A retrospective study on the epidemiological characteristics and establishment of early warning system of severe COVID-19 patients

OBJECTIVE: To explore epidemiological characteristics and risk factors of COVID-19 in Chongqing and establish early warning system that provides a feasible protocol for clinical assessment of COVID-19. METHODS: A retrospective cohort study of 133 confirmed COVID-19 cases was conducted from January to March, 2020. They were assigned to mild group (n=65) and severe group (n=68). Univariate analysis and multivariate Logistic regression analysis were used to identify the independent predictors of severe cases. An early warning system for COVID-19 was established, with accuracy evaluated by ROC analysis. RESULTS: The age of the severe group was significantly elder than that of the mild group (P 0.05). Multivariate Logistic regression showed that the age, shortness of breath, lymphocyte count, PCT level, LDH level, APTT level, and CRP level were independent predictors for severe COVID-19. COVID-19 prediction model (including independent risk factors) was established, showing a high accuracy and capability for predicting higher risk of severe COVID-19 (with a AUC value of 0.8842, sensitivity value of 84.33%, and specificity value of 96.89%). CONCLUSIONS: According to the epidemiological characteristics of COVID-19 in Chongqing, a positive correlation between age and severity of COVID-19 was found, but no association between epidemiological history and disease severity was seen. Prediction model has a high sensitivity and is easy to use, which provides a strong basis for the early clinical evaluation on the severity of COVID-19. This article is protected by copyright. All rights reserved.