This prospective study was approved by the Institutional Review Board of the 305 Hospital of People Liberation Army. Informed consent was provided by all participants.

We prospectively enrolled 156 patients with COVID-19 pneumonia who had been admitted to the 305 Hospital of People Liberation Army between 29 November 2022 and 10 February 2023. All participants presented with symptoms such as fever, sore throat, and cough and were diagnosed of COVID-19 pneumonia by means of a SARS-CoV-2 positive polymerase chain reaction test via nasopharyngeal swabs, then underwent an initial chest CT with consent to confirm the pneumonia. All of these patients completed chest CT and pulmonary function tests at the 6-month and 1-year follow-ups. The exclusion criteria included: age less than 18 years or more than 80 years, reinfection with COVID-19 pneumonia or other lung disease during follow-up period, refusal to be followed up or (and) inability to be contacted, poor CT images quality and unavailable pulmonary function results (patients cannot cooperate well in examinations) (Figure 1).

Non-contrast chest scans were obtained with a 64-section multidetector CT scanner (LightSpeed VCT; GE Medical Systems) or a 128-section multidetector CT scanner (Brilliance iCT; Philips Healthcare), with participants in the supine position during a breath hold following full inspiration. The scanning parameters were 120 kV and adaptive tube current, with the smallest field of view possible according to the body habitus. Axial reconstructions were performed with a section thickness of 1 or 1.25 mm using a bone filter. All 156 patients underwent initial CT scans and follow-ups CT scans using the same parameters.

All chest CT images were independently assessed by three senior radiologists (Y.Z., Lin.L., W.L., with 19, 18, and 11 years of experience in thoracic radiology, respectively), who were blinded to the baseline and clinical information of the participants. Any disagreement was resolved by discussion and consensus. The readers assessed the CT features using axial images. Multiplane reconstruction was used to resolve any interpretive doubts. Images were interpreted at a window of 1,000–2,000 Hounsfield units and a level of −700 to −500 Hounsfield units, respectively, to assess the lung parenchyma.

CT features were described according to the Fleischner Society glossary (7) as follows: ground-glass opacities (GGO), consolidation, reticulation, linear atelectasis, traction bronchiectasis, parenchymal bands, honeycombing, acute respiratory distress syndrome (ARDS) pattern, crazy paving pattern, organizing pneumonia, and pleural effusion. The CT evidence of pulmonary fibrotic-like changes was defined as presence of linear atelectasis, traction bronchiectasis, parenchymal bands, and honeycombing (6–8).

The chest CT score was calculated per each of the five lung lobes based on the extent of parenchymal involvement (9), as follows: (0) no involvement; (1) <5% involvement; (2) 5%–25% involvement; (3) 26%–50% involvement; (4) 51%–75% involvement; and (5) >75% involvement. The resulting total CT score was the sum of each individual lobar score and ranged from 0 to 25.

The following parameters were measured: forced vital capacity (FVC), forced expiratory capacity at the first second of exhalation (FEV1), total lung capacity (TLC), and diffusion capacity of the lung for carbon monoxide (DLCO) measured by means of the single-breath test. All pulmonary function test measurements were expressed as percentages of predicted normal values. Pulmonary diffusion was regarded as abnormal when the DLCO < 80% of predicted value (10, 11).

The statistical analyses were performed using the software SPSS 25.0 (IBM Corp., Armonk, NY, USA). Continuous variables were expressed as medians with interquartile ranges (IQRs) or means ± standard deviations (SDs). Categorical variables were reported as numbers and percentages. Between groups, variables were compared using Chi-squared test (or Fisher’s exact test if any cell frequency ≤5) or independent samples t-test, as appropriate. The measurements repeated over time were compared using linear mixed-effects models. Multiple comparisons between two time points were adjusted using the Bonferroni method. Spearman correlation coefficient test was used for the correlation analyses between chest CT scores and with pulmonary function parameters at 1-year follow-up. P < 0.05 was considered to indicate a statistically significant difference.

Overall, the study group was composed of 156 participants with a mean age of 62 years ± 14 (SD), and 62 participants were women (39.7%). The baseline and clinical characteristics were summarized in Table 1. Of the 156 participants, the median body mass index was 21.2 kg/m² (IQR, 16.3–29), and 55 (35.3%) were smokers. Ninety-three participants (59.6%) had different types of comorbidities and hypertension (82 participants, 30.3%) was the most common comorbidity. Nine participants (5.8%) had a history of malignant tumor, but none had used pneumotoxic medications in the 3 years prior to infection. The median hospital stay of the participants was 13 days (IQR, 5–22 days), and they underwent 6-month follow-up at a median of 176 days (IQR, 154–203 days) and 1-year follow-up at a median of 363 days (IQR, 341–378 days). Fourteen participants (9.0%) had available chest CT in 1 year prior to infection. One hundred and fifty participants (96.2%) received COVID-19 vaccine before infection, of which 143 (91.7%) received two complete doses of the vaccine and 7 (4.5%) received only one dose. Forty-five participants (28.8%) requiring the highest level of ventilatory support in the form of noninvasive ventilation (32 participants, 20.5%) or invasive positive pressure ventilation (13 participants, 8.3%). Participants were treated with medications mainly including paxlovid (99 participants, 63.5%), azvudine (48 participants, 30.8%), and glucocorticoid (50 participants, 32.1%).

Over time, the proportion of participants who showed abnormalities on chest CT gradually decreased (initial, 100%, 156/156; 6-month, 57.1%, 89/156; and 1-year, 37.8%, 59/156; P < 0.001) (Table 2). Compared to the initial CT, the proportion of participants with GGO (initial, 100%, 156/156; 6-month, 36.5%, 57/156; and 1-year, 12.8%, 20/156; P < 0.001) and consolidation (initial, 49.3%, 77/156; 6-month, 10.3%, 16/156; and 1-year, 4.5%, 7/156; P < 0.001) gradually decreased (Figure 2). Meanwhile, participants with reticulation increased from 11 participants (7.0%) to 21 participants (13.5%) at 6-month follow-up and decreased to 18 participants (11.5%) (P < 0.001) at 1-year follow-up. Among CT evidence of fibrotic changes, the proportion of participants increased (initial, 6.4%, 10/156; 6-month, 14.1%, 22/156; and 1-year, 14.7%, 23/156; P < 0.001) (Figure 3). Compared with the initial CT, chest CT scores of any abnormality, GGO, and consolidation decreased (all P < 0.001), whereas fibrotic changes increased (P < 0.001) at two follow-up time points. Meanwhile, reticulation showed insignificantly change between three CT scans (P = 0.09). Chest CT scores of any abnormality (P = 0.02) and GGO (P = 0.001) showed significantly decrease between 6-month and 1-year follow-up (Table 3).

Compared with 6-month, the ventilation function indicators of FVC (P < 0.001), FEV1 (P < 0.001), and TLC (P = 0.002) were significantly improved at 1-year follow-up, but there was no significant improvement in FEV1/FVC (P = 0.57). Meanwhile, the proportion of patients with FEV1, FVC, and TLC less than 80% predicted were all significantly decreased (all P < 0.001) at 1-year follow-up, but there was no significant change in FEV1/FVC < 70% (P = 0.09). Compared with 6-month follow-up, DLCO did not improve significantly at 1-year follow-up (P = 0.19), but patients with DLCO less than 80% predicted were significantly decreased (P < 0.001) (Table 4).

At 1-year follow-up, in the subgroup of patients with or without any abnormalities on chest CT, there was no statistically significant difference in all ventilatory function indicators, but the DLCO showed a significant reduction in the subgroup of patients with abnormalities (P < 0.001). The proportion of participants with DLCO less than 80% predicted was greater in the subgroup with abnormalities than in the subgroup without abnormalities on chest CT. Similarly, compared to participants of the subgroup with nonfibrotic changes, DLCO (P = 0.01) and DLCO less than 80% predicted (P < 0.001) showed significantly decrease in participants of the subgroup with fibrotic changes (Table 5).

The extent of fibrotic changes was strongly correlated with lower DLCO (r = −0.734, P < 0.001) while the extent of reticulation was moderately correlated with lower DLCO (r = −0.661, P < 0.001). GGO extent moderately correlated with lower FVC (r = −0.577, P = 0.001), as was the extent of consolidation (r = −0.598, P = 0.001). There was no more significant correlation between the other pulmonary function indicators and chest CT scores of abnormalities (Table 6).