The Predictive Value of Risk Factors and Prognostic Scores in Hospitalized COVID-19 Patients

Abstract

Introduction: Risk stratification in patients with COVID-19 is a challenging task. Early warning scores (EWSs) are commonly used tools in the initial assessment of critical patients. However, their utility in patients with COVID-19 is still undetermined. Aim: This study aimed to discover the most valuable predictive model among existing EWSs for ICU admissions and mortality in COVID-19 patients. Materials and methods: This was a single-center cohort study that included 3608 COVID-19 patients admitted to the University Clinical Hospital Center Bezanijska Kosa, Belgrade, Serbia, between 23 June 2020, and 14 April 2021. Various demographic, laboratory, and clinical data were collected to calculate several EWSs and determine their efficacy. For all 3608 patients, five EWSs were calculated (MEWS, NEWS, NEWS2, REMS, and qSOFA). Model discrimination performance was tested using sensitivity, specificity, and positive and negative predictive values. C statistic, representing the area under the receiver operating characteristic (ROC) curve, was used for the overall assessment of the predictive model. Results: Among the evaluated prediction scores for 3068 patients with COVID-19, REMS demonstrated the highest diagnostic performance with the sensitivity, PPV, specificity, and NPV of 72.1%, 20.6%, 74.9%, and 96.8%, respectively. In the multivariate logistic regression analysis, aside from REMS, age (p < 0.001), higher CT score (p < 0.001), higher values of urea (p < 0.001), and the presence of bacterial superinfection (p < 0.001) were significant predictors of mortality. Conclusions: Among all evaluated EWSs to predict mortality and ICU admission in COVID-19 patients, the REMS score demonstrated the highest efficacy.