Results

Patient characteristics

In this study, 148 (58.0%) of 255 patients with viral pneumonia were male with a median age of 67.0 (54.0-81.0) years (Table 1). One hundred and eight (42.4%) patients had a smoking history. A total of 201 (85.5%) patients had one or more coexisting illnesses. Hypertension, cardiovascular diseases, diabetes, and COPD were the most common coexisting conditions. Cough, fever, and sputum production were the most common symptoms. Chest pain/tightness was reported in 24.3% of the patients. On admission, the leukocyte count was normal or slightly increased in most patients. Elevated levels of C-reactive protein (CRP) were observed in most patients. Hypoxemia (PaO2/FiO2 < 300 mmHg) was found in 97 (38.0%) of the patients. All patients underwent chest radiography on admission, 180 (70.6%), and 72 (28.2%) of patients had findings of multi-lobular infiltration and pleural effusion, respectively. Generally, 79 (31.0%) patients had a bacterial infection, and 9 (3.6%) had a fungal infection. Common complications included respiratory failure (68 [26.7%] of patients), followed by acute cardiac injury and acute kidney injury. There were 136 (57.9%) patients who had received antiviral medication, and 211 (89.8%) were administrated with antibiotic treatment. Additionally, 59 (25.1%) patients were given systemic corticosteroids. Eighty-eight (34.5%) patients were diagnosed with severe pneumonia. The median length of hospital stay was 13.0 (8.0-21.0) days, and the 120-day mortality was 11.0%.
Most baseline characteristics, mixed bacterial/fungal infections, complications, risk assessment, and clinical outcomes were comparable among influenza, non-influenza, and mixed viral pneumonia. Patients with influenza had the highest rate of antiviral use (111, 68.1%) and the shortest length of hospital stay (12, IQR 7-17 days) compared to the remainder. Other significant differences were summarized in Table 1.

Pathogens distribution

Low respiratory tract specimens for viral identification were obtained in 238 (93.3%) patients (117 for sputum, 53 for ETA, and 68 for BALF respectively). Among all viral patients, Flu A was the first ranking viral pathogen by the detection rate of 49% (126/255), followed by Flu B, 15%; SARS-CoV-2, 8%; HRV, 7%; PIV and CoV, 4%; AdV, 3%; RSV, 2%; HMPV, 1%. No bocavirus was detected. Mixed viral infections were observed in 18 (7%) cases (Figure 1). Approximately forty percent of patients had bacterial/fungal superinfection (ranging from 38%-50%), except for patients with HMPV and SARS-CoV-2 (5%). Notably, there were no significant differences in the rate of bacterial/fungal infection by the viral pathogen (χ2=11.490, P = 0.244).
For the 88 (34.5%) patients involving mixed bacterial/fungal infection, the diagnosis was confirmed with blood culture (in 6 cases), pleural fluid (3 cases), low respiratory specimens culture or PCR (91 cases). Especially, the detection rate of specific bacteria/fungi etiology were similar among influenza, non-influenza and mixed viral infection group (Table S1). A total of 129 strains of pathogens were isolated from 88 patients, and 27 cases of superinfection were caused by dual or more pathogens.P. aeruginosa (20.5%, 18/88) and K. pneumonia (19.3%, 17/88) were the most commonly detected pathogens, followed by A. baumannii (18.2%) and H. influenza (17.0%). Of note, fungal infections were documented 9 (10.2%) patients (Figure 2). Up to 75% (6/8) of S. aureus were methicillin-resistant, 53% (9/17) of K. pneumonia resistant to β-lactam antibiotics, including third-generation cephalosporins and carbapenems, 33% (6/18) of P. aeruginosa and 31% (5/11) ofA. baumannii resistant to carbapenem. Other details were shown in Figure 2. Furthermore, 58.1% (18/31) of antibiotic-resistant bacteria were documented from patients with ventilation, 83.9% (26/31) from patients with severe pneumonia, and 48.4% (15/31) from those who died in 120-days after admission.

Severe clinical outcome of bacterial/fungal superinfection

Clinical and laboratory parameters were compared between hospitalizations involving viral infection alone to those with mixed viral-bacterial/fungal infections. A battery of clinical features was profoundly different between the two groups as indicated in Table 2. Patients with mixed viral-bacterial/fungal infections had more males, higher rates of coexisting illness, especially for hypertension, liver disease, and cancer. These individuals also had a higher incidence of cough, dyspnea and rales, higher respiratory rate, worse laboratory findings, more common abnormalities on chest CT of multi-lobular infiltration and pleural effusion, and worse clinical admission characteristics. Immune examinations between two groups demonstrated that serum levels of T cells were decreased in the bacterial/fungal superinfection group (P = 0.003), and the subset of CD8+ T cells declined further (P = 0.005). There were no differences in antiviral or antibiotic treatments between patients with or without bacterial/fungal superinfection. However, patients with bacterial/fungal superinfection revealed a dramatic increase in the use of antifungal drugs, systematic corticosteroids and mechanical ventilation, incidence of complications, and length of hospital stay as compared with those who suffered viral infection alone (Table 3). The in-hospital mortality was strikingly higher in patients with mixed viral-bacterial/fungal infections than in those without (22.7% vs 2.4%, P < 0.001), especially when they came from non-ICU wards (non-ICU: 15.8% vs 1.3%, P < 0.001; ICU: 45.2% vs 17.6%, P = 0.057, Figure 3). The Kaplan−Meier survival curves for patients with and without bacterial/fungal superinfection showed that the incidence of death within 120 days after admission was significantly higher in patients with bacterial/fungal superinfection (HR = 9.708 P < 0.0001) than in those without.

Risk factors of bacterial/fungal superinfection

To explore the risk factors associated with superinfection, we initially evaluated each variable that displayed statistical significance with P < 0.05 in the difference between patients with and without superinfection using univariate analysis, and candidate variables were shown in Table 2. Considering the total number of patients with superinfection (N = 88 ) in our study and to avoid overfitting in the model, ten variables were chosen for further multivariable analysis based on previous findings and clinical constraints, namely PaO2/FiO2, BUN, leukocytosis, lymphocytopenia, ICU admission within 24 h after hospitalization, need for mechanical ventilation on admission, comorbidity, D-dimer, rales, and age. Finally, PaO2/FiO2 < 300 (OR: 2.570, 95% CI: 1.370-4.821, P = 0.003), BUN ≥ 7.1 mmol/L (OR: 4.016, 95% CI: 2.148-7.509, P < 0.001), leukocytosis (OR: 2.769, 95% CI: 1.335-5.741, P = 0.006) and lymphocytopenia (OR: 1.998, 95% CI: 1.086-3.675, P = 0.026) were independent risk factors of superinfection (Table 4).