to ask it against.
The Legacy IT Problem: What the NHS Can Learn from Banking
Imagine if every bank in the country maintained its own ledger with no interbank settlement protocol. No shared understanding of who holds what. No way to reconcile when money moves between institutions. We'd call that insane.
But we don't have to imagine. We're watching it happen right now, in one of Britain's most regulated, best-funded sectors.
The UK's legacy banks are haemorrhaging customers, trust, and operational credibility to the likes of Monzo, Revolut, and Starling. The reason isn't better apps. It's infrastructure coherence. And legacy banking doesn't have it.
The UK Treasury Committee published data in March 2025 showing that nine major banks accumulated 803 hours of unplanned IT outages between January 2023 and February 2025 [12]. That is 158 separate failures, totalling more than 33 days of downtime. Barclays led with 33 incidents. NatWest clocked the most downtime: 194 hours. These aren't minor glitches. In January 2025, Barclays' mainframe suffered what the bank called “severe degradation.” Fifty-six percent of online payments failed over a three-day period that coincided with both payday and the HMRC self-assessment deadline. The Business Continuity Institute reported one family left temporarily homeless when the system collapsed mid-house purchase. The bank expects to pay up to £12.5 million in compensation for outages across the two-year period. Barclays confirmed the root cause: a software problem in its UK mainframe operating system. The bank still employs hundreds of engineers working in COBOL, a language over sixty years old. Many of the experienced COBOL specialists have retired. The bank now recruits sixth-form graduates into what it internally calls “baby dinosaur” roles, trying to maintain code that underpins deposits, overdrafts, and debit cards for millions of customers.
This isn't a Barclays problem. It's a structural one. A September 2025 survey of 150 UK banking leaders by SaaScada found that banks are spending £3.3 billion annually [13]. That is 24% of their entire IT budgets, purely on maintaining legacy core systems. More than half described their core as a “bottomless pit” of wasted money and time. Nearly seven in ten said growing demand for digital services is straining infrastructure that simply cannot keep pace. McKinsey's research tells the same story globally: only 30% of banks that attempt digital transformation successfully implement their strategy. Traditional banks are 40% less productive than digital natives. Fintechs release new features every two to four weeks; legacy banks take four to six months. Developers at traditional banks spend less than a third of their time actually writing code. The rest goes to maintenance, infrastructure, and integration work.
Here's what makes it obscene: the big four UK banks posted record pre-tax profits of £45.9 billion in 2024. They're on track for 14% higher in 2025. Record profits. Record fragility. The money is there. The will to fix the foundations is not.
Meanwhile, the challenger banks are proving what coherent infrastructure actually looks like. Monzo now has over 12 million customers, roughly 22% of UK adults, with revenue of £1.23 billion and 40% year-on-year growth. Revolut has surpassed 52 million customers globally and posted over £1 billion in profit. Forty percent of UK adults now hold a digital-only bank account, up from 24% just two years ago. When Monzo's systems faced a potential failure in August 2024, it transitioned to its backup platform within minutes, with zero service disruption. Its CTO told City AM: “Our systems are designed to be resilient to faults in any component. If there is an issue with one part of the system, the rest of the infrastructure seamlessly picks up the workload. It's fully tolerant.”
Fully tolerant. While Barclays' sixty-year-old mainframe buckled on payday.
The challenger banks aren't winning because they have better apps. They're winning because they built on clean infrastructure from the start. Cloud-native, API-driven, with coherent data flows across the entire platform. They don't have to reconcile between siloed systems because they never created the silos in the first place.
The NHS has exactly the same problem. But on a vastly wider scale, with far more organisations involved, far more complex governance requirements, and consequences measured not in frozen payments but in patient harm and death. If legacy banks, with their relatively simple data models, single-sector regulatory frameworks, and £45.9 billion in annual profits to fund modernisation, are struggling with system incoherence, what hope does a health service with hundreds of autonomous organisations, dozens of data standards, and seven ungoverned flows across every organisational boundary have?
The banking sector at least has competitive pressure forcing the conversation. The NHS doesn't even have a shared understanding of what the problem is.
Here's what makes this genuinely tragic. The people commissioning these tools are acting in good faith. They see a gap in care delivery and they fill it with technology. That's rational behaviour at the local level.
But at the system level, every app with its own data store is another island of context. And the gaps between those islands, the ungoverned boundaries where patients move between organisations, are exactly where harm happens. Where referrals disappear. Where responsibility is ambiguous. Where observations go unseen.
And the complexity isn't linear. Five apps don't create five times the integration challenge. They create combinatorial complexity. Every new procurement multiplies the number of ungoverned handoffs across the system. The NHS is spending money to increase entropy.
Ask yourself one simple question: if one of these apps makes a clinical observation about a patient, a deteriorating heart rate, a missed medication, a mood change, how does the rest of the system know?
The answer is: it doesn't.
The observation exists inside that app. In that app's data store. On that app's terms. The GP doesn't know. The consultant doesn't know. The community mental health team doesn't know. As far as the rest of the care system is concerned, the observation never happened. It's a clinically significant event trapped inside a silo, invisible at the exact moment someone else in the pathway needs to act on it.
This is where patients get harmed. Not because nobody noticed the problem, but because the observation had nowhere to flow.
The Problem with the Palantir Federated Data Platform (FDP)
“But what about the Federated Data Platform?” I hear you say. “Isn't that solving this?”
The Palantir FDP is a £330 million investment. It is the single largest digital programme in the NHS. And it is an analytics and operational visibility layer. Each trust gets its own instance. They can see their own waiting lists better, manage bed occupancy, coordinate theatre scheduling. That's genuinely useful work.
But the FDP gives trusts better visibility within their own boundaries. Flow is inherently inter-organisational. You can give every trust a perfect window into its own operations and still have a system that is completely blind to itself as a whole.
The FDP doesn't govern what happens when a patient crosses an organisational boundary. It doesn't answer: who has clinical responsibility right now? What was consented to? What was the intended pathway? What happened at the last handoff? The “federated” in the name is almost ironic. The data largely stays within trust boundaries, and the fundamental question of how care flows between organisations remains completely untouched.
Some trusts have noticed. Greater Manchester ICB, responsible for 2.8 million people, has refused to adopt the platform, stating that their existing local analytics already exceed its capability [14]. Leeds Teaching Hospitals warned that adoption would mean losing functionality, not gaining it [15]. The Chief Data and Analytical Officer Network wrote in an open letter that many trusts “already have similar tools in use that presently exceed the capability and application of what the FDP is currently trying to develop” [16]. Even the trusts gaining value from FDP are gaining operational value, not governance value. Better analytics on your own data is not the same as the ability to observe and manage flow across the system.
The NHS has spent £330 million and still cannot see itself.
Other countries have already proven this is solvable. Which makes the UK's inaction all the more damning.
There's no line item in an ICB budget for “flow.” Nobody's KPI is “reduce the number of ungoverned responsibility transfers between organisations.” The procurement model rewards point solutions and penalises systemic thinking.
So we keep buying apps. And the patient keeps falling through the gaps between them.
No app will ever solve the NHS's problems.
I need to say that again, because the entire digital strategy of the health service is predicated on the opposite belief.
No app, no matter how well-designed, how well-funded, how brilliantly implemented, will fix what is broken. Because the problem isn't inside any app. It's in the space between all of them. Apps operate within organisational boundaries. The crisis lives between them. Expecting an app to create inter-organisational governance is like expecting a brilliant accounting package to invent the interbank settlement protocol. It's a category error, and we are spending billions on it.
What's missing is infrastructure. The connective tissue that governs how identity, consent, clinical responsibility, and outcomes flow across organisational boundaries. Not as an afterthought bolted on later. As the foundation that everything else is built on.
Healthcare has seven fundamental governance flows that need to work across every boundary: identity resolution, consent management, data provenance, clinical intent, alert and responsibility transfer, service routing, and outcome traceability. Right now, not one of them is consistently governed at the inter-organisational level.
Until we build that infrastructure layer, every new app is just another silo. Another island. Another way for a patient to become invisible at the exact moment they're most vulnerable.
And the stakes are not merely clinical. They are existential, for the NHS and potentially for something larger.
The NHS already consumes 40% of day-to-day government spending. Clinical negligence liabilities stand at £60 billion [5]. Annual settlement costs are £3.6 billion and forecast to reach £4.1 billion within five years [6]. The system is not managing. It is not coping. By any honest measure, it is failing. Slowly, expensively, and in full view of the public who fund it.
What connects the banking crisis and the NHS crisis is not technology. It is a poverty of leadership courage at the most senior levels of government.
The Treasury Committee can document 33 days of bank outages [12], watch Barclays' mainframe buckle on payday, see £45.9 billion in annual profits flowing to shareholders while customers are locked out of their accounts, and still not mandate the infrastructure modernisation that would fix it. The Department of Health can watch 1.8 million medication errors at care transitions [2], see clinical negligence liabilities quadruple to £60 billion [5], receive Darzi's diagnosis that the system “remains in the foothills of digital transformation” [8], and respond with a Ten Year Plan that contains no plan for the one thing that would make everything else work.
In both cases, the pattern is identical. The evidence is overwhelming. The diagnosis is clear. The technical solutions exist. And the leadership response is to manage the symptoms rather than fix the foundations, because fixing foundations is unglamorous, multi-year, cross-departmental work that doesn't fit a political cycle.
Denmark mandated healthcare data standards by law, with financial penalties. The result: comprehensive data sharing within a decade. Estonia built unified health records on secure digital identity infrastructure. Both achieved through regulatory enforcement, not voluntary adoption. The technology exists. What's absent, in banking and in healthcare, is the political will to mandate the hard, boring, foundational work that makes everything else possible.
We know the current trajectory is unsustainable. Everyone knows. And yet the proposed answer is always the same: more money, more apps, more reorganisation. The public is asked to tolerate ever greater spend for ever poorer outcomes. But that tolerance is not infinite.
The consequences of this leadership failure will continue to harm both the pockets and the health of the United Kingdom. In banking, it means more outages, more frozen payments, more families left stranded on payday. In healthcare, it means more patients harmed at care transitions, more missed observations, more families asking why nobody was watching.
In communities up and down this country, high taxes and deteriorating services are already contributing to quiet but growing doubts about the legitimacy of the state itself. When people pay more and receive less, when they can't access their own money on payday and can't get an answer to “who is looking after my mother?”, they don't just lose faith in a bank or a hospital. They lose faith in the promise that the state can deliver for them at all.
That is the real risk. Not that the NHS fails to modernise. Not that another digital programme underdelivers. But that the foundational compact between citizens and the state, I pay you provide, fractures beyond repair. And once that compact is broken, it doesn't come back easily.
The NHS does not need another app. It does not need another platform. It does not need another £330 million programme that gives each trust a better view of its own silo.
It needs the grammar that allows everything it already has to work together safely. It needs flow infrastructure: the governed, observable, traceable transfer of clinical responsibility across every boundary in the system.
This is not a technology problem. It is a foundations problem. And unless the foundations are resolved, nothing built on top of them will stand.
The NHS deserves to be able to see itself. Before it's too late.
References
- Joint Commission, “Sentinel Event Alert 58: Inadequate hand-off communication,” September 2017.
- Elliott RA, Camacho E, Jankovic D, et al., “Economic analysis of the prevalence and clinical and economic burden of medication error in England,” BMJ Quality & Safety, 2021.
- Oskrochi Y, et al., “Medication Errors at Care Transitions in the NHS: A Systematic Review,” BMJ Quality & Safety, 2024.
- Oskrochi Y, et al., “Clinical Safety Assurance of Digital Health Technologies in the NHS: A Cross-Sectional Study Using Freedom of Information Requests,” JMIR, 2025.
- NHS Resolution, “Annual Report and Accounts 2024/25,” July 2025.
- National Audit Office, “Cost of settling clinical negligence claims has more than tripled in last two decades,” November 2025.
- Healthcare Safety Investigation Branch, “Our response to the NAO report on clinical negligence,” citing OECD estimates.
- Lord Darzi, “Independent Investigation of the National Health Service in England,” GOV.UK, September 2024.
- British Medical Association, “NHS in deep crisis, says doctors' leader,” September 2025.
- British Medical Association, “Survey of grassroots doctors in England reveals deep concerns about the government's 10 year plan,” 2025.
- The King's Fund, “The 10-year health plan: public perceptions of the three shifts,” 2025.
- UK Treasury Committee, “More than one month's worth of IT failures at major banks and building societies in the last two years,” March 2025.
- SaaScada, “Hidden costs of legacy tech: UK banks shelling out £3.3bn annually on managing core systems,” September 2025.
- The Register, “Greater Manchester says its NHS analytics stack is years ahead of Palantir wares,” May 2025.
- Democracy for Sale, “Palantir's NHS data platform rejected by hospitals,” May 2025.
- Digital Health, “NHSE planning guidance on FDP ‘at odds’ with its existing policy,” October 2025.
