The immediate aim of the NHS COVID-19 Vaccine Registry was to enable large numbers of potential volunteers to be identified in a short timeframe and generate faster recruitment to COVID-19 vaccine studies, being supported by the Vaccine Task Force. Here, we explore how such a momentous project got underway.
What were the objectives of the NHS COVID-19 Vaccine Research Registry and why is it important?
We sought to create a service providing researchers the ability to engage with volunteers who want to be involved. Finding the required number of volunteers was the key factor affecting the speed at which studies could help find vaccines to protect us all more quickly - in turn helping the NHS and saving lives.
We also needed a way of ensuring different trials were not chasing the same volunteers and the development of a central register addressed both of these challenges.
The long term aim of the service is to help foster a research ready population for future health research studies and harness the increased public awareness of research.
What were the key deliverables and the biggest pressures?
Firstly, we needed a platform where the public could securely express interest in volunteering for vaccine studies and share key information about themselves that could match them to studies.
Secondly, we needed a way to divide that volunteer base into enriched cohorts by age, COVID-19 status and other key criteria, such as health conditions. Finally, we needed a way for vaccine research teams to invite cohorts of volunteers to take part in their studies.
The time pressure to have the service ready for the first vaccine study meant we worked in phases and got the volunteer sign-up function stood up first. Once that was launched, we built other elements while volunteers started signing up. The deadlines were extremely tight, with a first prototype developed in just four weeks and a private beta version available within six weeks.
Not only did we have the build itself, but we had to resolve all the data issues as well in those six weeks, in addition to all the marketing of the service so we would be ready to engage volunteers.
Tell us about the team who built the registry
A number of teams from across National Institute for Health Research (NIHR) and NHS Digital were responsible for building the end-to-end service.
This included representatives from NHS.UK, the Data Services Platform, data analytics and NHS DigiTrials team within NHS Digital, and the project management and researcher teams from NIHR.
The project also involved representatives from Scotland, Wales and Northern Ireland as a UK-wide initiative, as well as Cognizant for the usage of the NHS Digital CRM (MS Dynamics), through on-going support and maintenance contracts.
It was very much a joint multi-team, collaborative effort.
What were the biggest organisational challenges the team faced?
This was the first time the teams had collaborated together and we were all working remotely. We solved this by embracing digital and remote collaboration tools, allowing the teams to work virtually as one.
Daily collaboration sessions to shape multiple streams of work at pace meant we were able to deliver a solution over a very short period of time to support the Vaccine Task Force.
The key challenges were around producing something the public could understand and use, while also meeting the needs of the researchers.
What are the key technologies that enable the project - what's under the hood?
The front end was developed using .net core, Azure functions and Azure Logic Apps for data capture. MS Dynamics was used to control the email validation journey alongside NHSmail.
Talk to us about personal data. What challenges were unique to this project and how did you solve them?
In terms of personal data, the key issue was that it was not possible to build a verification system to enable linkage to central data in the timeframe available. That meant any data needed had to be collected within the system itself.
As a four nation solution, there was a complex set-up phase where we had to generate the right processes and agreements in place for the service to hold data in England from Scottish, Welsh and Northern Irish citizens. The project team worked closely with the devolved administrations to get formal agreements in place to enable data to be collected in line with regulations, and we published a Data Privacy Impact Assessment (DPIA) to provide clarity and transparency around this.
Having a full verification system can act as a barrier to users and, given we were looking to reach as wide a range of people as possible, we wanted to reduce the complexity of any checking. We needed a level of security, without it being perceived as onerous or overbearing. This was achieved by sending volunteers a single verification email that enabled them to progress to the data entry screens for personal data.
As we are working in the research environment, arrangements for use of data were also subject to ethics approval. This meant we had to ensure the use of the service was appropriately included in ethics submission. The main lesson from all of these challenges was the actual legal side of things.
The wording of the Data Sharing Agreements, DPIAs and other documents is only half of the challenge; the other half is about relationships. There was no document that could have solved these challenges if people hadn't been willing to get round the table and work it out. There were many late night conversations early on to get everything set up.
Tell us about the data: what are you storing, handling and transporting. How did you manage this pipeline?
Data captured by the front end is transferred from Azure CRM via Azure Data Factory to an SFTP endpoint and is landed in an AWS S3 secure location, which triggers a Databricks pipeline action. This then processes the data, deriving several new calculated fields, such as age and geographic location.
The data collected and derived - for example name, date of birth, age, email address, contact telephone number and previous medical conditions - are stored as assets in the platform and can only be accessed by professionals within the organisation with a Data Sharing Agreement (DSA) approved by the information asset owner.
How did you ensure accessibility and what sorts of user testing did you do?
The service was developed following the NHS Digital Service Manual guidance and using components already available in existing live systems for consistency.
We followed internal reviews with Government Digital Service (GDS) and NHSX to ensure we were meeting government standards. There has since been an independent accessibility audit that assures the approach taken.
The system was tested through user research, at two hospital sites before launch, and then by several individuals with differing accessibility needs during development. An online Hotjar survey stated 77% of individuals with accessibility issues rated the service as very easy to use.
What was the biggest lesson the team learned along the way?
We scoped, designed, tested and developed the registry from a first prototype and into private beta without any definitive sense of public demand and fully live environment - a scenario that any development team will appreciate.
We’d done extensive end-to-end testing based on anticipated volumes, so we knew it worked, but an immediate spike in interest from the national launch did cause some initial problems.
Very high numbers of people tried to sign up at once and as brilliant as this was, it caused performance issues with the user journey taking longer than anticipated to be completed. In response, we worked to further improve performance and carried out additional capacity testing, which ultimately resulted in a better, more robust product.
It’s a good example of learning being applied in a live, pressured environment to enable continuity of service and improvement.
Finally, what makes you most proud about the NHS COVID-19 Vaccine Research Registry?
The registry has made a tangible difference to efforts to fight the pandemic and it has had an important role in the pace and quality of the world-leading vaccine research the UK has been able to deliver.
Pride comes from the fact more than 480,000 people have signed up to the registry, a number of whom are now actively participating in coronavirus studies. Studies which have supported and enabled the development of working vaccines, some of which are in use and others that will enter the global market in the months ahead.
- No. of volunteers: 483,029 (as of 20/04/21)
- Time to deliver: 6 weeks
- Studies supported: 5
- Total number of site visits: 2.2m
- Web access: Akami, MS Dynamics
- Data Processing: Azure Data Factory and AWS Transfer Family
- Data storage: AWS S3
- Data standards: FHIR, SNOMED CT
- Analytics: Tableau
- Release management: Azure DevOps