COVID-19 Data and ConnectivityThe model used existing de-identified health data to make accurate predictions, rather than existing approaches of relying on virus information and disease dynamics collected as the outbreak gathers pace.

The analysis is the first whole-population mortality model that considers risk data from before as well as during the pandemic. Crucially, the findings suggest that the model could predict excess deaths at the start of an outbreak before the new infectious disease is fully understood.

In March 2020, researchers used de-identified health records of 3.8 million adults in England to build a predictive model using three factors: pre-pandemic risk of death, the expected infection rate of the SARS2-CoV virus in the community, and the relative risk of death in people with COVID-19.

The research team accessed GP, hospital and death records, and searched for mentions of cardiovascular disease, diabetes and other high-risk conditions. Data were accessed safely and securely via the NHS Digital Trusted Research Environment (TRE) for England.

This new study, published in the Journal of the Royal Society of Medicine, provides an update to the 2020 model using de-identified electronic health records of 35.1 million people in England.

In this current study, the model was applied to the TRE. It predicted 100,338 deaths in people who caught COVID-19 over one year, close to the 127,020 actual figure that occurred between March 2020 and March 2021.

Scientists say the results highlight the value that securely held population-wide electronic health data can add to predicting the course of pandemics and could be vital to predicting excess deaths from future outbreaks.

“Infection modelling is usually the remit of the virologists and specialist infectious disease epidemiologists who look at viral spread. However, this is not enough to predict the path of a pandemic and an individual’s underlying health must be taken into account,” says Professor Amitava Banerjee, Chair in Clinical Data Science at UCL and senior author on the study.

Dr Mehrdad Mizani, Health Data Scientist at the BHF Data Science Centre and study author, said: Electronic health records are already available and – when accessed ethically and safely – are an untapped resource that should be part of pandemic planning and surveillance to help us keep ahead of the disease.”

The paper is available here .