Logistic regression analysis of death risk factors of patients with severe and critical coronavirus disease 2019 and their predictive value

Abstract

To analyze the risk factors of death in patients with severe and critical coronavirus disease 2019 (COVID-19) and their predictive value. Using the clinical and epidemiological database of Yangtze River Shipping General Hospital in Wuhan, the clinical and epidemiological data of 105 patients with severe and critical COVID-19 from January to March in 2020 were collected. Multivariate unconditional Logistic regression method was used to analyze the death risk factors of patients during hospitalization. The receiver operating characteristic (ROC) curve was drawn according to the multivariate analysis results to construct a death prediction model; the prediction value of the model was analyzed. The 105 patients with severe and critical COVID-19 were enrolled with 66 males (62.9%) and 39 females (37.1%). The age was (58.2±14.4) years old. Forty-two patients died in hospital and 63 survived. Among the dead patients, 69.0% (29/42) were male, and 78.6% (33/42) were over 60 years old. Compared with survival patients, the non-survival patients were older (years old: 59.2±12.5 vs. 51.2±11.4), and had more comorbidities, including coronary heart disease, hypertension, myocardial damage and thrombocytopenia (coronary heart disease: 33.3% vs. 11.1%, hypertension: 28.6% vs. 9.5%, myocardial damage: 73.8% vs. 11.1%, thrombocytopenia: 61.9% vs. 14.3%), and received more mechanical ventilation (92.9% vs. 44.4%), with significant differences (all P < 0.01). The variables of gender, age, basic diseases, mechanical ventilation and complications were included in the unconditional Logistic regression analysis, which showed that gender [odds ratio (OR) = 2.852, 95% confidence interval (95%CI) was 0.122-66.694], age (OR = 3.257, 95%CI was 0.466-18.584), coronary heart disease (OR = 7.337, 95%CI was 0.227-87.021), hypertension (OR = 5.517, 95%CI was 0.258-65.024) and concurrent myocardial damage (OR = 7.322, 95%CI was 0.278-95.020) and thrombocytopenia (OR = 3.968, 95%CI was 0.325-35.549) were independent risk factors for death in patients with severe and critical COVID-19 during hospitalization. According to the risk factors, the death prediction model was constructed and ROC curve was analyzed, which showed that the area under ROC curve (AUC) of death prediction model for predicting the mortality of patients with severe and critical COVID-19 during hospitalization was 0.804, the sensitivity was 83.8%, and the specificity was 82.3%. Various risk factors are associated with the death of severe or critical COVID-19 patients, such as gender, age, basic diseases and complications. The death prediction model is constructed by gender, age, basic diseases with coronary heart disease and hypertension, concurrent myocardial damage and thrombocytopenia, which has certain predictive value for the death of patients with severe or critical COVID-19.