Clinical and radiological features of COVID-19 have been mainly described in adults. Literature states that children show similar but less pronounced chest CT findings comparing to those of adults . In another 20-patient pediatric cohort, Xia et al.  reported 20% of normal chest CTs, approximately the same as in this study (29%). In other paper, Wang et al.  stated that normal CT findings can be seen particularly in the early stages of the disease. Indeed, 4 of our 7 CTs with normal findings were performed during the first week of symptoms.
Similarly to other studies, typical findings in this cohort included mostly peripheral bilateral GGO opacities. On the other hand, unlike in the cohort from Xia et al. , where 50% of the patients showed the halo sign, this typical sign was only seen in 3 of this study’s cases (13%). In the present series, a perilobular consolidation surrounded by GGO was also found, suggesting secondary organizing pneumonia. Despite commonly seen in adults , this finding is seldom described in children [1, 9] and not mentioned in the International Consensus .
Amongst this study’s indeterminate cases, diffuse GGOs without peripheral or subpleural distribution were the predominant findings and differential diagnosis with other viral pneumonias, atypical bacterial pneumonias and other non-infectious entities was considered . Conversely, one case showed bilateral lower lobe consolidations, with minimum GGO, which may be found in some advanced cases  although CT was performed at day 6 of symptoms. Time course of the imaging abnormalities has been evaluated by Pan et al.  and Wang et al.  in adults. Isolated GGO is typically seen after symptom onset and the authors noticed that opacities tend to evolve not only in extension  but also to mixed patterns including crazy paving, consolidation, subsegmental athelectasis, and perilobular abnormalities, the latter suggesting secondary organizing pneumonia . These temporal changes are at least partially responsible for different CT sensitivities along the course of the disease, probably both in adult and in the pediatric population, which may explain some differences between cohorts.
Due to ethical concerns and ionizing radiation exposure risks, serial evaluation of children is not feasible. However, comparing CT findings from different patients in distinct phases of the disease, some authors extrapolated the same temporal pattern from adults to children. For instance, Foust et al. considered that consolidation is part of a later stage of the disease (the “developed phase”), probably evolving from other findings like the halo sign and GGO . Subacute and chronic lung sequalae are yet to be evaluated and are not under the scope of our study.
Besides the halo sign, two other findings have been more commonly reported in children than in the adult population: peribronchial distribution of the opacities and bronchial wall thickening . However, we have only reported peribronchial distribution and bronchial wall thickening in 3 patients (13%). Nevertheless, regarding bronchial thickening, it is not clear if it is only related to SARS-CoV-2 infection or to coinfections (as in one of the cases in this series where a bronchiolitis pattern was seen in a child with concomitant SARS-CoV-2, rhinovirus, and Mycoplasma pneumoniae infections). Interestingly, none of the current study patients exhibited a crazy paving pattern (that has also been described as fine mesh with GGO), reported in 20% of the 20-patient study from Xia et al. .
Comorbidities constituted a confounding factor while evaluating chest CT in patients with suspicion for COVID-19 pneumonia (including patients with bronchopulmonary dysplasia and asthma). Moreover, coinfection, which is particularly frequent in the COVID-19 pediatric population, further complicated our assessment. Xia and colleagues  reported a coinfection rate of 40%, similarly to this study’s data (42%).
Mosaic pattern attenuation was found in 4 of 24 rt-PCR-positive patients (17%); however, it was difficult to correlate this finding with COVID-19 pneumonia. In one of these patients, who also presented bronchial wall thickening on CT, RSV and adenovirus coinfections were identified, suggesting that the mosaic pattern might be related to small airway disease. The other 3 patients had history of prematurity, which might explain the mosaic pattern independently of SARS-CoV-2 infection.
Another confounding factor for diffuse ill-defined GGOs was related to motion artifacts due to lack of collaboration in non-sedated children. Additionally, subpleural opacities may be related to the sedation, potentially leading to passive athelectasic bands that often appear in the dependent lungs.
Some differences were found between this study cohort and other published series, probably reflecting scarce data and lack of consensual guidelines for CT acquisition, namely regarding the appropriate timing.
Besides recognizing typical findings for COVID-19, radiologists should be able to pinpoint those features that favor different or concomitant diagnosis. In this cohort, only 3 laboratory-confirmed cases were considered as atypical, according to the consensus by Foust et al. . With this in mind, we emphasize that normal CT does not exclude COVID-19 and an abnormal CT, even with some suggestive findings, should not be used to diagnose COVID-19 .
There are some limitations in this study, namely that it is a retrospective study, with a small number of patients, focusing on patients that performed CT upon confirmed COVID-19 infection. The referral criteria to perform CT were not clearly defined in the early days of this pandemic. Hence, clinical severity was not clearly evaluated in this paper. Also, a significant proportion of our patients had comorbidities and coinfections.