Analysis of the clinical characteristics and early warning model construction of severe/critical coronavirus disease 2019 patients

Abstract

To analyze the clinical characteristics of critical patients with coronavirus disease 2019 (COVID-19), build an early warning model for severe/critical type, and aim at providing reference for the prediction of severe/critical COVID-19. The clinical data of COVID-19 patients treated in the Second People' Hospital of Fuyang City from January 20th to February 18th in 2020 were retrospective analyzed, including the demographic and epidemiological date, vital signs and hematology indexes, etc. on admission. Patients were divided into the normal type (set as normal group) and severe/critical type (set as severe group) according to the COVID-19 treatment plan classification standard published by National Health Commission of the People's Republic of China. The differences between two groups were compared, and the variables with statistical significance were incorporated in the multivariate binary unconditional Logistic regression analysis to screen the risk factors of severe/critical type. Risk factors were summarized to establish an early warning model, and the receiver operating characteristic (ROC) curve was carried out to evaluate the significance of the early warning model in the screening of critically COVID-19. A total of 155 patients with COVID-19 were admitted, including 125 patients of normal type and 30 patients of severe/critical type. (1) Compared with normal group, patients in severe group were older, and with higher proportion of basic diseases, higher body mass index (BMI), higher incidence of tachypnea, persistent high fever, peripheral blood oxygen saturation (SpO2) < 0.95, while the white blood cell count (WBC), CD4+T lymphocyte, CD8+T lymphocyte, lymphocyte count (LYM) were decreased obviously, the levels of interleukin-6 (IL-6), C-reactive protein (CRP) and serum amyloid a protein (SAA), and CT showed higher incidence of multi-pulmonary lobe lesions. There were no significant differences of gender, travel history from Wuhan, smoking history, shock index (SI) and CD4+/CD8+ ratio between the two groups. (2) Multivariate Logistic regression analysis showed that age ≥ 60 years old [odds ratio (OR) = 1.620, P = 0.031], combined with underlying diseases (OR = 1.521, P = 0.044), persistent high fever (OR = 2.469, P = 0.014), WBC < 2.0×109/L and/or LYM < 0.4×109/L (OR = 3.079, P = 0.006), pulmonary multilobar lesions (OR = 1.367, P = 0.047), and IL-6 ≥ 30 ng/L (OR = 2.426, P = 0.010) were the risk factors of severe/critical COVID-19. (3) The OR value corresponding to each risk factors were scored by rounding. Two points were scored for age ≥ 60 years old, with underlying diseases, persistent high fever and IL-6 ≥ 30 ng/L, 3 points for WBC < 2.0×109/L and/or LYM < 0.4×109/L, 1 point for pulmonary multilobar lesions, and totally calculated as early warning model scores. The early warning model score of the severe group was significantly higher than that of the normal group (9.33±2.79 vs. 5.04±2.38, t = 9.010, P = 0.001). (4) The ROC curve analysis showed the area under ROC curve (AUC) of early warning model on the early screening of severe/critical patients in COVID-19 was 0.944, and 95% confidence interval (95%CI) was 0.903-0.985; and the sensitivity and specificity were 93.3% and 72.0% respectively while the cut-off was 6.5. There are many differences between severe/critical and mild COVID-19 patients. The establishment of early warning model could help to screen severe/critical patients at an early stage, with certain significance for guiding treatment.