Millie Green has recently joined our team as a data scientist and in this blog she describes our OpenSAFELY co-pilot programme which is being used to support new users in accessing OpenSAFELY and producing detailed analyses using the NHS records of 58 million people. This is a new service that we have recently started to offer and, as OpenSAFELY is a rapidly changing platform, we will continually update and work through the final details of how we expect this co-pilot programme to work.

Onboarding new users to OpenSAFELY

OpenSAFELY has been created as a platform where researchers can work independently, using the tools and data (with appropriate permissions) without necessarily needing substantial engagement with our team. We have already begun onboarding the first wave of additional users for OpenSAFELY (which you can read about here) and full documentation is published on all aspects of the platform and all of the feature development and product support is in a public forum. On top of this extensive documentation and Q&A forum, as onboarding new users to OpenSAFELY is more complex than granting a simple data download, or a login and password, OpenSAFELY has begun a co-pilot programme.

What is the OpenSAFELY co-pilot programme?

All new users of the OpenSAFELY platform get access to our supportive co-pilot programme, where each new OpenSAFELY user is assigned a member of the OpenSAFELY team as their co-pilot for the duration of their project. The objective of this programme is to provide users with dedicated support in order to help them achieve a completed analysis within an agreed timeframe (usually 4 weeks).

All new users of OpenSAFELY have an initial meeting with their co-pilot, where they are introduced to the key concepts of the OpenSAFELY platform, where the available documentation is explained and where co-pilots describe how they would like to use their co-pilot for additional guidance and support. As a minimum, new users are offered at least a weekly meeting with their co-pilot to discuss progress, outline any outstanding issues, discuss any blockers and set goals for the remainder of the programme.

By the end of the co-piloting period, new users of the OpenSAFELY platform will be:

  1. Familiar with the OpenSAFELY framework
  2. Able to run code locally
  3. Approved OpenSAFELY users and are able to execute code against the real OpenSAFELY database
  4. Able to generate the study population for their project
  5. Able to run analysis scripts for their project
  6. Able to generate outputs, ready for review

The role of an OpenSAFELY co-pilot

As with any analysis in flight, there will be bright sunny moments where everything is going according to time and plan, and then there will be moments of turbulence! OpenSAFELY co-pilots are there to help users hold the course that they have set for their project. The role of an OpenSAFELY co-pilot varies, depending on the needs of the new user, but generally their primary job is to help ensure that new users are comfortable with the key concepts OpenSAFELY set out within the full user manual. This includes the following:

  1. The OpenSAFELY philosophy
  2. The programs and tools users must know (or be willing to learn) to use OpenSAFELY
  3. Codelists
  4. Study definitions and cohort extractor
  5. Job server
  6. Output review

The secondary job of an OpenSAFELY co-pilot is to provide guidance and support to new users when using the OpenSAFELY platform to deliver analyses for the first time. This might be in the form of directing a user to some helpful documentation, or it might be sharing their own knowledge and personal experience about the OpenSAFELY platform.

Finally, our OpenSAFELY co-pilots are there to be a friendly face, provide users with encouragement, and to share their tips and knowledge about using OpenSAFELY in order to help new users become knowledgeable OpenSAFELY users.

How can I start using OpenSAFELY?

While only approved users can execute code against the real OpenSAFELY database, anyone can access and use all the OpenSAFELY-specific tools needed to define analysis datasets, generate dummy data, and run analysis scripts in a computational environment that mimics the secure environment where real analyses are run. The best place to begin is by reading our getting started guide. The analysis workflow gives a high-level view of the steps involved in an OpenSAFELY research project and if you wish to install OpenSAFELY on your own computer, you can follow our installation instructions.

When you are comfortable with how OpenSAFELY works, we recommend following one of our OpenSAFELY walkthroughs (see this notebook) to guide you through the platform workflow on your own computer with dummy data, rather than using the documentation pages alone.

If you then decide that you would like to undertake a study using OpenSAFELY please contact the team by email at or contact us via the contact page on our website.