Chest CT with clinical factors helps distinguish COVID-19 from flu

Combining clinical variables with chest computed tomography (CT) could better differentiate the novel coronavirus disease (COVID-19) from influenza pneumonia, a recent study has shown.

The researchers retrospectively assessed 24 COVID-19 (mean age, 46.17±17.4 years; 13 men) and 79 influenza (mean age, 40.9±19.5 years; 50 men) pneumonia patients, looking at several chest CT features and clinical variables and calculating for their diagnostic sensitivity, specificity, and area under the curve (AUC). Univariate logistic analysis was used to screen potential distinguishing variables, subsequently confirmed through multivariate analysis.

Four clinical factors were used to distinguish influenza from COVID-19, including temperature, systolic blood pressure, cough, and sputum production. Together, these variables yielded an AUC of 0.819, as well as diagnostic sensitivity and specificity values of 0.783 and 0.747, respectively.

In terms of CT features, the researchers found nine factors with potential diagnostic value: central-peripheral, superior-inferior, and anterior-posterior distributions, patches of ground glass opacities (GGO), vascular enlargement in GGO, GGO nodules, interlobular septal thickening, air bronchogram, and bronchiectasis within focus.

All the chest CT features had diagnostic AUC of 0.927, sensitivity of 0.750, and specificity 0.962.

Multivariate logistic regression analysis was then performed to combine clinical factors and chest CT features. The combined model included six variables: systolic blood pressure, sputum production, vascular enlargement in GGO, GGO nodule, central-peripheral distribution, and bronchiectasis within focus.

The combined model also had better diagnostic performance, with AUC, sensitivity, and specificity values of 0.961, 0.87, and 0.97, respectively.