CT imaging features of 4121 patients with COVID-19: A meta-analysis.

ABSTRACT

OBJECTIVE: We systematically reviewed the computed tomography (CT) imaging features of coronavirus disease 2019 (COVID-19) to provide reference for clinical practice. METHODS: Our article comprehensively searched PubMed, FMRS, EMbase, CNKI, WanFang databases, and VIP databases to collect literatures about the CT imaging features of COVID-19 from 1 January to 16 March 2020. Three reviewers independently screened literature, extracted data, and assessed the risk of bias of included studies, and then, this meta-analysis was performed by using Stata12.0 software. RESULTS: A total of 34 retrospective studies involving a total of 4121 patients with COVID-19 were included. The results of the meta-analysis showed that most patients presented bilateral lung involvement (73.8%, 95% confidence interval [CI] 65.9%-81.1%) or multilobar involvement (67.3%, 95% CI 54.8%-78.7%) and just little patients showed normal CT findings (8.4%). We found that the most common changes in lesion density were ground-glass opacities (68.1%, 95% CI 56.9%-78.2%). Other changes in density included air bronchogram sign (44.7%), crazy-paving pattern (35.6%), and consolidation (32.0%). Patchy (40.3%), spider web sign (39.5%), cord-like (36.8%), and nodular (20.5%) were common lesion shapes in patients with COVID-19. Pleural thickening (27.1%) was found in some patients. Lymphadenopathy (5.4%) and pleural effusion (5.3%) were rare. CONCLUSION: The lung lesions of patients with COVID-19 were mostly bilateral lungs or multilobar involved. The most common chest CT findings were patchy and ground-glass opacities. Some patients had air bronchogram, spider web sign, and cord-like. Lymphadenopathy and pleural effusion were rare.