Prediction of mortality in adult COVID-19 patients using chest CT severity scoring systems: A comparative analysis of different scores

Abstract

Purpose: To compare the accuracy of mortality prediction of four CT severity scoring systems for COVID-19: CT severity score three levels, CT severity score, Total severity score, and Chest CT score. Methods: This was a retrospective study of 278 patients hospitalized with COVID-19 confirmed by a positive polymerase chain reaction (PCR) and in whom a CT scan was performed to assess the severity of lung involvement. This assessment was performed using four different scoring systems, including the CT severity score three levels, the CT severity score, the Total severity score, and the Chest CT score. Results: A total of 278 COVID-19 patients had chest CT scans, of whom 59 (21.22%) died and 219 (78.78%) survived. The ROC curves showed outstanding performance for the four chest CT severity scoring systems: 0.9580 for the CT severity score; 0.9532 for the CT severity score three levels; 0.9474 for the Total severity score; and 0.9327 for the Chest CT score. The comparison of these four ROC curves revealed no statistically significant difference between the four scoring systems (X2 = 3.89; p = 0.2740). Conclusion: The four chest CT severity scoring systems used predicted mortality in COVID-19 patients with excellent agreement and outstanding performance.