Predicting the risk of death in patients with covid-19 infection

Abstract

The article presents the associations of clinical, laboratory, and clinical and instrumental features of patients with COVID-19 with increased risks of death and survival. The final model for predicting the risks of developing a lethal outcome in COVID-19 was determined, which has high classification qualities (optimal threshold value of the calculated model is equal to -1.6149; sensitivity – 97.1%; and specificity – 82.6%. The purpose of our study was to determine the risks of developing fatal outcomes in patients with COVID-19 based on their clinical, laboratory and instrumental features. The study was performed at the Department of Infectious and Pediatric Infectious Diseases, Parasitology, Phthisiology and Pulmonology of the Kharkiv National Medical University in accordance with the current bioethical norms and rules. All patients signed informed consent. 179 patients with COVID-19 aged 20–88 years (average age was [58.75±13.82] years) were observed. Medical and statistical calculations were performed using the IBM SPSS 25.0 software package. The associations of indicators with the binomial dependent variable were calculated using multiple logistic regression analysis with the calcula-tion of β coefficients. The significance of differences was determined using the Mann-Whitney U-test with a threshold of statistical significance p=0.05. Based on the results, the final prog-nostic model of the risk of developing a lethal outcome of COVID-19 indicates an increased risk of death in COVID-19 with increasing age (by 13.9%), leukocyte count (by 14.4%), D-dimers (by 0.001%) on day 5–7. According to the model, an increase in the probability of sur-vival in COVID-19 was significantly proved with an increase in hemoglobin (by 6.1%) at the day of hospitalization, monocyte count (by 17.1%) on day 5–7 and the use of ceftriaxone (by 87.8%).