Objective: To create a prediction model of the risk of severe/critical disease in patients with Coronavirus disease (COVID-19). Methods: Clinical, laboratory, and lung computed tomography (CT) severity score were collected from patients admitted for COVID-19 pneumonia and considered as independent variables for the risk of severe/critical disease in a logistic regression analysis. The discriminative properties of the variables were analyzed through the area under the receiver operating characteristic curve analysis and included in a prediction model based on Fagan’s nomogram to calculate the post-test probability of severe/critical disease. All analyses were conducted using Medcalc (version 19.0, MedCalc Software, Ostend, Belgium). Results: One hundred seventy-one patients with COVID-19 pneumonia, including 37 severe/critical cases (21.6%) and 134 mild/moderate cases were evaluated. Among all the analyzed variables, Charlson Comorbidity Index (CCI) was that with the highest relative importance ( p = 0.0001), followed by CT severity score ( p = 0.0002), and age ( p = 0.0009). The optimal cut-off points for the predictive variables resulted: 3 for CCI [sensitivity 83.8%, specificity 69.6%, positive likelihood ratio (+LR) 2.76], 69.9 for age (sensitivity 94.6%, specificity 68.1, +LR 2.97), and 53 for CT severity score (sensitivity 64.9%, specificity 84.4%, +LR 4.17). Conclusion: The nomogram including CCI, age, and CT severity score, may be used to stratify patients with COVID-19 pneumonia.
【저자키워드】 COVID-19, Prediction model, age, Charlson Comorbidity Index, lung computed tomography, 【초록키워드】 Pneumonia, severity, lung, risk, Laboratory, Computed tomography, Probability, sensitivity, specificity, clinical, Patient, characteristic, Belgium, disease, Analysis, Predictive, Logistic regression analysis, independent variable, likelihood ratio, cut-off, positive, MedCalc, variable, highest, analyzed, collected, evaluated, conducted, calculate, CCI, Ostend, patients with COVID-19, variables, 【제목키워드】 SARS-CoV2, Model, development,