OpenSAFELY in brief

⚠️ This is a simplified overview – for details and technical specifics, see OpenSAFELY in detail.

OpenSAFELY is a software platform for analysing large, sensitive datasets – without compromising any of the secrets inside. It’s open for research, but keeps private data safe.

It’s different to most other data platforms. Rather than allowing researchers to make copies of the raw data, OpenSAFELY leaves the data where it’s already stored. Instead, OpenSAFELY takes the research code to the data – then brings back aggregated results in the form of tables and graphs.

Researchers only get to see insights. They never get unconstrained access to the raw data.

OpenSAFELY is used in various different ways, but the most important way is with data from the NHS – particularly GP records.

Why NHS data matters

GPs take notes every time they talk to a patient – those notes are part of your health record. Taken together, health records from millions of people are combined in huge datasets.

Various people (mostly scientists, but also government policymakers and other researchers) are very keen to analyse these datasets.

But because we’re talking about GP health records, those datasets are full of some of the most sensitive and private information. Privacy is a vital part of medical practice. Nobody wants the things they tell their GP to become public knowledge.

So there’s a tension: some people want open access to the data for research. Others want to make sure the secrets inside that data are kept safe.

Studying the data at scale helps to:

make the NHS work more efficiently
improve healthcare services
reduce costs
save lives

That’s why this data is so important. And that’s why scientists are so keen to use it for research. But at every step along the way, it’s vital that we protect the privacy of every individual patient.

OpenSAFELY was built to solve this problem.

How we use OpenSAFELY with NHS patient data

The NHS England OpenSAFELY service, funded by NHS England, uses the OpenSAFELY platform to analyse NHS healthcare data.

Researchers and other specialists wanting to run analyses have to make an application to NHS England, which controls the application process and decides which proposals will be approved.

Openness is very important (that’s one reason why it’s called “Open”SAFELY. We ask everyone using the service to openly document how they intend to use it, to share the code they write, and much more. In the spirit of openness, we maintain live lists of approved projects, jobs running in OpenSAFELY, and published research.

Your privacy comes first

OpenSAFELY was designed from the outset to protect patient privacy – that’s the most important thing.

We promise that:

Your medical privacy is our top priority.
OpenSAFELY respects Type 1 Opt outs and the National Data Opt-out: if you’ve opted out, your patient data is automatically excluded from research projects.
No-one can access your data except the owners of the datacentres where it’s already stored. Researchers using OpenSAFELY get to see summarised insights, but don’t get unconstrained access to the raw data.
We take security very seriously. OpenSAFELY uses state-of-the-art computer security techniques to make sure that no-one can use the service without permission; and that everyone using the service uses it responsibly and safely.
Your data will not be sold. We couldn’t even sell it if we wanted to – it’s not ours to sell. We provide the tools that help scientists do research using the data, but we don’t control the data itself.

The legal basis

OpenSAFELY is made possible by formal instructions from the government, known as “Directions”. Until now, it ran under the COVID-19 Directions, which only permitted research projects that related to COVID-19.

In 2025, the government issued new Directions that make it possible to conduct research on any healthcare topic.

How OpenSAFELY is used in other ways

On top of the NHS-funded service, we often undertake collaborative research projects with scientists at other institutions. For example, we’ve done lots of projects like this with colleagues from the London School of Hygiene and Tropical Medicine.

Projects like this are different because:

they are funded separately, and
researchers from the Bennett Institute and third party organisations work together to conduct a scientific study

→ Find out more about collaborative research projects

In addition, we also work with other organisations on data infrastructure projects. These usually create opportunities to put the technology behind OpenSAFELY to work in entirely new ways: perhaps with new datasets, running in new datacentres, and in areas outside healthcare.

For example, we’re currently collaborating with the National Institute of Teaching on OpenSAFELY Schools, to develop ways of analysing education data using the same principles of privacy and security.

→ Find out more about our data infrastructure collaborations

How OpenSAFELY is different from other data platforms

OpenSAFELY doesn’t let researchers take copies of datasets, or even copies of small bits of datasets. The data always stays put.
OpenSAFELY generates dummy data to help researchers run tests as they go along – so they don’t need access to the real data for that either.
OpenSAFELY makes everything as open as possible, to allow for, and encourage, external scrutiny and scientific rigour

Where the money comes from

Funding for the OpenSAFELY platform comes from Wellcome, UK Research and Innovation, the National Institute for Health and Care Research, and the Peter Bennett Foundation.

NHS England provides funding for the NHS England OpenSAFELY service.

Who does the work

OpenSAFELY is run by a small team of software engineers, data scientists, clinicians, pharmacists and epidemiologists working at the Bennett Institute, part of the University of Oxford.

Find out more

→ About our patient and public involvement and engagement activities

→ About our panel of Digital Critical Friends, independent members of the public who help guide our work