The single patient record will run late. The Treasury priced that in two decades ago.

The Department of Health and Social Care has done something unusual for a flagship programme still two years from its own deadline: it has told the public, in writing, that the date might not hold. An impact statement published alongside the 10-Year Health Plan warns that the single patient record, due in 2028, faces a “systematic tendency for project appraisers to be overly optimistic” about cost and duration, and points to the programme’s scale and its dependence on ageing legacy systems as the reason. Read generously, that looks like an organisation that has learned from two decades of scars and chosen candour over false confidence. It is a more interesting document than that, because the line DHSC is quoting was not written about the single patient record. Whitehall wrote it in 2003, for every IT project it has approved since.

That phrase comes from HM Treasury’s Green Book supplementary guidance on optimism bias, based on a 2002 review of large public procurement by the consultancy Mott MacDonald. It has applied to every central government business case for over two decades, and it singles out one category for the harshest correction of all: equipment and software development, including IT, carries an upper-bound cost uplift of 200%, against 24% for standard buildings and 66% for the riskiest civil engineering work. The Treasury has assumed, since before most current NHS digital leaders started their careers, that a government IT programme’s first honest estimate can be out by a factor of three. DHSC citing optimism bias is not a fresh warning. It is a reference to a two-decade-old instruction that every NHS digital business case is already required to follow.

Whether that correction actually works has one clean historical test. The National Programme for IT launched in 2002 with a budget of £2.3bn over three years, promising every NHS patient a shared electronic record. The government dismantled it in 2011. The National Audit Office’s own review put total spend at £9.8bn; the programme never delivered its central promise. £9.8bn against a £2.3bn baseline is a 326% overrun, past even the Treasury’s 200% worst-case allowance for exactly this kind of project. Apply that same standard Green Book starting point to the single patient record’s own business case, and the implied worst case is a programme costing three times its published estimate. NPfIT’s actual overrun was bigger than that. The government’s most pessimistic built-in assumption about NHS IT was not pessimistic enough to describe what happened to NHS IT last time.

That is not one bad decade. In 2013 the health secretary promised a paperless NHS by 2018. The government missed the target, pushed it to 2020, then effectively abandoned that date too after Professor Bob Wachter’s 2016 government-commissioned review concluded 2020 would also fail and proposed 2023 instead. By his own account in 2023, even that revised target delivered “reasonably good” progress rather than the maturity he had hoped for. NHS England launched an £8.1bn digital transformation programme in 2019, due to run to 2023/24; the National Audit Office’s 2020 assessment found it progressing “slower than expected”, with funding it judged unlikely to be sufficient for the ambition. NHS England awarded the Federated Data Platform to a Palantir-led consortium in 2023, targeting 240 NHS organisations live by 2027; fewer than 100 were connected by 2025, and ministers are now weighing an early exit from the contract. Four programmes, four technologies, four governments, the same shape of delay, and the single patient record has not yet reached its delivery phase.

This is not unique to health. The Infrastructure and Projects Authority’s delivery-confidence ratings show the share of major government projects rated amber/red or worse rising from 16% of the portfolio in 2013 to 31% by 2016, and holding around 28-34% through the early 2020s before the IPA redesigned the rating scale. A third of the government’s most important projects have looked troubled in any given year for well over a decade. NHS IT sits inside that pattern, not apart from it, but it carries the single largest optimism-bias correction the Treasury applies to any category of spending, which amounts to an official admission that government IT forecasts are its least trustworthy numbers of all.

None of this makes the single patient record a bad idea, and DHSC flagging the risk is not dishonesty. But a cited optimism-bias correction is not a mitigation, it is an admission that the number has already been marked up once, and NPfIT shows the mark-up has not always been big enough. DHSC’s own risk assessment for the SPR names the same causes as NPfIT’s post-mortem: dependency on a single vendor, clinical resistance to sharing data, and the cost of connecting decades of separate systems at once. Those are governance and integration problems, not arithmetic ones, and no percentage uplift in a business case template fixes them. If the single patient record beats the pattern, it will be because someone manages those three risks harder than NPfIT’s sponsors did, not because the spreadsheet carried a bigger number.

Sources

• Department of Health and Social Care, Health Bill: single patient record and information sharing – impact assessment (gov.uk)
• HTN Health Tech News, “DHSC publishes impact statement on 10-Year Plan noting single patient record will take several years”, January 2026
• HM Treasury, Green Book supplementary guidance: Optimism Bias (gov.uk)
• National Audit Office, review of the final benefits statement for programmes previously managed under the National Programme for IT, June 2013
• UK Parliament Public Accounts Committee, “Dismantled National Programme for IT in NHS: report published”
• National Audit Office, Digital transformation in the NHS, May 2020
• Professor Robert Wachter, Making IT Work: Harnessing the Power of Health Information Technology to Improve Care in England, September 2016 (gov.uk)
• PublicTechnology, “NHS estimates Palantir data platform will deliver returns of five times its costs”, October 2025
• Infrastructure and Projects Authority, Annual Report on Major Projects (2013, 2016, 2021, 2023-24 editions) (gov.uk)