Computed Tomography and Clinical Features Differentiating Coronavirus Disease 2019 from Seasonal Influenza Pneumonia

Abstract Objectives: To investigate computed tomography (CT) and clinical features could help differentiate coronavirus disease 2019 (COVID-19) from seasonal influenza pneumonia. Methods: We retrospectively evaluated the clinical features and chest CT findings of Chinese patients with COVID-19 and seasonal influenza pneumonia treated during the same period. Results: The 24 patients with COVID-19 (mean age, 41 years; 13 men) and 79 patients with seasonal influenza pneumonia (mean age, 41 years; 50 men) differed significantly in mean temperature, respiratory rate, and systolic blood pressure; in central-peripheral, superior-inferior, and anterior-posterior distribution but not lateral distribution of pulmonary lesions; and patchy ground-glass opacity (GGO), GGO nodules, vascular enlargement in GGO, air bronchogram, bronchiolectasis in GGO or consolidation, interlobular septal thickening, and crazy-paving pattern. Separate regression models were developed with clinical features, CT features (including anatomical distributions), and a combined model informed by the first two. The combined model had the best diagnostic performance for identifying COVID-19: a cut-off value of 0.38 was 74% sensitive and 100% specific and had an area under the receiver operating characteristics curve of 0.94. This model was based on sputum production, vascular enlargement in GGO, and central-peripheral distribution (random vs subpleural). Conclusions: The combination of sputum production, vascular enlargement in GGO, and central-peripheral distribution should be extremely helpful in the differential diagnosis of COVID-19.