Preparedness is not the same as deciding to leave. My view is that the NHS should be able to show that a critical platform could change without losing data, disrupting care or weakening control. At a Health and Social Care Committee hearing in November 2026, the Health Secretary declined to say whether an exit plan was being prepared, saying only: "We have detailed plans for all eventualities." Source: https://www.digitalhealth.net/2026/11/nhs-england-to-demand-palantir-clears-fdp-benefit-claims/

Architecture matters. FDP is an overlay above existing datasets, not one central database: each trust and ICB has its own instance and remains its data controller, while NHS England runs a national instance. NHS England says lock-in is avoided through the incremental term, end-contract planning, in-house engineering, partnering and movable standards. A minister told MPs the NHS owns IP on tools that could be transferred to a new supplier, such as the NHS canonical data model. These are claims to verify, not assume. Source: https://commonslibrary.parliament.uk/research-briefings/cbp-12264/

I would test whether data and standards can leave with the NHS, rather than remain usable only inside the current supplier's tools. Compatibility matters: the NHS Confederation has raised interoperability concerns with existing local tools, while the CDAO Network says less digitally mature trusts face greater lock-in risk. Nuffield Trust calls procuring and transitioning by March 2027 "challenging". A migration exercise should use dependent local applications and interfaces, and prove the canonical model is reusable by another provider. Source: https://commonslibrary.parliament.uk/research-briefings/cbp-12264/

Permissions and audit trails need scrutiny. May 2026 reporting said some supplier staff had access to identifiable data; NHS England later said it had incorrectly communicated who required access. I would test documented access approvals and independently reviewable logs showing who accessed identifiable data, under what authority and when, with exceptions explained. Source: https://commonslibrary.parliament.uk/research-briefings/cbp-12264/

Spine2 offers a historic example. It replaced the old BT-run Spine over the August 2014 bank holiday weekend after rehearsals and a "ten-minute plan"; a like-for-like switch kept messages sent by end systems unchanged, followed by 45 days of intensive monitoring. Insourcing came later: the Data Transfer Service moved in-house as MESH about 18 months after go-live, while a database support contract remained. I would copy that discipline, not assume its design fits today. Sources: https://www.digitalhealth.net/news/29478/spine2-is-alive https://www.digitalhealth.net/2016/02/nhs-spine-2-managed-entirely-in-house/

The independent evaluation by Imperial College Projects is expected to be published only after the extension decision. I would want independent exit testing before any irreversible cutover, with findings given to those accountable for the service. None of these checks requires a decision to switch. Source: https://commonslibrary.parliament.uk/research-briefings/cbp-12264/

My practical checks would be:

  • Usable exports and mappings for every required dataset.
  • Dependent local applications and interfaces still work.
  • Access permissions and audit logs tested.
  • Clinical continuity, fallback and recovery rehearsed.
  • NHS ownership of reusable IP, tools and standards.
  • Independent findings, with remaining risks explicitly accepted.

Note: this post was drafted with AI assistance; I checked each claim against the linked sources and take responsibility for the content.

Loading comments…