OpenSAFELY, built by the Bennett Institute for Applied Data Science within our department, covers GP records for tens of millions of patients in England. Researchers never see identifiable data. They develop their analyses using dummy data, submit their code, and it runs automatically against the real records. That privacy-first design earned the platform a Queen Elizabeth Prize for Higher and Further Education in 2025 – and, more importantly, the trust of the GP profession, patient groups, and privacy campaigners.
The platform proved what whole-population GP data could do during the pandemic. Its first major study analysed 17 million patient records and provided some of the earliest robust evidence on who was most at risk of dying from COVID-19 – including the finding that higher death rates among Black and South Asian communities could not be explained by pre-existing health conditions alone.
That evidence directly shaped government shielding guidance and workplace protections. OpenSAFELY went on to track vaccine coverage across nearly 58 million records, exposing disparities in uptake by ethnicity and deprivation, and its Service Restoration Observatory gave NHS England real-time data on how primary care activity was recovering after lockdown.
In total: over 100 published papers in journals including Nature, the BMJ, and The Lancet, from more than 30 organisations including NICE, UKHSA, and NHS England. Earlier this year, a £17 million Wellcome investment expanded the platform to incorporate NHS Talking Therapies data alongside GP records, opening up new research into which mental health treatments work best for which patients.
Now, the same platform is open for any health question. NHS England, which controls the data, will review all applications submitted by the deadline and inform applicants of decisions by 31 May 2026.
The infrastructure was built for COVID because that was the emergency. But the privacy protections, the scale, and the rigour apply to every important question in primary care – chronic disease, treatment outcomes, service delivery, and more.
How to apply
The applications section of the OpenSAFELY website has everything needed to get started: a tutorial, a live online sandbox with dummy data, a feasibility tool for exploring SNOMED code usage, and 150,000 words of technical documentation.
NHS England charges £25,000 per project. Researchers who do not yet have funding are encouraged to apply and indicate this on the form.
The application window closes at 17:00 on 30 April 2026.