COVID-19 early warning score: a multi-parameter screening tool to identify highly suspected patients

Cong-Ying Song; Jia Xu; Jian-Qin He; Yuan-Qiang Lu

The copyright holder for this preprint (which was not peer-reviewed) is the . https://doi.org/10.1101/2020.03.05.20031906 doi: medRxiv preprint In our study, we firstly developed a diagnostic model based on multivariate logistic regression analysis. Since this diagnostic model was a logistic regression equation and inconvenient to operate in clinical environment, we converted the diagnostic model into an scoring tool by assigning values to each variable in the equation. Besides of that, we enrolled two more parameters into the final COVID-19 EWS, including age and meaningful respiratory symptoms. Based on the statistical analysis, we found that elderly people were more susceptible to COVID-19 and develop severe, which may due to their lower disease resistance and more basic diseases; some significant symptoms, such as cough, expectoration and dyspnea, can help to strengthen the association between fever and lower respiratory symptoms and weaken the weight of fever caused by other reasons. As a result, the COVID-19 EWS contained easy-to-get parameters on clinical manifestation, epidemiological characteristics, basic vital signs, laboratory and radiologic data, which provide a more comprehensive assessment of patients. Through re-evaluation in the validation dataset and comparison with other methods, results showed that both of our diagnostic model and COVID-19 EWS had relatively high and stable sensitivity and specificity.