This prospective study was conducted by enrolling adult patients who needed mechanical ventilation and were admitted in the ICU due to COVID-19 for 6 months from July 1 to December 31, 2021. Among patients with radiologically confirmed pneumonia, those classified as ‘Severe’ or ‘Critical’ according to the World Health Organization clinical classification were included. Thirteen patients were recruited from a tertiary university hospital and two patients from a national public hospital. This study was approved by the ethical committee of the Seoul National University Bundang Hospital (approval number: B-2107-694-302; NMC-2021-06-083) and conducted in accordance with the amended Declaration of Helsinki. Written informed consent was obtained from all patients or their families. This research was supported by the Research of the Korea Centers for Disease Control and Prevention (grant number: 2021ER190400).

Ultrasound examinations were performed at three key time points to monitor the progression of the patient’s lung changes. It was performed immediately after intubation in the ICU (within 48 hours), at the time of transfer from the ICU to a general ward after tracheal extubation, and on an outpatient visit day 1 month after discharge. ICU discharge, which also marked the transfer to the general ward, was determined by the patient’s ability to transition to a nasal prong following tracheal extubation, with an oxygen requirement of less than 6 L and SpO₂ of 92% or more. Hospital discharge criteria included either no oxygen requirement or an oxygen requirement of less than 3 L that could be managed with home oxygen therapy. An outpatient visit was scheduled 30 days post-discharge for follow-up blood tests, imaging tests, and to assess the patient’s overall condition. Bedside LUS was performed by three sonographers with 2-10 years of experience in ultrasound. The machine used in the isolation ICU was a GE VENUE GO iq ultrasound (GE Healthcare, Seoul, Korea) equipped with a convex C1-5-RS probe, with a frequency of 3.5 MHz, a depth of 10 cm [7]. According to international evidence-based recommendations for point-of-care LUS in the emergency setting [8], a complete eight-zone LUS examination was performed with the patients in the supine position. The machine used for outpatient treatment after discharge was a dual-probe V-scan iq ultrasound (GE Healthcare), equipped with a linear probe, with a frequency range of 3.4 to 8 MHz and a depth of 8 cm to minimize inconsistencies in imaging views. The patient underwent LUS in 12 zones, including posterior imaging, while in a sitting or supine position (Figure 1). The captured images and videos were recorded and stored in the picture archiving and communication system (PACS) for subsequent offline analysis. Pleural thickness and the number of B-lines were recorded as actual numerical values in millimeters, which were considered for additional analysis to provide a comprehensive assessment of lung pathology.

In 12 pulmonary zones, typical lung findings were observed, and the number of B-lines, degree of pleural thickening, and presence or absence of consolidation that could suggest lung fibrosis were checked. The LUS scoring system used in our study was developed by our team to assess pulmonary fibrosis in postCOVID-19 acute respiratory distress syndrome (ARDS) patients. It combines key ultrasound indicators—pleural thickening, B-lines, and consolidation—based on existing literature [9-11]. The final score was obtained using the LUS scoring system. The maximum B-line was set at 2 points, the degree of pleural thickening at 2 points, and, when present, consolidation at 1 point. When 12 pulmonary zones were applied, 30 points on each left and right side were calculated as the maximum score (Figure 1, Supplementary Table S1, Supplementary Figure S1).

Due to logistical constraints, high-resolution computed tomography (HRCT) could not be performed at ICU admission or during quarantine. Instead, HRCT examinations were scheduled during routine outpatient visits, typically 1 month after discharge. These scans were conducted either the day before or on the same day as the outpatient visit, with the LUS examination performed immediately after the face-to-face consultation. Using the lung fibrosis texture-based automatic quantification analysis technique, we measured the extent of reticulation, honeycombing, consolidation, and ground-glass opacity on HRCT. This approach involves whole-lung segmentation and texture quantification to evaluate the presence of fibrotic and inflammatory features. Typically, fibrosis in interstitial lung disease includes reticular structures, honeycombing, sclerosis, and ground-glass opacities. The term ‘mixed disease pattern’ refers to the coexistence of multiple abnormalities, such as consolidation and ground-glass opacities, in the same patient, indicating overlapping inflammatory and fibrotic processes. In this study, fibrotic changes after severe COVID-19 pneumonia were not accompanied by honeycombing or reticulation. Therefore, we defined the HRCT fibrosis score as the sum of consolidation and ground-glass opacity, reflecting this mixed disease pattern [12-14]. The HRCT fibrosis score used in this study is calculated as the sum of ground-glass opacity and consolidation, measured using quantitative CT imaging. This scoring method is designed to assess the extent of post-COVID-19 fibrotic changes in the lungs. A higher fibrosis score reflects a greater degree of lung abnormalities, indicating more extensive fibrotic sequelae. All HRCT images were interpreted and scored by radiologists who were blinded to the clinical data, ensuring an objective and unbiased evaluation of the fibrosis extent.

Dichotomous variables are expressed as frequency (%) and continuous variables are expressed as median (interquartile range [IQR]). The relationship between the LUS and CT scores was studied using Pearson correlations. Statistical significance was set at p<0.05. A linear mixed model was used to analyze the trends in the average LUS score, B-line number, and pleural thickness over time. Ultrasound was performed in eight areas during the initial stage and in 12 areas at the time of leaving the ICU and at the time of discharge. However, to maintain consistency, the analysis was conducted using data from the eight areas that were consistently examined across all time points. This approach ensured a valid comparison of trends over time.

Clinical data were collected from 15 consecutive patients hospitalized for COVID-19 infection during the study period. Three patients were excluded due to refusal and two due to a ‘do not resuscitate’ status. The study group included 12 COVID-19 patients who underwent LUS evaluation (Figure 2). Table 1 shows the baseline characteristics and treatments administered to all study participants. The average patient age was 53±10 years, and 57% were men. The most common comorbidity was hypertension, followed by diabetes and heart disease. None of them had an underlying pulmonary disease, and the majority were non-smokers. The average PaO₂/FiO₂ ratio was 79.13, indicating relatively severe respiratory failure, and the average duration of mechanical ventilation was 8.5 days.

The LUS score was calculated by assigning points to the degree of B-lines and pleural thickness, along with the presence of consolidation. Due to the conflicting trends observed between B-lines and pleural thickness, the combined score for these parameters did not show significant variation. The sum of LUS scores across the eight examined zones was 19.1 (IQR, 13.0 to 26.5) at admission, 19.2 (IQR, 16.5 to 23.5) at discharge, and 17.5 (IQR, 15.5 to 19.25) in the outpatient setting, and there was no statistically significant difference (p=0.658). The HRCT fibrosis score did not show a statistically significant correlation with the actual pleural thickness (mm) measured using LUS (see Supplementary Tables S2, S3 for details).

Consequently, we conducted a trend analysis using the actual measured values, specifically the number of B-lines and the degree of pleural thickness in millimeters. This approach allowed for a more precise evaluation of changes in LUS findings. The mean values of B-lines for each lung area were 2.3±1.20 at admission, 0.45±0.19 at ICU discharge, and 0.15±0.23 at the outpatient visit approximately 1 month post-discharge. Similarly, the mean values of pleural thickness for each lung area were 2.05±0.19 at admission, 2.30±0.17 at ICU discharge, and 2.40±0.20 at the outpatient visit. The number of B-lines decreased significantly (confidence interval [CI], -1.35 to -0.62; p<0.001), and pleural thickness increased significantly (95% CI, 0.89 to 0.35; p=0.001). These measurements were taken to monitor the progression and resolution of lung abnormalities over time (Figure 3).