Construct And Validate Clinical Features Of Severe COVID-19 And Its Clinical Trajectory Prediction Model

Objective: To describe the clinical features and risk factors of severe COVID-19.To develop and validate a predictive model for the risk of developing and dying from severe COVID-19.METHODS: This is a multicenter retrospective study in which clinical data were collected from3609 patients from 15 countries worldwide who were hospitalized with positive nucleic acid tests for neocoronavirus from January 2020 to October 2020,randomly divided into model training and validation groups,and clinical indicators with non-zero regression coefficients were selected in a LASSO regression model to screen for the occurrence of severe COVID-19 The best risk factors for the occurrence of severe COVID-19 and death were selected in the LASSO regression model.The clinical indicators selected in the LASSO regression model were then integrated,and a multifactorial logistic regression analysis was introduced to construct the column plots.The test criteria were defined as statistically significant at P≤0.05.The subject operating characteristic(ROC)curves were plotted,and the area under the curve(AUC)was calculated to test the performance of the column line graph in the training and validation groups;the calibration curves were used to assess the degree of agreement between the actual results and the predicted results of the column line graph in the training and validation groups.If the AUC > 0.75,the prediction model we built is considered to have good prediction performance,while in the range of 0.5 to 0.75 is considered to be acceptable.Finally,the net benefit was calculated and the decision curve was plotted,using the decision curve analysis(DCA)method,which was used to determine the clinical utility of the column line graphs.Results: 1,Old age,high CRP,and lymphopenia were three independent risk factors for severe COVID-19.2,Body temperature(Temp),glutamate transaminase(AST),white blood cells(WBC),C-reactive protein(CRP),blood sodium(Sodium),(blood potassium Potassium),urea nitrogen(BUN),platelets(Platelets),respiratory rate(RESP),heart rate(HR),arterial blood lactate(LAC),and finger pulse oximetry(SPO2)were significantly correlated with the occurrence of severe COVID-19.LASSO regression was used to select 12 variables to construct a line graph predicting the risk of developing severe COVID-19.The calibration curves showed good agreement with the area under the curve(AUC)values of 0.963 and 0.0.962 for the training and validation groups.The decision curve analysis indicated that the training and validation groups showed a large positive yield within the broad risk range,respectively.3,Age(age),temperature(Temp),C-reactive protein(CRP),blood potassium(Potassium),urea nitrogen(BUN),platelets(Platelets),and heart rate(HR)were associated with the occurrence of death in severe COVID-19 The LASSO regression was used to select 7 variables to construct a line graph to predict the risk of death in severe COVID-19.The area under the curve(AUC)values for the training and validation groups were 0.816 and 0.844,and the calibration curves showed good agreement.Decision curve analysis indicated that the training and validation groups showed good positive yield over a wide range of risks,respectively Conclusions: 1.Old age,high CPR,and lymphopenia are independent risk factors for severe COVID-19;2.We established a relatively accurate column line graph to predict the risk of developing severe COVID-19 more accurately;3.We have created a relatively accurate column line graph that can predict the risk of developing death in severe COVID-19 with relative accuracy.