Introduction

In December 2019, a novel coronavirus (SARS-CoV-2) able to infect the respiratory tract in humans emerged in Wuhan (China), causing a disease known as COVID-19. A possible complication of SARS-CoV-2 infection is a severe acute respiratory syndrome (SARS) due to interstitial pneumonia [1]. On March 11, 2020, the WHO declared COVID-19 a global pandemic. As of June, 2021 more than 175 million people have been infected by SARS-CoV-2 worldwide and 3.8 have died [2].

Several studies reported a range of clinical and laboratory features among hospitalized COVID-19 patients, including increased levels of inflammatory markers [3]. The frequency of respiratory and functional impairment after COVID-19 is still debated but several studies found reduced lung volumes, reduced diffusing capacity of the lung for carbon monoxide (DLCO) and reduced exercise tolerance following hospital discharge [4,5,6,7]. A comprehensive follow-up strategy for COVID-19 patients after clinical recovery has been advocated [8]. We performed a study to investigate the prevalence of respiratory impairment in a cohort of COVID-19 patients after hospital discharge and to determine the relationship between the severity of pulmonary involvement during hospitalization and the extent of residual clinical and functional abnormalities.

Material and methods

Study population and subgroups

In the post-COVID-19 outpatient program at Fondazione Policlinico Universitario “A. Gemelli” IRCCS, Rome (Italy), a multidisciplinary team evaluates patients after hospital discharge. Patients presenting between April 22nd and May 27th, 2020 were invited to participate in the study. Inclusion criteria were: previous hospitalization for COVID-19; radiological evidence of interstitial pneumonia at the time of hospital admission; nasopharyngeal swab negative for SARS-Cov-2 in the 48–72 h before study enrolment.

Based on arterial blood gas (ABG) analysis values during hospitalizzation, three subgroups were defined using the worst PaO2/FiO2 value (p/F): mild (p/F ≥ 300), moderate (≤ 200 p/F < 300), and severe (p/F < 200). Such values, derived from the Berlin Criteria for ARDS, have been used in clinical practice to stratify severity of respiratory failure [9]. Written informed consent was obtained from each participant. The study protocol was approved by the Institutional Review Board of the Università Cattolica del Sacro Cuore (Rome, Italy) (approval number 0024185/20). Clinical evaluations, exams, and procedures were performed in accordance with the Declaration of Helsinki.

Study design and assessments

In this cross-sectional study, all patients underwent physical examination, resting ABG, pulmonary function tests (PFT) with DLCO, and 6MWT at the time of the study visit. ABG was analyzed using the ABL90FLEX radiometer (A. de Mori Spa Milano, Italy). The Biomedin Spirometer (software Baires version 5.1 revision 3, Biomedin SRL, Padova, Italy) was used to perform PFT and DLCO with the single breath-hold method (software Baires version 5.1 revision 3, Biomedin SRL, Padova, Italy). Lung function tests were performed according to current international guidelines [10, 11]. The 6MWT was used to assess the sub-maximal level of functional capacity. After 6 min of rest, the patient was instructed to walk along a 50 m corridor as fast as possible for 6 min wearing a finger/forehead pulse oximeter (Nonin 3100 Wristox pulse oximeter with nVISION software; Nonin Medical Inc, Plymouth, MN, USA) to record percutaneous oxygen saturation (SpO2) and heart rate (HR). At the end of the 6 min (or before, if the patient was unable to walk any further for fatigue, dyspnoea, or chest pain, or if saturation dropped below 80%) the distance covered was recorded and the patient was invited to sit and rest for 6 min. A drop in oxygen saturation ≥ 4% from baseline was considered to be clinically relevant During the study visit, a visual analog scale (VAS) score was used to measure the levels of dyspnoea and cough. Using a 100 mm linear scale, where 0 mm represents absence and 100 mm represents the worst dyspnoea and cough, patients were asked to report the levels of these two symptoms at the onset of the disease (i.e. immediately before hospital admission), during hospitalization, and at the moment of the study visit [12]. Retrospective data collected for this study included chest imaging findings, pharmacological treatments, p/F values, the type of respiratory support, and duration of hospitalization.

Statistical analysis

Descriptive statistics such as means with standard deviations or medians with interquartile ranges were used for continuous variables after checking for normal distribution of data. Frequencies or percentages were used to describe categorical variables. Dyspnea and cough VAS scores collected from the same patient were compared using analysis of variance (ANOVA) for repeated measures. Between-group comparisons of demographics and clinical data were performed using one-way analysis of covariance (ANCOVA) for continuous variables and the chi-square test for categorical variables. Estimated means of physiological variables across study groups were reported after adjustment by age, included as covariate in the ANCOVA model due to a significant age difference between study groups. P-values less than 0.05 were considered statistically significant. SPSS (version 24, IBM, New York, NY, USA) was used to perform statistical analyses.

Results

Characteristics of the study population

One hundred and fifty-seven patients included in the post-Covid follow-up program were screened for inclusion in the study. Twenty-six patients were excluded due to positivity at the nasopharyngeal SARS-Cov-2 swab; eighteen patients were excluded due to lack of radiological evidence of COVID-19 pneumonia at the time of hospitalization; twenty-seven patients were excluded because they had not been hospitalized (discharged from the Emergency Department). Eighty-six patients were included in the analyses.

The characteristics of the study population are reported in Table 1. Study visit occurred 35 (SD: ± 21) days after hospital discharge. At the time of hospital admission, chest imaging (i.e., chest X-ray or chest CT scan) revealed bilateral ground-glass opacities (GGO) with or without consolidations in 70 patients (81%). Sixteen patients (19%) had unilateral lung involvement. During hospitalization, 56 (65%) patients required supplemental oxygen and 15 patients (17%) were admitted to the intensive care unit.

Table 1 Characteristics of patients during the hospitalization for Covid-19

Pulmonary function testing (Fig. 1) showed overall preserved lung volumes, with mean percent predicted total lung capacity (TLC) of 89.6% (± 14.6%) and mean percent predicted forced vital capacity (FVC) of 104.6% (± 18.5%). Mean percent predicted forced expiratory volume in one second (FEV1) was 102.8% (± 16.0%). Mean percent predicted residual volume (RV) was the only respiratory volume reduced under the 5th percentile (74.8 ± 18.1%). Percent predicted DLco was also mildly reduced (77.2 ± 16.5%).

Fig. 1
figure 1

Overall cohort pulmonay function tests. Total Lung Capacity (TLC), Forced Vital Capacity (FVC), Forced Expiratory Volume in the first second (FEV-1), Residual Volume (RV) Diffusion Lung capacity for carbon monoxide (DLco)

Mean partial pressure of oxygen (pO2) was 91.4 mmHg (± 8.0) and mean alveolar-arterial oxygen gradient (d(A-a)) was 13.0 mmHg (± 7.5). Approximately one month after hospital discharge, patients reported more dyspnoea than pre-admission values (VAS score estimated mean difference: 15.3 mm; p < 0.001) (Fig. 2; Table 2).

Fig. 2
figure 2

Overall cohort trends of the VAS scores for dyspnoea and cough before hospitalization, during hospitalization, and at follow-up

Table 2 Dyspnoea and cough in overall population in three-time points: before hospitalization, during the hospitalization, and at study visit-time; VAS: Visual Analogic Scale

Comparison of study groups by p/F ratio

Most patients (n = 76, 88%) had an ABG performed during the hospitalization and were therefore included in this analysis. Among excluded patients, 8 patients had no ABG and 2 patients had an ABG performed with unknown oxygen inhaled fraction.

Gender, smoking status and comorbidities were not different across groups. Six patients (21%) in the mild hypoxemia group had a history of asthma. Patients in the severe hypoxemia group were older (63.1 years, p = 0.014 vs other groups), had a longer hospitalization time (23.0 days, p < 0.001 vs other groups) and were treated with anti-IL-6 drugs and enoxaparin more frequently (respectively 81% and 95%, p < 0.001 vs other groups).

Lung volumes were generally lower in the severe hypoxemia group, including lower percent of predicted FVC (p = 0.005), lower percent of predicted FEV1 (p = 0.009) and lower percent of predicted TLC (p = 0.012) (Table 3). In the severe hypoxemia group mean percent predicted TLC was 80.4% (± 3.1), indicating a residual restrictive impairment after 35 days from hospital discharge. DLco was also more reduced (64.9 ± 3.2% predicted) in the severe hypoxemia group than in the other two groups (p < 0.001).

Table 3 Characteristics of patients stratified by p/F during hospitalization

As expected, the alveolar-arterial oxygen gradient increased progressively across study groups, ranging from 10.1 mmHg (± 1.4) in the mild hypoxemia group to 16.6 mmHg (± 1.6) in the severe hypoxemia group (p = 0.011). Compared to patients in the severe hypoxemia group, patients in the mild hypoxemia group demonstrated greater exercise tolerance (+ 80.0 m in 6MWD; p = 0.004) and higher nadir in SpO2 (+ 2.5%; p = 0.005). Dyspnoea and cough levels at the time of study visit were similar across groups.

Discussion

The findings of this study suggest that respiratory abnormalities persist over time in COVID-19 patients who experienced a more severe form of disease during hospitalization. Several studies already reported a reduction in lung volumes and DLco levels as well as reduced exercise tolerance following hospital discharge [4,5,6,7]. Our study expands these findings in a larger Italian cohort. To our knowledge, this is the first study establishing the relationships between the severity of acute respiratory failure (as measured by the p/F ratio) and a wide range of blood gas and physiological parameters.

We identified a persistence of dyspnoea in the overall study population, a finding consistent with a study by Wong and coworkers, which reported dyspnoea in half of 78 COVID-19 patients after hospital discharge [13].

In order to explore the impact of disease severity on residual respiratory abnormalities, patients were stratified into three groups, according to levels of respiratory failure during hospitalization. No significant differences were observed regarding therapies, except for enoxaparin and anti-IL-6 drugs, administered more frequently in the severe group. The limited use of corticosteroids was likely due to the fact that evidence for dexamethasone use appeared towards the end of study completion [14]. We cannot exclude that a more extensive use of corticosteroids would have changed our findings.

Patients with mild and moderate disease had normal lung volumes. In contrast, a mild reduction in RV was found in the severe hypoxemia group. Whether this finding results by altered lung compliance in this group [15] remain to be determined. Moreover, TLC was at the lower limit of normal in the severe group: this finding suggests a link between severity of COVID-19 pneumonia and reduction in lung volumes. Whether such abnormalities were due to the presence of fibrotic sequelae after acute interstitial pneumonia could not be determined, since our cohort did not undergo a chest CT scan at the time of the study visit. Moreover, we identified normal DLco values in the mild and the moderate hypoxemia groups and reduced values in the severe hypoxemia group. This could reflect the degree of microvascular and epithelial damage, likely to be more consistent in the severe cases [16]. Patients recovering from ARDS from any cause may have persistent functional impairment one year after hospital discharge [17]. Therefore, these findings might not be COVID-19-specific.

Our study had several limitations. CT imaging was not available at the time of study visit: as such, the relationships between functional impairment and residual fibrotic changes remain unknown. The follow-up time in this study is short, and further studies are warranted to clarify whether respiratory abnormalities persist in the longer term. The use of p/F ratio to classify COVID-19 severity is not ideal as it may not be reliable in non-intubated patients [18]. Finally, the levels of dyspnoea and cough before and during hospitalization were collected at the time of the follow-up clinical evaluation: they may therefore not measure accurately the severity of symptoms at those timepoints.

Conclusion

Severe COVID-19 pneumonia may result in respiratory abnormalities and a reduction in exercise tolerance, which can be present at least one month after hospital discharge. Moreover, a low p/F ratio during the acute phase of the infection seems to correlate with a residual reduction of lung volumes, and residual reduction in DLCO. Further follow up is required to determine the degree of pulmonary and exercise impairment following hospitalization for COVID-19 pneumonia.