| Updated:
Internal NHS data indicates that Palantir’s hospital discharge tool cut delays by as much as 28 per cent, City AM understands, challenging recent reports from critics of the tech giant that it had made “no noticeable improvement” to the healthcare system’s efficiency.
According to an internal NHS England report seen by City AM, trusts using the Optica AI tool recorded a 14.1 to 14.9 per cent fall in discharge delay days among long-stay patients, compared with a 2.9 to 3.5 per cent increase at trusts that had not adopted it.
For patients managed directly through Optica, NHS England calculated a 28 per cent reduction in delay days.
The figures were drawn up late last year in response to an analysis by the Health Foundation, which was subsequently cited in reports questioning the effectiveness of Palantir’s Federated Data Platform (FDP).
The Health Foundation found “no noticeable improvement” in hospital discharge performance associated with Optica.
NHS England’s response disputes the way that conclusion was reached, arguing the analysis included large numbers of shorter hospital stays which the tool is not primarily designed to address.
The NHS document has not been published and its findings do not establish that Optica alone caused the reductions.
The disagreement instead centres on how the impact of the software should be measured and which patients should be included.
It comes as the government prepares to review Palantir’s £330m NHS contract ahead of a February 2027 break clause, the Department of Health and Social Care (DHSC) has confirmed.
Health Department defends Palantir deal
“As per standard practice, this contract will be reviewed ahead of the February 2027 break clause,” a DHSC spokesperson told City AM.
However, the department defended the programme, adding that “thousands more patients are benefitting from the NHS Federated Data Platform every month, including through the reduction in the number of unnecessary days spent in hospital after treatment.”
The decision is likely to receive heightened scrutiny after separate official figures showed the expected lifetime cost of the wider FDP programme has risen to £1.1bn, while its forecast financial benefits have been revised down to £808m.
The £1.1bn figure covers the wider rollout and operation of the programme rather than Palantir’s contract alone.
The National Infrastructure and Service Transformation Authority (NISTA) attributed the higher cost to “increased delivery at scale”, while the reduction in expected benefits followed updated rollout plans, improved data and more detailed assumptions about usage and performance.
Despite the changes, the FDP retains a green delivery rating from NISTA, making it one of 29 out of 172 major government programmes to receive the rating.
Dispute over Palantir’s impact
At the centre of the conflicting findings is Optica, Palantir’s AI tool used to coordinate the discharge of patients who may require additional care or support before leaving hospital.
The Health Foundation compared performance more broadly across hospitals that had and had not adopted Optica.
NHS England argued in its unpublished response that this diluted the apparent impact because most hospital admissions involve shorter and simpler stays, while Optica is primarily intended for patients with more complex discharge needs.
Its analysis instead examined patients staying for more than seven, 14 and 21 days.
Across those groups, trusts using Optica recorded reductions in delay days of between 14.1 and 14.9 per cent, compared with increases of between 2.9 and 3.5 per cent at trusts without the tool.
“The Health Foundation report appears to misunderstand what the tool does,” a Palantir spokesperson told City AM.
“Optica is focused on supporting the more efficient discharge of long stay patients with complex needs. And that is where it is delivering results.”
“By lumping in the vast number of cases where patients are in hospital less than a week, they are comparing apples and oranges, and diluting the impact of the product.”
However, Tom Bartlett, founder of Bartlett Data and a former deputy director of data engineering at NHS England, told City AM there were “problems with both of the methodologies that have been used”.
One issue is that adoption can differ significantly within the same hospital.
A trust counted as an Optica user may have some wards using the software heavily and others not using it at all, making trust-wide comparisons less precise.
“If you’ve got a hospital that’s got 10 wards and two of them are using Optica, you should be doing an analysis of those two wards, not of the 10,” Bartlett said.
He nevertheless disputed the Health Foundation’s conclusion.
“I think [the NHS analysis] is more accurate than it’s been given credit for,” he said, describing the finding of “no noticeable improvement” as “absolutely wrong”.
Bartlett told City AM further analysis would be needed to isolate wards actively using the software and determine more precisely how much of any improvement could be attributed to Optica.
Unpublished NHS figures
The NHS response was prepared after the Health Foundation raised concerns privately late last year, but has not subsequently been published even after the think tank’s findings became public.
Dr David Nicholl of Doctors’ Association UK, which has campaigned against the Palantir contract, urged caution over the figures.
“I am so sceptical about anything that NHS England put out about FDP,” he told City AM, citing conversations with NHS staff who had questioned whether the platform offered benefits over software already in use.
“I haven’t heard anyone who’s had a good word for it,” he said.
The Office for Statistics Regulation has separately scrutinised NHS England’s presentation of figures used to promote the benefits of the FDP.
The watchdog told City AM that its previous investigation concluded on 22 July, but said it had since received “further, separate casework” relating to the FDP which it was now considering.
There is no set timetable for that work, and the OSR said it was not expecting the unpublished analysis seen by City AM to be formally submitted to it.
It has asked NHS England to keep it informed about promised improvements to the presentation of FDP data, including adding a more prominent caveat to its benefits website.
“We also set out our expectation that any figures cited by ministers or in official briefings about the FDP and its effectiveness are communicated in a clear, accurate manner,” the regulator told City AM.
£330m deal faces 2027 decision
The scrutiny comes as the government approaches a decision over the future of one of its highest-profile tech contracts.
Palantir won the £330m contract in 2023 to provide the technology behind the FDP, which is aimed at connecting data held across different parts of the NHS and provide hospitals with tools covering areas including patient discharge and waiting lists.
The wider programme is now forecast to cost £1.1bn over its lifetime and deliver £808m of financial benefits, although Palantir said the latest published benefit-to-cost measure remained £4.92 for every £1 spent.
The company also pointed to NISTA’s green assessment and said the most recent annual increase in costs was below the five per cent threshold used to identify material changes.
“Palantir software is helping the NHS to deliver better patient care – including additional operations, a reduction in discharge delays and an increase in patients finding out whether they need cancer treatment,” a spokesperson said.
Nicholl said his opposition to the contract had increasingly shifted from concerns around Palantir itself towards whether the NHS was receiving sufficient value for the money being spent.
“It’s a colossal sum of money,” he said. “For something that doesn’t do any better than what we’ve already got.”
Bartlett, however, warned that terminating the contract would carry costs of its own because the NHS would need to procure and roll out replacement systems across hospitals already using the FDP. “If the supplier is delivering, why would you pull it out?” he said.
DHSC has not indicated which way the government is leaning. Its confirmation that the contract will be reviewed ahead of February 2027 means ministers will have to assess the competing evidence over the coming months, including whether the improvements identified in NHS England’s analysis can be independently attributed to the platform.
+ There are no comments
Add yours