Presenting characteristics
The study population consisted of 108 hospitalized patients with confirmed COVID-19 (Table 1). There were 25 patients (23%) under 50 years-old, 83 (77%) over 50 years-old, the median age was 66 years (interquartile range [IQR]: 51 to 72; range: 23 to 86), and 60 patients (55.6%) were men. The median duration from initial symptoms to dyspnea was 5 days (IQR: 0 to 7) and from first symptoms to admission was 10 days (IQR: 6 to 15). Forty-eight patients (44%) had an underlying disease, and the most common underlying diseases were hypertension (n = 44, 40.7%), diabetes (n = 18, 16.7%), and cardiovascular disease (n = 16, 14.8%). The most common symptoms at the onset of illness were fever (n = 93, 86%), cough (n = 78, 72.2%), and sputum production (n = 44, 40.7%), followed by diarrhea, fatigue, abdominal pain, headache, and vomiting. Sixty-five patients (60.2%) had dyspnea. In addition, one patient deteriorated and received endotracheal intubation and extracorporeal membrane oxygenation (ECMO). However, she could not be brought back to life eventually (mortality rate (1/108, 0.93%).
Laboratory parameters
The white blood cell counts of patients on admission were above the reference range in 10 patients (9.3%) and below the reference range in 10 patients (Table 2). Fifty-six patients (52%) had decreased LY counts, 62 patients (57%) patients had elevated D-dimer levels, 30 patients (28%) had elevated serum creatinine levels, 46 patients (43%) had elevated creatine kinase-MB (CK-MB) levels, 33 patients (31%) patients had increased procalcitonin levels, 87 patients (81%) had elevated CRP levels, and 76 patients (70%) had elevated levels of ESR.
Chest CT features
Chest CT images at admission were taken for 84 of 108 patients, and all CT images indicated abnormalities (Table 3). The typical findings were bilateral ground-glass opacity (GGO), frosted glass density, and consolidation (Figure 1). GGO was present in 45 patients (53.6%), followed by frosted glass density (n = 38, 45.2%), and consolidation (n = 1, 0.93%). None of the patients had pleural effusion. Analysis of involved zones indicated that the base of the lung was most affected region (apex of diaphragm: n = 79, 94.1%), most patients had central and peripheral distribution (n = 81, 96.4%), and most patients had bilateral involvement (n = 78, 92.9%).
Chest CT scores
We divided the patients into three subgroups based on the chest CT score (0 to 100, 101 to 200, and above 200), and then compared the characteristics of these subgroups (Table 2). Analysis of blood cell counts indicated that the WBC, neutrophil, and lymphocyte counts were significantly lower in patients with higher CT scores. Patients in the three groups also had significant differences in D-dimer, blood urea nitrogen (BUN), lactate dehydrogenase, troponin, procalcitonin, ESR, CRP, and serum ferritin.
Comparison of the left and right lung layers indicated the right lung score was higher, but this was not statistically significant (Figure 2A). A comparison of CT scores of the two lungs at different areas indicated that the chest CT score significantly increased from the top to the bottom of the lungs (Figure 2B).
Association of chest CT score with inflammatory mediators
Correlation analysis indicated that chest CT score had significantly positive correlations with WBC count; neutrophil count; prothrombin time(PT); the levels of D-dimer, ESR, CRP, procalcitonin, serum ferritin, IL-6, and IL-10; and a negative correlation with LY count (Figure 3).
Chest CT score and laboratory parameters before and after treatment
After treatment, the chest CT score, and the levels of ESR, CRP, and procalcitonin were significantly lower than before treatment, but the LY count was significantly greater (Figure 4).
Association of glucocorticoid use with chest CT score and laboratory parameters
We also divided patients into a glucocorticoid group (n = 24, 22%) and a non-glucocorticoid group (n = 84, 78%) and compared their laboratory parameters and chest CT scores after treatment (Figure 5). The results indicated that glucocorticoid use had significant positive relationships with WBC count and LY count and significant negative relationships with the levels of NEU, CRP, activated partial thromboplastin time(APTT), and PT. Notably, the CT score declined more in users than non users of a glucocorticoid.
Correlations of changes in chest CT score with changes in other laboratory parameters
We further examined the correlation between the change in chest CT score from before to after treatment (ΔCT score) with changes in other laboratory parameters from before to after treatment (Table 4). The results indicated that ΔCT score had a positive correlation with ΔLY count and a negative correlation with ΔESR.