Healthcare technology
British health technology is measured almost entirely by what has been installed. The harder questions are what it costs to keep running, and what happens to a record when a patient crosses a trust boundary.
The installation is nearly finished. The hard part is not
NHS England's frontline digitisation programme has a date attached to it. The published position is that 95% of trusts will have implemented or upgraded an electronic patient record by March 2026, with the remaining 5% delivering their plans to do so, against £1.9bn invested to bring every trust to a baseline level of digital capability. On the programme's own terms that is close to done, and it is a genuine achievement in a service that spent the previous decade buying nothing.
It is also the least interesting fact about the estate. A record system is a purchase. What a trust does with the record is an operating decision it takes every day for the next fifteen years, funded from a budget that the capital grant did not cover. The programme's next stated focus is maximising the impact and usability of what has been installed, which is a quiet admission that installation and use are different projects with different sponsors.
What the review found underneath
Lord Darzi's investigation into the NHS in England, published in September 2024, put the capital position bluntly. It identified a shortfall of £37 billion of capital investment against comparable countries over the 2010s, and noted that £4.3bn was moved out of capital budgets between 2014-15 and 2018-19 to cover in-year deficits. His summary of the technology position was that the NHS is in the foothills of digital transformation.
The productivity figures in the same investigation are the ones health technology has to answer for. Between 2019 and 2023 hospital staff numbers rose sharply while activity did not keep pace: 7% fewer daily outpatient appointments for each consultant, 12% less surgical activity for each surgeon, and 18% less activity for each clinician working in emergency medicine. A great deal of technology was installed during those years. Whatever it did, it did not show up there.
That is not an argument against the spending. It is an argument about attribution, and it is the reason this publication treats "digital productivity" claims in health with more scepticism than the equivalent claims in a bank. A hospital's throughput is set by beds, discharge pathways, diagnostics capacity and the availability of social care. A record system sits upstream of all of it and can improve without any of them moving.
The legacy estate nobody counts
The government's own review of digital capability puts the share of systems classed as legacy between 10% and 50% across NHS trusts, a range so wide that it is really a statement about how little is known centrally. Two neighbouring trusts can differ by a factor of five and both be inside the published band.
That variation is why automation lands unevenly in health. A trust running a modern record system with documented interfaces can put a classifier in front of a referral queue and measure the result. A trust whose clinical data lives behind three systems and a shared inbox is buying integration work and calling it automation. The distinction matters commercially, because the two look identical in a business case and diverge completely in year two, which is the argument this publication sets out in full when it examines what automation costs to run.
Read the contract, not the launch
The NHS Federated Data Platform is the clearest current example of why the terms matter more than the announcement. NHS England's own explainer records that a consortium led by Palantir Technologies UK Limited was awarded the contract in November 2023, that there is a maximum term of seven years but a committed term of only three, extendable by two years and then by two further single years, and that the funding covers up to 240 NHS organisations.
A three-year committed term inside a seven-year envelope is not a footnote. It sets the date by which the platform has to have proved itself to survive, and it tells every trust considering whether to build on top of it exactly how much certainty it is being offered. Reporting the headline value without the break structure gets the story backwards.
Foundry4 covers health technology as a strand of public sector technology rather than as a separate trade, because the procurement rules, the legacy problem and the accountability questions are the same ones that apply in a department or a council. Where the analysis turns on how the underlying records are structured and shared, it sits with the data and analytics coverage instead.