How mathematical modelling saved lives during the COVID-19 pandemic
A mathematical model developed at Swansea University helped to save lives and support the NHS in Wales during the height of the Covid-19 pandemic.
At the start of the COVID-19 pandemic, trends in the disease in Wales were not always in sync with other areas of the UK. However, much of the planning evidence was based on UK-level scenarios. The result of this in one Welsh health board was considerable over-provision of critical care staffing at a key period.
Health and disease control decision-making in Wales is devolved to the Welsh Government, who convened their Technical Advisory Cell (TAC) to provide emergency scientific advice for the pandemic. TAC identified an urgent need for bespoke mathematical models for the spread of COVID-19 and its impact on local hospitals in Wales, and the Swansea Modelling Team was created in response.
Developing a modelling tool
Professors Gravenor and Lucini brought together an interdisciplinary team that combined expertise in epidemiology, mathematics and software engineering. Under great time pressures, the team had developed their main research tool, the Swansea Model.
The team used Welsh data, including demographics, age-specific cases, hospitalisations, critical care admissions and deaths, to model each local authority area separately. They then used the model to estimate the R (reproduction) number, and to understand the spread of the virus in Wales and how it might be controlled.
The research team designed a code to schedule detailed policy scenarios as requested by key stakeholders and to explore options of changes to specific factors.
The Swansea Model was extensively reviewed over several iterations by colleagues on TAC, the TAC modelling sub-group and the NHS National Modelling Forum. The research team worked daily with clinicians, NHS planners, and politicians in order to get realistic data to input, to focus on the most pressing questions, and to clarify the model’s limitations and predictions.
Reducing hospital admissions and deaths
The Swansea Model informed health policy decisions by the Welsh NHS and Government. It was used in every health board in Wales and by the NHS Wales Informatics Service.
The model enabled the health service to provide critical care staffing across Wales at safe and efficient levels based on early and accurate scenarios for planning hospital capacity, and to accurately plan for severe call demand on the Welsh Ambulance Service based on early and accurate scenarios.
It was also used as an evidence base for major national interventions, for example, the October ‘firebreak’ in Wales which reduced a runaway R rate to below 1. The model provided accurate estimates of the rebound time after the firebreak, to support planning. Over the period, here were an estimated 5,000 fewer hospital admissions, 350 fewer ICU admissions, a 33% reduction in peak ICU occupancy, and 1,100 fewer deaths.
The model was also used to inform the follow-up interventions that brought the epidemic under control by late December after a challenging winter and new-variant increases in transmission rates, and was used to model the impact of emerging vaccination programmes and the road maps out of lockdowns in early 2022.
Research team
Professor Biagio Lucini, Professor Mike Gravenor, Ed Bennett, Mark Dawson and Ben Thorpe: the Swansea Modelling Team – Swansea University
Research partners
Wales Technical Advisory Cell, Wales TAC Modelling Subgroup, Welsh Government, Supercomputing Wales, The Scientific Pandemic Influenza Group on Modelling, UKRI, Microsoft.