Imperial College London
Early in 2020, a key part of the COVID-19 response in the UK was to identify people at risk of poor outcomes in order to “shield” them from infection. 6.8% of the English population were initially identified as at high risk of having severe outcomes from COVID-19 and were added to the Shielded Patients List. We have researched the processes of creating and then applying this new category of person as a case study on how data are used in personalisation. This Shielded Patients List was created initially using a clinical algorithm applied to healthcare data over a period of weeks, then added to by clinicians, and later expanded to include more people at high risk of severe outcomes by utilising a personalised risk score algorithm for predicting COVID outcomes derived from healthcare data during the first wave of the COVID-19 pandemic. The technologies that underpin the creation of this list- data sharing, data interoperability, the application of algorithms, user generated data, data systems, machine learning – all existed before the Covid-19 pandemic, but the creation of the Shielded Patient List represents a novel shift in data practices and how they are imagined within the medical and scientific fields, particularly with regards to healthcare data and wide scale personalisation and population health management. As a part of this work, we collaborated with Patient Experience Research Centre (PERC) to develop and administer a questionnaire to those who were on the Shielded Patients List to understand how they were impacted by the creation of this category. We also examined the Shielded Patients List methodology to explore key features of personalisation such as the need for participation (in data sharing), precision (in defining a category), prediction (of an outcome given that category) which then leads to some action.