Prediction models learnt from Wuhan
The models are derived in a population of 775 people with confirmed COVID-19 admitted to one of two hospitals in Wuhan, China. The population is a low to moderate severity population (4.3% died and 9.7% had a ‘poor outcome’ defined as acute respiratory distress syndrome, intensive care unit admission, mechanical ventilation, extra-corporeal membrane oxygenation therapy or death). The models perform well in the original population.
Our prediction models are supposed to support decision making at different points of the health care pathway.
- Community triage: Models using Demographic, premorbid Conditions and Symptoms (DCS). The model of death prediction using DCS is NOT listed below as it's not validated very well internally.
- Hospital admission: Models using Demographic and Laboratory results (DL).
NB: Risk prediction models are only ever a complement to clinical judgement. They are not a replacement.
Step 1:choose your model
Step 2:enter patient values
% of patients with these characteristics are predicted to .
Based on the Wuhan derivation data, this patient would be categorised as . Interpret this with CAUTION – validity in other contexts has not been established.
- Low risk (estimated average rate in the derivation data was %)
- High risk (estimated average rate in the derivation data was %)
- Very high risk (estimated average rate in the derivation data was %)