Is chest X-ray severity scoring for COVID-19 pneumonia reliable?

PURPOSE: To explore whether chest X-ray severity scoring (CX-SS) could be reliable to assess the severity of pulmonary parenchymal disease in COVID-19 patients. MATERIAL AND METHODS: The study consisted of 325 patients whose COVID-19 was confirmed by RT-PCR test and who underwent chest X-ray and computed tomography (CT) studies to assess parenchymal disease severity. Only 195 cases included in the final analysis after exclusion of cases with previous chest disease and cases having more than 24 hours interval between their X-ray and CT chest studies. Both chest X-ray and CT severity scores (CT-SS) were recorded by 2 experienced radiologists and were compared to the clinical severity. Interobserver agreement was assessed for CX-SS and CT-SS. RESULTS: In relation to the clinical severity, the sensitivity of the CX-SS for diagnosis of moderate to severe parenchymal disease was high (90.4% and 100%) and low for mild cases (66.2%), while the specificity was high for mild to moderate parenchymal disease (100%) compared to severe cases (86.7%). The sensitivity, specificity, and diagnostic accuracy of the CT-SS were higher than CX-SS. Pearson correlation coefficient demonstrated a strong positive correlation between CX-SS and CT-SS (rs = 0.88, p < 0.001). The inter-observer agreement for CX-SS was good (k = 0.79, p = 0.001), and it was excellent for CT-SS (k = 0.85, p = 0.001). CONCLUSIONS: CX-SS is reliable to assess the severity of COVID-19 pulmonary parenchymal disease, especially in moderate and severe cases, with the tendency of overestimation of severe cases.