Case study: Making automated Covid testing a reality with Acutest
Read about how Acutest supported an NHS Trust with their automated Covid-19 test laboratory in 2021.
In early March 2021, frequent testing of pupils and staff was the defence against renewed Covid-19 infection and school closure but, despite the availability of free lateral flow tests, the following months were punctuated by whole-school closures as children and teaching staff fell ill with Covid-19.
To meet this challenge, an NHS Trust and a renowned UK University pioneered the first high-volume, automated testing laboratory in the UK designed to process tens of thousands of saliva samples every day from schools. The aim was to give early warning to schools so that infection could be contained, and the schools could remain open. The heart of the laboratory was its information management system (LIMS), that tracked every sample through its journey from receipt to resulting, and communicating test results to each child’s guardians, the schools’ coordinators, GP surgeries and Public Health England.
Project goals and achievements
LIM’s targets were to:
- Process as many samples as possible for 18-hours every day of the year without interruption or downtime with an initial target of 12,000 per day.
- Complete the sample journey from receipt to results dispatched within four hours to allow schools to rapidly respond without having to close the school.
- Maintain absolute accuracy between the person who provided the sample and the result of the test, whether negative or positive.
- Highlight any tests that may have suffered issues and to identify, locate and resubmit those samples for retesting.
In live operation it exceeded its initial target of 12,000 samples per day, achieving a stretch target of 30,000 samples per day a full five months before scheduled go live and 50,000 when fully operational—an achievement that could only happen in an agile development programme.
Acutest’s approach
Following Agile best practice from the start, the programme used the Business Driven Development methodology to draft requirements and acceptance criteria. Close working between developers, biomedical specialists, and testers was enabled by adopting the ‘Three Amigos ’ approach. The stories and scenarios developed gave a clear and above all shared understanding of what needed to be delivered and acceptance criteria to know when it was done.
Throughout the programme, we followed a risk-based approach to prioritise the stories that were selected for development in each sprint. Those stories that were at the highest risk of failure were the first to be designed, prototyped and tested in order to fail fast and fail early while there was time to fix defects. Development and test resources were assigned stories according to their capability to mitigate the risk inherent in the functionality of the story. The rigour of testing was also matched to the intrinsic risk of each user story—high-risk stories attracting frequent automated testing and the use of session-based testing.
The develop/test/deploy cycle ran in sprints of one to two weeks with the duration flexed according to the demand urgency for the functionality under construction. The Three Amigos approach used to formulate requirements continued on into sprint planning and involved the development team, biomedical scientists and test. This approach to joint working facilitated early prototyping and simulation which proved to be very successful in surfacing defects early in the development process.
Quality Assurance issues overcome
Although not a fully continuous delivery programme, the frequency of delivery from an agile programme created issues in the always-on environment of the Covid-19 testing lab. The LIMS and the API through which it interacted with the laboratory robotics could be tested in isolation whether through simulations in the development environment, user acceptance in the test environment or non-functional testing in the pre-production environment. In contrast, end-to-end integration testing which was required for the laboratory’s certification was a major test challenge. End-to-end testing involved the laboratory robots, the conveyors, the sample handling systems, and the assay tools—physical components of the assay process that only existed in live. Near continuous delivery meant weaving testing in with live operations without compromising accuracy of live tracking and resulting or sample processing capacity.
This was only achieved by:
- Simulating the laboratory universe in which LIMS operated to prove that it functioned as expected.
- Running focused tests using the live laboratory during time set aside for laboratory cleansing and biosecurity.
While not being failings of the LIMS, non-functional testing unearthed usability issues that were slowing the approval of test results by the Biomedical Scientists. As sample volumes ramped up to 30,000, these delays accumulated and were creating a bottleneck which threatened our ability to meet the four-hour receipt-to-test target and put undue pressure on the Biomedical Scientists. To resolve the issue quickly, the development process switched to a more evolutionary approach:
- Sprints were shortened to one-week to accelerate delivery of user interface solutions.
- Co-development between development resources and Biomedical Scientists was put in place to cut the ‘loop’ out of the feedback loop, allowing us to shortcut development and enable continuous prototyping.