The COVID pandemic has led to millions of elective surgeries around the world being cancelled or delayed. We urgently need to restart these operations if we are to avoid the most disastrous effects of untreated non-infectious diseases and health conditions. But, with COVID infections set to continue for months – if not years – patients, surgeons and healthcare decision makers need to better understand what having COVID at the time of surgery might mean for the risk of complications or death. Only then will we be able to make informed choices about when and how to safely restart elective surgeries and best manage already-stretched critical care resources.
Working with the global CovidSurg Collaborative, this project is using large-scale data and machine learning to develop a tool that can predict a person’s risk of dying within 30 days of having any type of surgery while infected with COVID-19 (within the 7 days before or the 30 days after surgery). The CovidSurg Risk Stratifier tool integrates data from 8,500 patients across more than 750 hospitals in 69 countries. It uses patient and clinical factors that are highly associated with post-operative death – like the patient’s age, the surgical specialty and anaesthetic type – to calculate an individual’s mortality score. The aim is to help surgeons, healthcare decision makers and surgical patients better understand risk and make informed choices about surgical care during the COVID-19 pandemic.
Surgical decisions necessitate personalized approaches that inform surgeons, patients and healthcare professional of individual risk of particular patient. CovidSurg Risk is machine learning-based model, validated against real-world prospective patient data, that predicts the risk of death for patients undergoing surgery with COVID-19.
—George Gkoutos, Chair of Clinical Bioinformatics at University of Birmingham
Impact and outcomes
When surgery is unavoidable, a patient’s CovidSurg mortality score can inform consent processes and decisions about how to allocate resources for post-operative and critical care. It may also help reduce vast surgical backlogs by identifying patients who have a very low mortality risk – and therefore opportunities to increase the number of operations that can safely take place.
The CovidSurg Risk Stratifier tool has global application; the model is based on international data and the surgical and patient factors used to calculate the final risk score are all readily available – even in low-resource settings, where its use may be especially important given vaccination rollouts are likely to take years.
The project is looking to broader the tool’s application to cancer surgeries.
COVIDSurg Collaborative (led by the NIHR Global Surgery Research Unit) and the University of Birmingham.
HDR Midlands: Laura Bravo, Victor Roth, Cardoso, Luke Slater, Andreas Karwath, Simon Ball and George Gkoutos
COVIDSurg: Dmitri Nepogodiev, Omar Omar, Elizabeth Li, Joana Simoes, James Glasbey and Aneel Bhangu
Developing a risk prediction model for COVID-19 patients needing intensive care
This project is using large-scale real-time data and analysis to help hospitals rapidly and more reliably predict when a patient with severe COVID might need intensive care.
Better Care Projects
Projects Access to large-scale ‘research ready’ patient data is allowing a range of academic, industry and clinical innovators to develop the robust and regulated digital health tools and...