The COVID-19 pandemic started in December 2019 and quickly spread around the world. A key challenge in all healthcare settings is to identify patients with possible COVID-19 at presentation, to inform both clinical management and infection prevention and control interventions. Diagnosis relies in most cases on PCR of nose and throat swabs for the SARS-CoV-2 virus. However, the current turn-around time for the PCR test is about 24 hours, so the test result is not available at the time of admission. Instead cases are triaged according to how likely patients are to have COVID-19 based on their symptoms and other features.
There is no validated tool for clinical triage and misclassification is common. Furthermore, a large proportion of patients are triaged to the ‘Amber’ group which requires either isolation in side rooms or cohorting in half-occupancy wards to ensure sufficient physical distancing for infection control. This puts significant pressure on the already limited hospital bed capacity. There is therefore an urgent need for better tools to risk stratify a patient’s risk of COVID-19 at the point of presentation.
We propose to use detailed demographic, clinical, physiological, laboratory and radiology data to explore how these might be used to improve patient triage at presentation to hospital.