US big data company Palantir is facing more opposition to its bid to win the £480m NHS England Federated Data Platform (FDP) contract as doctors across the UK call on the government to step in and pause the tender process. A new report outlining seven risks posed by choosing Palantir has been sent to MPs, and says that nearly half of NHS patients will opt out of sharing data if a private company is selected to develop and run the platform.
The paper, The NHS Federated Data Platform and Planatir – 7 Key Risks, has been sent to MPs today, Politico reports. It puts forward concerns around lobbying and inside dealing during the procurement process for the FDP, a vague scope for the project and a failure to design for patient consent.
Produced jointly by law firm Foxglove and non-profit membership organisation Doctor’s Association UK (DAUK), the report says that Palantir has a “poor reputation” and would “exacerbate” problems with public trust in the NHS. Both organisations are part of a joint effort with other campaign groups to sue the UK government over its plan to “hand over confidential data” to Palantir.
Critics of US-based Palantir have long believed that the company’s secretive nature and close ties to US security agencies such as the CIA make it unsuitable to work with public bodies in the UK. But it has been heavily involved in the development of the FDP, which is being set up to help data sharing across the NHS to improve patient care.
The report also criticises NHS England, highlighting that data projects within the national health service have failed because of a “lack of public trust” and says that senior leadership has been steered away from developing in-house solutions or drawing from open-source projects.
The two stakeholders provide alternatives for NHS England and the UK government, saying that none of the Palantir sceptics lacks belief in the potential of data. It points to Open SAFELY, an open source platform for health record data, and OneLondon, which joins up healthcare across the capital, as case studies of how data platforms in healthcare can be designed to be transparent.
Foxglove and UK doctors point to a ‘rigged’ procurement process for FDP
According to the report, the FDP procurement fails to meet the standard of being “transparent and scrupulously fair” due to lobbying by Palantir and the waiving of procurement procedures during the Covid-19 pandemic.
Content from our partners
“Prior to gaining a tow-hold in the NHS, at a time when procurement rules had been suspended, Palantir had no track record in health,” the report explains. It points to Palantir’s Foundry Platform, which is normally used in sectors such as aerospace and energy, but said it focused on the healthcare market in 2019.
“Liam Fox and officials in the Trade Department held closed-door meetings with Palantir bosses at Davos,” the report continues. “Briefing notes say these meetings were set up to discuss the ‘untapped’ commercial potential of NHS data.”
Foxglove and Doctors Association UK claim that from mid-2019 the US firm continued to lobby the NHS through closed-door dinners and private meetings in California. By 2020, Palantir was selected to provide its Foundry platform for a new ‘datastore’ – the contract was worth a token £1. As reported by Tech Monitor, by the contract’s first renewal, the firm increased the price to £1m and then to £23m by the second and has since won another £11.5m extension.
“For Palantir, the most valuable outcome of these Covid-19 contracts was not the £34m, but the opportunity to install itself, without any competitive tender process, in the heart of the NHS,” the report states.
The scope for FDP remains vague with patient data at risk
The report hones in on the scope of FDP and that NHS England has failed to ensure it will be a good value for money due to its vagueness.
It references a report conducted by the National Audit Office (NAO) in 2020, Lessons Learnt From Major Programmes, which examined different programmes for the government. NAO said that the scope of the programmes was one of the “root causes” of challenges seen during the programmes.
Foxglove and DAUK say that this problem is now seen in the scope of the FDP, which they say has been “difficult for even experts to follow.” The initial tender states five uses of the FDP, however, it also says that these uses will expand, which leads the report to conclude that nobody knows exactly what the government is planning to do with it.
“The documents also state the customer base for the FDP is ‘wider than just NHS England’ without clearly explaining who else will have access,” the report states. “Integrated care systems will have access, but so too, it is suggested, will providers commissioned under a contract to provide NHS services, which potentially includes a raft of public and private organisations.”
Foxglove and DAUK also argue that the lack of precision in the FDP scope increases the likelihood of the NHS getting poor value for money: “It increases the risk of being left with a system which doesn’t fulfil the NHS’ core needs, and of paying for features it doesn’t need at all,” it claims, pointing out that any changes to the scope later in the project are liable to increase project costs.
But another risk of the FDP project is the failure to design the database for patient consent. The report says that this is one of the “greatest risks of this system” and another lesson not learned from previous NHS data projects, pointing to the failure of care.data and its subsequent abandonment in 2016.
“By the time care.data was abandoned in 2016, over 1.6 million NHS patients had opted out of the scheme – enough to seriously degrade the quality of the data,” Foxglove and DAUK say. It also says that the NAO found there had been “insufficient focus on how to gain patient acceptance to share their data.” This happened again in 2021 when another programme known as General Practice Data for Planning and Research was put on hold after over 1m patients in a month opted out of data sharing.
Research from YouGov shows this could happen again with the FDP – nearly half of the English adults who haven’t yet opted out of NHS data-sharing said they would be likely to do so if the FDP was introduced and run by a private company. This means an additional 20 million patients removing consent.
Palantir would create a monopoly lock-in
Foxglove and DAUK say that for the FDP to be successful, it needs to be able to work alongside other systems such as OneLondon. It also says that if the NHS wanted to switch to another supplier, FDP needs to enable that.
However, it says that the feedback from data scientists and NHS managers says that selecting Palantir would create the opposite effect: “By technical design, Palantir has explicitly avoided creating a system that you can easily extract data from or that works well with other industry–standard data analytics systems. Palantir’s system pushes people to its own proprietary systems, and switching costs will be very high.”
It points to Jupyter, an industry-standard data science tool used by most NHS data scientists. “Palantir offers its own proprietary tool that does not work with this system.” It also says that “insiders” report that another pilot of a central government use of Foundry, unrelated to health, took months to have data extracted.
A monopoly could also mean higher prices for the NHS more generally. The Competition and Markets Authority (CMA) flagged a similar issue when it announced its investigation of the United Health proposed acquisition of EMIS. As reported by Tech Monitor, the government agency said that competition could be “substantially reduced” if the acquisition went ahead and could result in contract costs going up for the NHS for “lower quality offerings.”
Palantir’s ‘failed pilots’ and ‘bad reputation’ are causes for concern
The last two risks from the report call out Palantir’s ‘dodgy’ reputation and examples of failed pilots, which could be foreshadowing what could happen with the FDP.
“To defend the FDP plan, Palantir and some officials have leant heavily on two supposedly successful uses of the Foundry software – in supporting the vaccine rollout, and in a pilot of the software in a single digitally immature NHS trust, Chelsea and Westminster,” it says.
But the authors argue that these isolated examples are not success stories and that there is a lack of transparency around the data or documentation used to support the claims, and that NHS managers, data scientists and health tech experts have told them differently.
“These experts say that the FDP system is poorly designed,” they say. The report also says that several Foundry pilots have failed, but this has never been acknowledged, that further pilots of the ‘elective care recovery system’ have been indefinitely suspended and that FDP unnecessarily duplicates existing NHS infrastructure.
Palantir told Tech Monitor that its pilots haven’t been paused or suspended in 11 NHS Trusts and that it is currently live in four of the Trusts referred to in the report. Another is “on to be live.”
“We’re proud that our software supported the COVID vaccine rollout and, since then, has helped deliver a 28 per cent waiting list reduction and faster cancer treatment at Chelsea and Westminster NHS Foundation Trust,” a Palantir spokesperson said.
“As a software company, we don’t collect or monetise customers’ data – we simply provide tools that help them organise and understand their own information.”
