New research offers an external and independent assessment of the first QCovid prediction tool used to estimate the risk of a person being hospitalised, or dying from COVID-19.
This study was designed to evaluate the validity of the original QCovid algorithm for COVID-19-related deaths in the Welsh adult population.
QCovid’s development was led by Oxford University in early 2020 and was subsequently used by the National Health Service (NHS) to identify high-risk individuals for the shielding programme and later for vaccine prioritisation.
This latest study that uses Welsh population data to validate QCovid was led by the Population Data Science group at Swansea University, Health Data Research (HDR) UK, in collaboration with teams at the universities of Oxford, Edinburgh, and Leicester, as well as the Office for National Statistics. It was funded by HDR UK and the Medical Research Council.
Read the full study here in IJPDS – the International Journal of Population Data Science
The team used anonymised, individual-level, population-scale data held in the Secure Anonymised Information Linkage (SAIL) Databank. This included all individuals aged 19 to 100 years, living in Wales on 24th January 2020 (the date of the first confirmed COVID-19 case in the UK) and registered with a general practice (GP) contributing data to SAIL.
The SAIL Databank contains electronic health record (EHR) data from approximately 83% of all GPs in Wales. SAIL achieves this through a partnership with NHS Wales’ Digital Health and Care Wales (DHCW), acting as a trusted third party, enabling the data anonymisation and transfer process to SAIL.
The researchers studied the data retrospectively in SAIL using robust statistical analysis methods. The team evaluated and determined that the QCovid prediction tool has a high degree of accuracy for the Welsh population.
This was achieved by analysing the number of COVID-19 deaths attributed to each influencing factor used in the QCovid model for the population of Wales, for example, age, ethnicity and the pre-existence of health conditions, amongst other factors. By studying data in this way, pooled together from an anonymised individual-level, the team could compare the real-world risks of COVID-19-related deaths with the QCovid prediction tool.
This study of the Welsh population replicates a recently published study validating QCovid in 35 million adult residents of England by the Office for National Statistics.
The availability and use of the SAIL Databank was important in this study. It provided a population-scale health data source that was independent of the original study population in England, used in QCovid’s initial validation. Being able to repeat the study and achieve similar results in diverse populations is an important component of scientific research, and is crucial for validating prediction systems that use routine data where the results may be used to plan clinical management of individual patients.
Building on previous research, this latest study provides an important contribution as to the validity of the QCovid tool to predict the inherent risks of COVID-19 across the UK, and it’s possible application to other populations.
These findings helped guide risk management decisions and target vaccination and treatment programs for the most vulnerable individuals in society. The researchers are in the process of conducting further analyses to determine the validity of the latest QCOVID models in predicting hospitalisation resulting from COVID-19 infection and the effects of the vaccine programme in Wales.
HDR UK Wales and Northern Ireland Research Officer & Data Scientist, Jane Lyons, who led the research, said, “Validation studies are an important part of all research and we have been able to show that the original QCovid models predict COVID-19 deaths in the Welsh population as effectively as they have in the English population. This study is a great example of a collaborative team science approach across multiple organisations in the UK. Following on from this study, we are currently in the process of validating the updated QCovid risk prediction algorithms in estimating the risk of COVID-19-related deaths and hospital admissions in Welsh adults following one or two doses of COVID-19 vaccination.”
SAIL Databank’s co-director and QCovid co-investigator, Ronan Lyons, said, “This study highlights the importance of the SAIL Databank in assessing the relevance and validity of algorithms developed in England and other countries to the Welsh population. The results were fed to policy makers in the Welsh Government COVID-19 Technical Advisory Group and have helped to influence policies designed to protect the Welsh population.”