In 2024 we were approached by Prof Tjeerd van Staa and Francine Jury from the Division of Informatics, Imaging and Data Sciences to support the development of an antibiotics Knowledge Support System for General Practitioners in the UK. This was an ambitious project together with the BRIT2 team in the Faculty of Biology, Medicine and Health.
The Problem
Over-prescription of antibiotic drugs has become a concern in the UK. The over-exposure of the population to antibiotics has resulted in the evolution of drug-resistant strains of bacteria. This is partly due to antibiotics often being prescribed as a quick solution for relatively minor conditions, where the patient would have recovered easily without the drug anyway. To illustrate the scale of the problem, in 2016 the government set a target to reduce the level of inappropriate prescribing by 50% by 20211 (p15). This target was not met.
The Aim
The aim of the research was to trial whether a software intervention within GP surgeries could reduce the number of antibiotics prescribed. This software would run on the GP’s own desktop machine alongside the electronic patient record and would be launched when a patient presented with symptoms of interest, e.g. a cough or chest infection.
The GP would answer a small number of questions about the symptoms, and then the software would run an analytical model developed by the BRIT2 team. This model informed the GP of the likelihood of any further problems should the antibiotic not be prescribed. The GP would then make their decision and could print out a patient information sheet if they decided not to prescribe an antibiotic.
RSE Input
The BRIT2 team had already developed a prototype piece of software to do this; however, their software developer had left the team. The software was therefore untested and requirements were evolving quickly as the project and trial developed. The RSE team made substantial alterations and fixes in order to make the software robust, accessible, and reliable. This was important as without this the software would not be accepted by the NHS’ IT teams for deployment.
The original prototype was only compatible with a single supplier of electronic patient records. This placed a limit on the number of GP practices that could use the software. Erdem Atbas from the RSE team, assisted by Andrew Jerrison who provided design and advice, added compatibility with a second supplier to increase the range of surgeries that could participate in the trial. They also re-engineered the software to make it easier to add more suppliers in the future. In addition, the software was made more extendable to allow for easier adaptation to future studies that were not necessarily focussed on antibiotics.
Technical Aspects
The technology stack for this project was C#, .NET Windows Presentation Foundation (WPF) which allowed us to reuse the existing code. WPF is a popular stack for Windows app development and allowed us to use some existing .Net development expertise within the team.
The software created was a desktop application that was launched by the GP when they were looking at a patient record. It interfaced with the electronic patient record software to discover the relevant medical information needed to run the analytical model.
The GP could then answer a few questions about the patient’s symptoms and the software would display the information that the GP needed to determine whether the patient needed an antibiotic. The software could then optionally be used to update the patient's medical record with details of the decision.
Outcomes
Although the results from the trials have not been analysed yet, the BRIT2 team has already developed a good understanding of the challenges in developing and deploying software for GP practice-based research, and the RSE team was instrumental in supporting that.
Working with the Research IT software engineers has been a masterclass in what true team science looks like. Andrew and Erdem’s professionalism, flexibility, and creative solutioning transformed a complex, ambitious research challenge into an elegant, scalable digital tool that can be repurposed to address multiple healthcare challenges. They didn’t just support our research team, they were an integral part of it, helping us deliver on our research objectives and achieve more than we could have envisioned when we first drafted the grant application.
If you are interested in having a Research Software Engineer work with you on your latest research project or, if you are interested in help and advice from the team, please get in touch.
(1) UK 5 Year Antimicrobial Resistance (AMR) Strategy 2013-2018