The UK is in a unique position when it comes to health data. Our GP records are perhaps one of the richest and most complete sources of health data in the world. Tapping this resource has the potential to uncover associations, links, or other patterns that could both change healthcare for individual people and the way healthcare is delivered for the better.
The rapid response in a time of huge uncertainty due to the emergence of COVID-19 underscored the potential of such a resource to understand which traits, pre-existing conditions, or groups of people were most at risk from infection and respond accordingly to protect them.
But how to do this safely, securely, accountably, and quickly without exposing patients sensitive – and confidential – data, or risking this data falling into the wrong hands by moving it about the country for analysis?
The OpenSAFELY team managed this where others had failed in a novel way, that, importantly, still safeguarded people's data.
Rather than copying, moving, or downloading the data to analyse elsewhere, and risk infringing people’s privacy (even if pseudonymised first). OpenSAFELY went straight to the horse’s mouth, as it were.
They worked with the organisations that store the GP data as part of business as usual for the NHS – specifically TPP and EMIS – and developed an open source software platform that enables researchers to run code across the GP data remotely, without them ever needing to see the data or to move it.
Importantly, they also worked with patients’ groups, representative bodies, and privacy advocate groups to ensure that the approach was acceptable to all.
It meant that trusted researchers – following a rigorous application process – could safely and securely run analyses and ask questions on almost real-time data from around 58 million people without that data ever leaving its original secure database.
All queries run against the GP data are logged, all approved projects, and all analytical code is shared in the public domain so that everyone can see what the data is being used for, and others can re-use or sense check the analytical code.
So, what does all this mean? What can be done with access to all these patient records?
The truth is, we're at the tip of the iceberg at the moment. But so far there are 84 separate projects using OpenSAFELY-TPP and OpenSAFELY-EMIS, from researchers across the UK. See box.
The lessons learned from building OpenSAFELY have also been used as evidence in the Government-commissioned review Dr Ben Goldacre is leading on better, broader, and safer access to health data for research.
However, the project has also had more tangible and immediate impact as, indeed, it was set up and hoped to.
For example, OpenSAFELY-TPP and OpenSAFELY-EMIS were used to provide answers to urgent clinical questions early in the pandemic, identifying those most at risk of hospitalisation or death from COVID-19, the risks associated with living with children, and the effectiveness of certain COVID-19 treatments.
A further important impact was the discovery and confirmation of ethnic differences in infections and COVID-19 hospitalisation in England, with some minority ethnic populations facing both higher risks of infection and worse outcomes than the white population. Finally, OpenSAFELY was included as a case study in a series of citizens’ juries commissioned by NHSX, the National Data Guardian, and will be included as a case study in the forthcoming NHS data strategy.
The citizens’ juries, carried out by colleagues at the NIHR ARC Greater Manchester, found that the jurors considered OpenSAFELY to be the most transparent, trustworthy, and secure of the three data sharing initiatives looked at.
To summarise, OpenSAFELY-TPP and OpenSAFELY-EMIS have demonstrated the power of leveraging one of the UKs greatest and most unique healthcare assets – NHS GP records – and continues to do so. As a fully portable, modular, and open-source software platform, it can be used to unlock such benefits from any healthcare database in England or elsewhere.
You can find more details, findings, and links to the tools themselves at www.opensafely.org, as well as on twitter @OpenSAFELY.