2.3 Data collection and analysis
We reviewed the clinical charts of all the pediatric patients for demographic information, symptoms, date of symptom onset, admission date, discharge date, dates and results of nucleic acid tests for COVID-19, and other identified concurrent infectious pathogens.
Two radiologists (WX and ZL with 12 and 18 years of experience, respectively) independently reviewed the chest CT images on PACS, only decision reached in consensus was reported. The initial CT images were stratified into one of two groups: normal or abnormal group. The CT images in the abnormal group were further assessed for imaging features including: (a) unilateral or bilateral distribution, (b) lobes involved, and (c) lesion characteristics. Lesion characteristics were subcategorized into: (a) ground-glass opacity, (b) consolidation, (c) nodule, and (d) thickened interlobular septa.
The involvement of lung was quantified according to a previously published paper, which had applied it in adults with COVID-1916. We divided each lung into upper zone (above the carina), middle zone, and lower zone (below the inferior pulmonary vein). And each zone was scored for percentage of lung involved on a scale of 0-4 (0 for 0% involvement; 1 for less than 25% involvement; 2 for 25% to less than 50% involvement; 3 for 50% to less than 75%; 4 for more than 75% involvement). Overall CT score of lung involvement was the summation of scores from all six lung zones.