Chest computed tomography findings in hospitalized COVID-19 patients: a systematic review and meta-analysis

Most studies evaluating chest computed tomography (CT) features in coronavirus disease 2019 (COVID-19) have been small-sized and have presented varied findings. We aim to systematically review these studies and to conduct a meta-analysis of their results to provide a well-powered assessment of chest CT findings in patients with COVID-19. PubMed and EMBASE databases were systematically searched to identify published studies that evaluated chest CT findings in COVID-19 patients. Data regarding study characteristics and CT findings, including distribution of lesions, the lobe of lung involved, lesion densities, and radiological patterns, were extracted. Arcsine transformed proportions from individual studies were pooled using a random-effects model to derive pooled proportions (PPs) and 95% confidence intervals (CIs). A total of fifty-four studies (n=2693 confirmed COVID-19 patients) were included in the final review. Prevalence of different CT findings varied across studies; however, the most common findings were bilateral pulmonary involvement (PP: 74.1% [68.4%, 79.5%]; I2 = 85.76%), ground glass opacification (PP: 64.6% [57.6%, 71.4%]; I2 = 91.52%), involvement of the left lower lobe (PP: 71.2% [58.9%, 82.1%]; I2 = 90.91%), and subpleural distribution of lesions (PP: 57.2% [39.0%, 74.3%]; I2 = 93.08%). Multivariate meta-regression revealed a positive association between prevalence of air bronchograms and average age of the population (p=0.013). Bilateral ground glass opacification, a subpleural distribution of lesions, and involvement of the left lower lobe were the most notable chest CT findings in COVID-19 patients.