The Culture of Silence: How Flawed Claims Reached Parliament While Patients Waited for Surgery

The NHS Federated Data Platform is software, supplied by Palantir under a £330m contract, that helps hospitals join up data scattered across their systems. Whether it actually improves hospital performance has become one of the most contested questions in UK health policy, and the contract's future may hang on the answer.

The stakes are not abstract. As of early 2026, over seven million people in England are on a waiting list for hospital treatment. Many are in chronic pain. Some have conditions that will deteriorate the longer they wait. In the year to August 2025, over 79,000 patients were removed from NHS elective waiting lists because they died before being treated. When an operating theatre sits half-empty because the scheduling is poor, or a slot opens up and nobody fills it in time, these are not data problems. They are people who do not get their operation that week, and may wait months for the next opportunity, if they are still alive. Theatre utilisation is a dry phrase for something with a direct human cost.

Patients removed from NHS waiting lists because they died, by length of wait. Source: Hyphen analysis of FOI data from 127 acute NHS Trusts, year to August 2025.

There are three sets of evidence on whether this tool helps. All three point in the same direction. None of them prove the scale of benefit that has been claimed.

Yet somewhere between the evidence and the parliamentary debate, "the scale is uncertain" hardened into "the benefits are exaggerated and untrue." This is the story of how that happened, and why the best available evidence was never brought into the conversation.

Evidence set one: the official figures

NHS England publishes headline numbers for the FDP: as of March 2026, 111,589 additional patients through operating theatres, discharge delays cut, hundreds of thousands of patients removed from waiting lists.

How is the theatre figure calculated? The published methodology is simple. Take the period before a hospital switched on the tool, work out the average weekly number of operations, and project that forward. Anything above the projection counts as "additional."

The problem is what this ignores. There is no adjustment for COVID recovery, seasonal patterns, or anything else that might have increased operations anyway. A hospital that adopted the tool just after a COVID slump would record its natural recovery as a benefit of the software. NHSE's own methodology page acknowledges this, stating: "We cannot therefore draw conclusions about cause and effect as other variables have not been controlled for."

The waiting list numbers have a different weakness. The removals span inpatient lists (87,842), outpatient lists (218,389), and referral-to-treatment pathway closures (993,477). These include deceased patients, duplicate entries, people who moved abroad or went private, and patients whose condition improved or who were treated elsewhere. Useful housekeeping that improves data quality, but not new treatment for anyone, and the headline figures that Palantir cites bundle them together in ways that blur the distinction.

So the official figures deserve the scepticism they have received. The direction may well be right, but the method cannot prove it.

Evidence set two: the Medact article in the BMJ

In mid-April, the BMJ published a news article by an activist at Medact, a campaign group that opposes Palantir's involvement in the NHS. Using publicly available hospital data, the Medact article showed that Chelsea and Westminster, the original pilot site, improved at the same rate as neighbouring London hospitals that never used the FDP. It also showed waiting lists at the Trust rising in several specialties.

The Medact article landed hard. It was cited in a Westminster Hall debate, repeated by the Good Law Project, folded into campaign briefings, and treated by health journalists as the definitive debunking of the FDP's benefits.

As a challenge to the official figures, it is a fair piece of analysis. It uses the same data source as NHSE's own methodology to show that the "additional operations" could simply be the post-COVID recovery that every hospital experienced. On that point, the article is on solid ground.

Evidence set three: the study nobody mentioned

Here is what almost nobody in this debate seems to know. A month before the Medact article appeared, the BMJ's own journal arm published a peer-reviewed evaluation of the same pilot.

The study, in BMJ Health & Care Informatics on 12 March, was conducted by researchers at Imperial College London, funded by Chelsea and Westminster, with Trust staff facilitating data access. The analysis was performed independently by the Imperial team, and the paper was externally peer reviewed. It took a more careful approach than the NHSE methodology. Rather than a simple before-and-after comparison, it modelled what would have happened at Chelsea and Westminster without the tool, using 90 weeks of data and over 31,000 surgical bookings, and measured the gap between that projection and reality.

It found that theatres were used more efficiently after the tool arrived than the model predicted they would have been without it: booked utilisation was 15% higher than expected, actual utilisation 12.2% higher, and the number of bookings per session 10.9% higher. The authors were candid about the caveats. COVID disruption muddied the baseline, the tool wasn't adopted evenly across teams, and recovery trends couldn't be fully separated from the tool's effect. Their conclusion was careful: the tool was associated with better scheduling, and more research is needed.

In other words: promising, not proven.

The conflation

The Medact article never mentioned this study, not even in passing.

That omission mattered, because the two pieces measure completely different things. The Medact article's graphs track the total volume of theatre activity: how much surgery happened. The peer-reviewed study measured efficiency: how well each hospital used the theatre time it had. Think of it as the difference between how many miles a car drove and how many miles it got per gallon. A hospital can drive the same total miles as its neighbours while becoming far more fuel-efficient, and that is exactly what the study found.

The Medact article's graph: total weekly capped touch time in thousands of minutes for Chelsea and Westminster (blue) against regional peers and national median. This measures the volume of theatre activity. Source: Model Hospital data, as presented in the BMJ news article.

The peer-reviewed study's graphs: booked utilisation, actual utilisation, and bookings per session as percentages, with interrupted time series modelling, counterfactual projections (black lines), and 95% confidence intervals (shaded areas). These measure scheduling efficiency, a fundamentally different question. Source: BMJ Health & Care Informatics, March 2026. Reproduced under CC BY-NC 4.0

The waiting list graphs are even less relevant. The peer-reviewed study made no claims about waiting lists at all. Waiting lists rise and fall with referrals, staffing, beds, and discharges, none of which a theatre scheduling tool controls.

The Medact article's second graph: annual elective inpatient waiting lists by specialty at Chelsea and Westminster, showing rising trends. The peer-reviewed study made no claims about waiting list volumes. Source: NHS RTT statistical data, as presented in the BMJ news article.

So the Medact article demolished the official promotional figures, which deserved it, but was received as having demolished the evaluation too, which it never touched. Without the missing citation, readers had no way to tell the difference.

The conclusion hardens

Watch what happened next.

On 16 April, Martin Wrigley MP told Parliament that the BMJ had shown the Chelsea and Westminster benefits to be "exaggerated and untrue." That phrase now sits in Hansard, in a debate that shaped the ministerial response on the contract's future.

Campaign groups amplified the Medact article. Medact itself built it into a briefing urging Trusts to refuse the FDP. A Yorkshire Bylines piece by the No Palantir in NHS Campaign went further, claiming I had admitted the FDP's benefits were "weak and unproven."

What I had actually written, on LinkedIn in April, was that the official figures needed stronger methodology, and that the benefits were "directionally right even if the quantum is unproven." Having now read all three evidence sets side by side, that sentence turns out to be a more accurate summary of the total evidence than anything said by either camp since. But it was flattened into a confession.

At each step, the claim hardened. "The scale is uncertain" became "the figures are contested" became "exaggerated and untrue" on the floor of the House of Commons. The peer-reviewed evidence was available the entire time. Nobody referenced it.

I made the same mistake. I agreed with the Medact article publicly, told the editor of Digital Health that the official figures weren't solid, and deliberately avoided citing them when Naga Munchetty interviewed me on BBC Radio 5 Live. I assumed a critique published under the BMJ masthead would have engaged with the BMJ's own peer-reviewed evaluation. It hadn't, and I should have checked - now I have.

Why nobody corrected the record

Chelsea and Westminster had peer-reviewed evidence supporting their work, published in the BMJ's own journal family. When the Medact article challenged their pilot, they said nothing. The article itself notes the Trust didn't respond to a request for comment. The study's authors don't appear to have responded either. NHS England, which commissioned and funded the FDP programme, also stayed silent on the substance.

This was not simply the result of external pressure from campaigners, although that pressure is documented. The Medact article quotes an anonymous consultant describing a hostile environment for those who raise concerns about the FDP. But the silence on the other side is an institutional choice, made at management level in both the Trust and NHSE. Both organisations allowed the conflation to stand because the political climate around Palantir made any public engagement feel too risky. The result is a collusion of silence between the campaigners who control the narrative and the institutions that could challenge it but choose not to.

That silence had a price. A flawed conflation stood unchallenged for months, reached Parliament, and is shaping a billion-pound decision.

To be fair, the silence is not total. The national FDP programme has a funded communications team. Chelsea and Westminster's FDP team won the Tech for Good Award at the National Technology Awards and were finalists in several other categories. Rebecca Llewellyn celebrated the team publicly on LinkedIn as "the best tech team in the UK." Build with FDP hackathons are running. The web pages are maintained. But this activity is directed inward, towards the health tech community and the teams already committed to the platform. In the arenas where the evidence is being contested, in the BMJ, in Parliament, in the national press, there is near-total silence. No rapid response was submitted to the BMJ. No correction reached Hansard. The vacuum left by that strategic silence has been filled entirely by the platform's critics, and the people with the evidence to challenge them have watched it happen.

And the silence extends beyond Chelsea and Westminster. HSJ's Joe Talora has reported that flagship Trusts signed up to the FDP are still not using it, and said on The Briefing Room podcast that some signed up with no intention of ever doing so: "They've done what they're asked to do. They're not going to use it." HSJ has been consistently sceptical of the FDP, and that editorial stance colours its reporting. But the underlying fact Talora has uncovered is a problem regardless of the publication's position on the platform.

NHS organisations that have evaluated the FDP and concluded it adds nothing to their existing capabilities have every right to say so. Greater Manchester ICB did, with published reasoning. That is a legitimate position. But Trusts that signed up to satisfy a central mandate and then left the tool unused are doing something different. They are neither testing the evidence nor contributing to it. The Medium Term Planning Framework now effectively mandates FDP uptake for data warehousing, the canonical data model, and core products supporting elective recovery and cancer. Beyond theatre scheduling, the platform's capacity to replace the shadow IT that proliferates in every Trust, the spreadsheets, local databases, and ad hoc reporting tools that fragment data and duplicate effort, is a benefit in its own right. Trusts that sign up, tick the box, and then walk away are forgoing that opportunity without ever putting it to the test. In a system where 79,000 people died on elective waiting lists last year, calculated non-adoption dressed up as compliance is not a neutral act. The CQC's well-led framework asks whether Trusts use data and technology effectively to improve care. Ticking a box and walking away does not meet that test either.

What I am not arguing

The Medact article's investigative reporting stands on its own merits. The links between senior government figures and Palantir, the consultant's account of pressure on staff, the political reaction: these are legitimate journalism, and questions about governance in a £330m contract should be asked.

Nor is the peer-reviewed study beyond challenge. Its authors list their own limitations honestly, and the right description of their finding is promising rather than proven. Nobody should wave it around as proof the FDP transforms the NHS.

The point is narrower. "Not yet proven" and "exaggerated and untrue" are not the same thing, and the gap between them is exactly where a billion-pound policy decision should be made carefully rather than loudly. The Chelsea and Westminster evaluation is the most rigorous assessment of FDP effectiveness that anyone has published. It went through peer review, it was transparent about its limitations, it found a positive signal. Yet it has been effectively erased from the debate by a news article that never engaged with it and a silence from the people who should have been shouting about it. Whatever you think of Palantir, that is no way to make policy.

What should happen now

Anyone who disagrees with the peer-reviewed study should challenge it where its authors can reply: a rapid response in BMJ Health & Care Informatics. Anyone who has cited the Medact article as proof the evaluation was flawed should read both and notice that they never addressed the same claims.

NHS England should replace its crude benefits methodology. The current approach of projecting a pre-adoption average forward cannot distinguish the tool's effect from the COVID recovery, and the weakness of those figures has done more damage to the FDP's credibility than any campaign group. The evaluation shows what better measurement looks like. Commission more of it.

And the people who know what the FDP is and isn't delivering at Trust level need to find their voices. The current vacuum is being filled by claims neither side can verify, while the evidence that does exist sits in a journal both sides apparently forgot to read. There are seven million people on waiting lists who cannot afford for this debate to be settled by whichever story travels fastest.

Tom Bartlett is the founder of Bartlett Data Ltd and former Deputy Director of Data Services at NHS England, where he led the engineering team that built the Federated Data Platform. He writes in a personal capacity.