Clinical characteristics of COVID-19 complicated with pleural effusion

This project was a retrospective single-center study, which included 476 COVID-19 patients hospitalized in Renmin Hospital of Wuhan University (Wuhan, Hubei province, China) from January 20, 2020 to March 23, 2020. Patients were excluded from the study if they met any of the following criteria: (1) age < 18 years; (2) relevant data were not available; (3) PE was caused by chronic heart failure, malignant tumors, tuberculosis and other infection diseases by clinical history, imaging examination or thoracentesis. Severe and critical cases were defined according to the guidelines of COVID-19 diagnosis and treatment plan (trial version 8) developed by the National Health Commission of China (http://​www.​nhc.​gov.​cn/​). The patients were categorized as follows: (1) general type: patients have fever, respiratory symptoms, and imaging findings of pneumonia. (2) severe type: patients have one of the following: (a) respiratory distress, respiratory rate ≥ 30 beats / min; (b) resting oxygen saturation ≤ 93%; (c) arterial blood oxygen partial pressure / oxygen concentration ≤ 300 mmHg (1 mmHg = 0.133 kPa). (3) critical type: cases have one of the following features: (a) respiratory failure and need for mechanical ventilation; (b) shock; (c) organ failure requiring intensive care. All 476 patients with COVID-19 were divided into two groups in terms of CT findings. Group 1 included 153 patients with PE while Group 2 had 323 patients without PE. Flowchart for patient selection is shown in Fig. 1. The study was conducted in accordance with the principles of the Declaration of Helsinki, and was approved by the ethics committee of Renmin Hospital of Wuhan University (NO. WDRY2020-K128). Written informed consent was obtained from each participating patient.

Chest CT scan was performed in all 476 patients with COVID-19 by employing an Optima CT680 scanner (GE Medical Systems, Milwaukee, WI), which set at 210 mA and 120 kV with a minimum slice thickness of 1 mm. All images were analyzed in a consistent manner by two experienced chest radiologists. Image analysis was based on the lesion features of each patient including: (a) lesion distribution, (b) number of involved lobes, (c) lobe of lesion distribution, (d) lesion patters(e.g., ground glass opacities, pulmonary consolidation, linear opacities) and (e) other findings (e.g., adjacent pleural thickening, pleural effusion, pericardial effusion, thoracic lymphadenopathy, pulmonary emphysema) [14]. The alterations caused by underlying lung diseases such as tuberculosis and lung cancer were excluded in this study.

The demographic data, clinical features (including medical history, comorbidities, signs and symptoms), laboratory findings and chest CT results were obtained through electronic admission records. Information on date of symptom onset, initial clinic visit, hospital admission, result of SARS-CoV-2 RNA detection, and the type of COVID-19 was also taken. The onset date was defined as the date when symptoms were noticed. Data were reviewed by a trained team of experienced physicians and independently analyzed by three researchers.

Nasopharyngeal and oropharyngeal swab specimens were collected and tested by fluorescence RT-PCR assay by using a SARS-CoV-2 RNA kit (Shanghai Geneodx Biotech Co. Ltd.), approved by National Medical Products Administration (NMPA) and recommended by Chinese Centers for Disease Control and Prevention (CDC) [15].

Baseline chest CT showed that 100 of 153 patients (65.36%) developed bilateral PE. Fifty of 153 patients (32.68%) exhibited only ground glass opacities (Fig. 2a). Five patients (3.27%) showed only pulmonary consolidation (Fig. 2b) and 7 (4.58%) presented with only linear opacities. Forty patients (26.14%) displayed ground glass opacities with pulmonary consolidation. Thirty-one patients (20.26%) had ground glass opacities with liner opacities. Only 6 patients (3.92%) showed pulmonary consolidation with liner opacities. Fourteen patients (9.15%) were found to have all three lesion patterns. As for other findings, 31 of 153 patients (20.26%) were complicated with pleural thickening (Fig. 2c); 12 (7.84%) with pericardial effusion (Fig. 2d), and 8 (5.23%) suffered from pulmonary emphysema. Lymphadenopathy was uncommon in this series (Table 1).

All 476 patients, including 153 patients with PE and 323 without PE, were included in this study (Table 2). There existed no significant difference in most underlying diseases between PE group and none-PE group, apart from chronic disease (P = 0.020) and other diseases (P < 0.001). Meanwhile, PE group had a higher incidence of fever (P = 0.012), cough (P < 0.0001), breath shortness (P = 0.014) and slower heart rate (P < 0.0001). According to their laboratory findings, patients with PE had higher levels of white blood cells (P = 0.026), neutrophils (P = 0.012), lactic dehydrogenase (LDH, P = 0.001), C-reactive protein (CRP, P < 0.001) and D-dimer (P < 0.0001), and lower levels of lymphocytes (P = 0.043), platelets (P = 0.001), hemoglobin (P = 0.022), partial pressure of oxygen (PO₂, P = 0.001) and oxygen saturation (SpO₂, P = 0.001). Moreover, patients with PE had higher incidence of severe or critical COVID-19 (P < 0.001) and longer hospital stay time (P < 0.0001). By the end of March 23, 10 patients in PE group died, while only 3 patients deceased in no PE group, suggesting that the mortality rate was statistically (P = 0.001) higher in patients with PE than in those without PE (Fig. 3).

A total of 48 variables were subjected to the LASSO regression analysis, and the results showed that PE, LDH, D-dimer and TBIL were significantly related to incidence of critical COVID-19 when the partial likelihood deviance was smallest (Fig. 4a and b). A model containing LDH, D-dimer, PE and TBIL, in the form of nomogram, was created to predict the progression to critical condition in COVID-19 patients (Fig. 5a). The risk predicted by the nomogram was virtually consistent with the actual outcomes, indicating the nomogram was well-calibrated (Fig. 5b). Then, decision curve analysis (DCA) was drawn to assess the clinical utility of the nomogram, and if it is clinically useful for the identification of patients who would progress to critical condition (Fig. 5c). Finally, ROC analysis was performed to evaluate the discriminative power of the nomogram. The nomogram exhibited good discriminative performance at AUC of 0.817 for predicting the progression to critical COVID-19 (Fig. 5d).