The severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) disease (COVID-19) pandemic represents a clinical and public health emergency and the national healthcare systems suffers from the high incidence of difficult-to-treat cases.1,2 While the infection can be asymptomatic, the disease can cause multi-organ dysfunction.2

Occurrence of acute respiratory failure is the most important cause for immediate hospitalization.2–4 Up to 20% of COVID-19 patients need intensive care unit (ICU) care, with 30%–100% treated with mechanical ventilation.5 Mortality of ICU patients ranged from 26% to 61.5%.6 Among critically ill patients, severe acute hypoxemic respiratory failure is the dominant finding, whereas hypercapnia is rare.

According to the Berlin definition,7 the severity of hypoxaemia defines the severity of acute respiratory distress syndrome (ARDS), based on ratio of arterial oxygen tension to fraction of inspired oxygen (PaO2/FiO2). Until now, the clinical features of COVID-19-related ARDS (CARDS) are still unclear.8 An important question is whether (or not) CARDS is a distinct form of ARDS that requires a different treatment strategy.

It seems that patients affected from COVID-19 respiratory failure meet criteria for moderate to severe ARDS.9,10 Baseline respiratory mechanics are not different in CARDS patients who eventually die from those who survive being extubated or remaining intubated.

After the acute phase, differences are observed suggesting differential trajectories of respiratory failure.11

However, the current knowledge on pulmonary pathogenesis and lung function impairment in the post-acute phase is still limited12,13 due to the recommendations on lung function tests during the pandemic phase published by the European Respiratory Society.14 In this regards a good alternative to gather information about the ability of the lung to exchange gases could be the arterial blood gases analysis (ABG).

Knowing PaO2, PaCO2 and FiO2, the alveolar-to-arterial oxygen (AaDO2) gradient can be calculated; AaDO2 gradient enables indeed a more precise evaluation of the pathophysiological basis of hypoxemia than the more widely used PaO2/FiO2 (P/F).15

Published data about gas exchange impairment of patients surviving the acute phase of COVID-19 are lacking.

Aim of this study is to assess the role of AaDO2 gradient and P/F in the post-acute phase of COVID-19 pneumonia.

COVID-19 survivors discharged from medical wards after a negative molecular test for SARS-CoV-2 were admitted to four clinical centres of the Istituti Clinici Scientifici Maugeri, Italy, and enrolled between April 1st and September 1st, 2020 to undergo clinical evaluation and multidisciplinary rehabilitation. The rehabilitation programme was implemented following the Italian position paper16: interventions were chosen considering age, clinical severity, length of immobilization, and comorbidities.16

The study was approved by the central ethical committee (CEC2279).

Clinical, radiological, and functional data were collected (Tables 1 and 2).

Quantitative variables were described with means (standard deviations, SD) or medians (Interquartile ranges, IQR) in case of parametric or non-parametric distribution, respectively. Absolute and relative (percentage) frequencies were used to describe qualitative variables. Student t or Mann–Whitney test was computed to assess differences for parametric and non-parametric quantitative variables. A p-value less than 0.05 was considered statistically significant. The statistical software STATA version 16 (StataCorp, Texas, USA) was used to perform all statistical computations.

One-hundred and forty-five consecutive patients were recruited. The mean (SD) age was 70 (10.8) years and 99 (68.3%) patients were male; 21 (14.5%) had preexisting pulmonary disease (1 TB sequelae, 13 COPD, 7 asthma) and 34 (45%) were current or former smokers. Overall, 59 (40.7%) CARDS patients were originally discharged from ICU and 86 patients from non-ICU departments; among patients from non-ICU departments 30 (34.9%) underwent both non-invasive ventilation (NIV) and supplemental oxygen, 47 (56.7%) supplemental oxygen only, and 9 (10.5%) did not receive any therapies (Table 1).

Blood gases analysis was performed at admission or discharge in 137 (94.5%) and 66 (45.5%) patients, respectively (Table 2); blood gases analysis was carried out both at admission and at discharge only in 58 (40%) patients (Table 2).

No statistically significant differences were observed for intrapulmonary gases exchanges (SaO2, PaO2, PaCO2, P/F, AaDO2) respiratory rate D-dimer between patients originally intubated when compared with those non-intubated during the acute COVID-19 phase (Table 2).

A statistically significant post-rehabilitation improvement was observed in 58 patients evaluated with ABG both at admission and at discharge, for the following parameters: AaDO2 (p: 0.004), D-dimer (p: 0.0003), respiratory (p: 0.003) and heart rate (p<0.0001) but not for P/F (Table 1). These findings are confirmed among the 33/58 patients not ICU-admitted (AaDO2 (p: 0.01), D-dimer (p: 0.0004), respiratory (p: 0.02) and heart rate (p<0.0001) and not on P/F), while among the 25/58 patients originally admitted at ICU the AaDO2, D-dimer and respiratory rate lose the statistical significance (Table 2).

Stratifying further these 58 patients by gender, age, length of stay (LoS) of hospital rehabilitation, BMI, smoking history and hypertension, AaDO2 retains statistical significance in males (p: 0.002), aged>70 years (p: 0.03), LoS<24 days (p: 0.002), obese (p: 0.007), smokers (p: 0.02) and those affected by hypertension (p: 0.002).

Our preliminary data on patients admitted for rehabilitation after recovery from COVID-19 suggest the following:

• a)

  The finding that intrapulmonary gases exchanges between originally intubated vs non-intubated patients during the acute COVID-19 phase do not differ significantly may suggest an atypical ARDS, although the effect of both the selection process and the small sample size cannot be excluded.

• b)

  The D-dimer as well presents no differences between the two groups of patients described above, potentially suggesting a multifactorial damage (alveolar damage, parenchymal damage and vascular damage): more damage would be expected among the previously intubated patients, likely to have suffered a more severe acute COVID-19 phase.

• c)

  In patients with ABG both at admission and discharge (n=58), a statistically significant improvement was observed at discharge for AaDO2 gradient and the same results were confirmed for 33/58 patients not ICU-admitted. By contrast, AaDO2 gradient lost the statistical significance among the 25/58 CARDS patients originally admitted at ICU. This might suggest that AaDO2 is more sensitive than P/F in the COVID-19 post-acute phase to monitor the lung damage in those not admitted to the ICU.

• d)

  Stratifying further these 58 patients AaDO2 gradient retains statistical significance in males, aged>70 years, LoS<24 days, obese, smokers and those affected by hypertension suggesting it may be a sensitive marker in severe patients.

• e)

  Alveolar-to arterial oxygen, which can be calculated knowing PaO2, PaCO2 and FiO2, can provide a more accurate evaluation of hypoxemia than P/F, because this could mirror changes in PaO2, FiO2 or both.12 More evidence is needed to understand the role of the AaDO2 gradient as a marker of lung function impairment. Case reports from post-mortem findings and biopsies showed mononuclear inflammation and frequently diffuse alveolar damage, with necrosis of alveolar epithelial cells, hyaline membrane formation, and hyperplasia of type II pneumocytes. In addition, consolidations due to fibroblastic proliferation with extracellular matrix and fibrin forming clusters in airspaces, as well as vascular damages, were described.8 All together alveolar, epithelial and vascular impairment could justify either ventilation-perfusion mismatch or intra-pulmonary shunting with an increase in AaDO2. In our study, the AaDO2 gradient might improve in the medium-term among the patients previously admitted at ICU (or remained unchanged due to irreversible damage of the lungs).

Additional studies are needed to ideally plan a longer follow-up ABG in monitoring the COVID-19 post-acute phase.