Critics say the claim that the medical data sold by UK's Department of Health of millions of NHS patients to American drug companies is anonymized is misleading
Toby Helm / The Guardian : Tweets: @jonashworth Tweets: Jonathan Ashworth / @jonashworth : Shocking & needs urgent investigation from ministers. Patient data must be absolutely confidential between doctor and patient, not a plaything for US tech corporations to make profits. https://www.theguardian.com/ ...
Context & Ripple Effects
The row over NHS data sales has been building for over a year: the Department of Health sold records of millions of patients to American drug companies in late 2019, a move defended at the time as the price of better care, while the NHS separately promised that data shared with Google's DeepMind would be anonymized before analysis. Now shadow health secretary Jonathan Ashworth is calling the anonymization claim itself misleading and demanding urgent ministerial investigation, turning a commercial arrangement into a question of whether the NHS's anonymization promises hold up at all.
First-order effects
- Jonathan Ashworth's call for urgent investigation puts ministers on the defensive over a sale already completed, forcing the Department of Health to defend its anonymization standard in public rather than in contract terms.
Second-order effects
- The credibility of the anonymization pledge is the same asset the NHS leans on for its other data partnerships — including the DeepMind arrangement — so every challenge to one deal raises the consent and trust bar for the next.
Third-order effects
- The pattern has since escalated, not receded: the NHS moved to scrape medical histories of 55 million patients for third-party sharing in 2021, and Biobank was later found to have shared data of roughly 500,000 volunteers with insurers despite pledging not to — suggesting anonymization assurances function as policy cover rather than enforceable limits.
The trend: UK public-health data sharing is drifting from case-by-case partnerships toward population-scale extraction, with the gap between anonymization promises and practice becoming the recurring fault line.