Meet the experts: How Paul Aylin continues to play a key role in the development of NHS surveillance tools for Dr Foster

Paul Aylin is director of the Dr Foster Unit at Imperial College London, where he draws on more than 20 years of experience using health data to inform his research. Here, he details his journey from public health doctor to statistical expert and discusses what the future holds for the Unit.

25 Nov 2020 | 3 min read

Professor Paul Aylin

Alerting tool changes course of Dr Foster

Originally, I trained as a medic, eventually qualifying as a public health doctor. Following my training, I stumbled into work in death certification at the Office for National Statistics, looking at how to improve the quality of data, and then moved on to work as an environmental epidemiologist within the small area health statistical unit at Imperial College London, where I had access to large data sets including cancer registry and HES data.

The Bristol Royal Infirmary cardiac scandal broke while I was in this role, and I put forward a proposal for work on the inquiry to provide evidence on mortality rates at the Bristol Royal infirmary children’s cardiac unit compared with other centres in England. From there, I became more involved in health service research and was invited to work for the Shipman inquiry, looking at statistical techniques to detect outliers in mortality data.

I met Professor Sir Brian Jarman through the Bristol Inquiry and began working with Dr Foster, and soon realised the potential for more detailed analyses of hospital outcomes. It was at this point I realised I had a deep interest in health data and its application as a catalyst for improvement. We took the techniques we had learned with the Bristol and Shipman inquiries and developed the idea of an ongoing surveillance tool for NHS hospitals, to help alert them of clinical areas that might need further investigation.

Dr Foster took this idea and operationalised it, using data that was only a couple of months old – revolutionary in those days as health data analyses were often a year or more out of date. That immediacy was very innovative, and the move altered the course of Dr Foster, essentially taking it from a publishing company to a healthcare analytics company.

Developing a broad range of research topics

Within the Dr Foster Unit, we do a lot of work around developing and validating approaches to analysing big routinely collected health data sets. We have recently taken the mortality alerting system that we helped to develop and evaluated that, asking hospital trusts what they do with that data and what impact it has on performance. As part of that work, funded by the National Institute of Health Research (NIHR), we found relationships between other indicators of quality of care and our mortality alerts, as well as subsequent falls in mortality following our alerts which helps validate the work we do.

At Imperial, I work closely with a number of other research groups, across acute care, primary care, and mental health. The other unit I work with is the NIHR Health Protection Research Unit. We are looking at anti-microbial resistance and healthcare associated infections, using population databases to examine patterns of prescribing and how these patterns have changed.

This year I’ve been involved in papers looking at Covid-19, including the NHS response, antibiotic prescribing, hospital admissions, and the impact of the virus on other non-COVID19 patients.

Rich population health data provides exciting prospects

Within the Dr Foster Unit, we are developing our expertise around machine learning techniques, and have initially applied some of that learning to data to anticipate when a hospital might reach a crisis point within emergency care and have to divert patients to other hospitals.

The Unit is also leading on integrated care and how we can use these new, very rich population data sets, broadening from Dr Foster’s roots in acute care. We are aiming to delve deeper into AI techniques, and how we can apply that to understand complex patients and improve their healthcare. With AI, we can look not just at similarities between patients but their trajectory, too, bringing in a temporal dimension as well as the side-by-side comparison. Dr Foster has access to large, complex data sets that can help us model interventions, which is exciting.

For more information on the Dr Foster team, click here.