COVID-19 6 months after hospital discharge: pulmonary function impairment and its heterogeneity

To the Editor:

Until now, reports about pulmonary function in previously hospitalised subjects for coronavirus disease 2019 (COVID-19) are at discharge [1] or at 3–4 months [2–4]. The first study at 6 months is that of Huang et al. [5], which enrolled 1733 discharged subjects, 349 of whom underwent a pulmonary function study.

We consecutively enrolled from 15 March to 15 June 2020, during the first pandemic wave in Italy, 135 discharged COVID-19 patients, aged ≤80 years, in a follow-up study (Assessment of Cardiac and pulmonary consequences in patients recOvered from coviD-19 infection, the ACOD study) approved by the regional ethics committee (CER Liguria), aiming to collect data at 6 and 12 months after discharge from Hospitals (Santa Corona, Santa Maria di Misericordia, San Paolo) serving an area of 280 000 habitants.

This letter reports timely preliminary data on respiratory function at 6 months from discharge.

Written informed consent was collected from all subjects. Spirometry and pulmonary diffusion capacity tests were performed following the American Thoracic Society/European Respiratory Society statements [6, 7] with a Vyntus Body Plethysmograph (Vyaire Medical GmbH, Hoechberg, Germany). To minimise cross-infections, diffusing capacity of the lung for carbon monoxide (D_LCO) was measured by the single-breath method using the Diffusion SB RT Module for Body Vyntus (Vyaire Medical, GmbH). Abnormal data were that with a Z score >2sd (less than lower limit of normality (<LLN) or greater than upper limit of normality (>ULN)) by applying The Global Lung Function Initiative Network (GLI) reference values [8, 9]. Appropriate correction to D_LCO for haemoglobin was considered [7].

Descriptive statistics are reported as mean±sd. Differences between two groups were analysed for statistical significance by t-test (unpaired) and between more than two groups by ANOVA, Kruskal-Wallis or Chi-squared test, where appropriate. A two-sided p<0.05 was considered significant for all comparisons.

Table 1 summarises demographic and pulmonary function characteristics of the 135 enrolled subjects at follow-up for moderate-to-severe COVID-19, subdivided by the treatment for their acute respiratory failure (arterial oxygen tension <60 mmHg): 1) oxygen supplementation only, 2) continuous positive airway pressure (CPAP, by helmet), and 3) invasive mechanical ventilation (MV). No differences were found between the three groups, apart from age and gender.

At follow-up, impaired respiratory function was found in 64 (47%) of the enrolled subjects, characterised by an older age (62±11 versus 55±10 years; p<0.001) and a higher modified Medical Research Council (mMRC) dyspnoea scale (1.58±0.76 versus 0.36±0.48; p<0.01), without any differences in the ratio of males/females (44/20 versus 47/24) or in body mass index (BMI) index (28±5 versus 28±5 kg·m⁻²).

In 46 (34%) subjects D_LCO was impaired (61±14% of predicted), associated with reduced (62%) or normal (38%) D_LCO corrected for alveolar volume (K_CO, 81±15% of predicted).

Table 2 reports the pathophysiological classification [10] of the impairment found: pulmonary restriction (total lung capacity <LLN), isolated reduction of diffusing capacity of the lung for carbon monoxide (D_LCO <LLN), airway obstruction (forced expiratory volume in 1 s (FEV₁)/forced vital capacity (FVC) <LLN), and isolated air trapping (residual volume >ULN). No differences were found between groups in terms of age, male/female ratio, BMI or mMRC score. K_CO % of predicted was higher in the group with a restrictive pattern versus that with isolated D_LCO reduction (p=0.04).

At follow-up, D_LCO reduction was mainly associated with pulmonary restriction (53%), as expected, and less frequently with airway obstruction/airway trapping (8%), but also isolated (38%). In the latter case, subjects did not report pulmonary thromboembolism during hospitalisation or indirect signs of pulmonary hypertension at follow-up.

When comparing our data with that of Huang et al. [5], in moderate-to-severe COVID-19 (n=260), the impairment of respiratory function is similar among subjects requiring supplemental oxygen and less among those requiring CPAP or MV. Specifically, based on this subdivision, D_LCO was impaired in 31 and 40% of subjects (versus 29 and 56%), pulmonary restriction was present in 14 and 26% of subjects (versus 10 and 35%), and FEV₁/FVC <LLN in 8 and 10% of the subjects (versus FEV₁/FVC <0.7 in 8 and 2%). It is noteworthy that Huang et al. [5] did not use the Z score criterium, with a possible overestimation, and that the mean age of those who underwent spirometry in their study was unknown. At 8 months, using the Z score criterium, Barisione et al. [11] found fewer subjects with impaired D_LCO (20%)_, but in mild-to-severe COVID-19 and after having carefully excluded all subjects with comorbidities (including obesity) potentially affecting D_LCO.

As in the study of Mo et al. [1], which was carried out at hospital discharge, for about 29 out of 46 D_LCO-impaired patients, the K_CO was still within the normal range, which might indicate that the D_LCO decrease was more than the K_CO decrease, or, in other words, that lung volume is contributing to the gas exchange impairment. However, the relationship between alveolar volume (V_A), D_LCO or K_CO is complex and any interpretation a surmise. By using diffusing capacity of the lung for nitric oxide (D_LNO), Barisione et al. [11] suggest that a decreased alveolar membrane diffusive conductance (DM) is more frequent and persistent than the reduction of pulmonary capillary blood volume (V_C) in the recovery phase, at 8 months from discharge.

Regarding the restrictive pattern, in the study of Mo et al. [1], at discharge, it was interpreted as a consequence of a critical illness (due to a transient impairment in mechanical properties of the chest wall and respiratory muscles). In our study, after 6 months, is more suggestive of a change in the elastic properties of the lung.

Airway obstruction (n=11) or isolated air trapping (n=10) was present in 15% of the subjects at follow-up. Even subtracting known (n=4) or underdiagnosed (one current and three former smokers) COPD, a value of 9% among the discharged is still higher to the expected rate within a population of their age. Air trapping (an increase of residual volume (RV) and RV/TLC ratio) can be interpreted as an involvement of small airways not yet detected using conventional pulmonary function tests (i.e. FEV₁/FVC ratio). An inflammatory process in the small airways could contribute to airway closure, by interfering with surfactant activity, by increasing the volume of intraluminal material or, more consistent in our 6 months after discharge subjects, by airway remodelling [12].

Finally, Huang et al. [5] reported data on 89 subjects not requiring supplemental oxygen (mild pneumonia). As our subjects were all affected by acute respiratory failure, we have no data to compare. However, we found that among 20 subjects undergoing spirometry for exertional dyspnoea 6 months after SARS-CoV-2 infection recovered from at home, 12 presented an impaired respiratory function. Apart from four former smokers probably affected by underdiagnosed emphysema (TLC 120±9%, RV 169±30%, D_LCO 74%±2 of predicted) and one asthmatic subject (FEV₁/vital capacity 66%, FEV₁ 48% of predicted), the other seven were nonsmokers and previously healthy subjects reporting a reduced D_LCO (65±9% of predicted), associated with a restrictive pattern (TLC 67±8% of predicted) in three of them. These results are in line with those reported by Trinkmann et al. [13] on nonhospitalised subjects, but at 3 months [10] and after a first evaluation in the emergency department.

There are some limitations in our study. First, the lack of baseline pulmonary function data before COVID-19. However, patients with chronic respiratory disease were a minority, as were current or former smokers, and none of the subjects had a history of pulmonary fibrosis. Secondly, the association with computed tomography chest images were not analysed in this preliminary report.

In conclusion, our study reveals that after 6 months from discharge for moderate-to-severe COVID-19 about half of the enrolled subjects presented an impaired respiratory function and a significant exertional dyspnoea. Although it is tempting to speculate on the pathophysiology of the type of impairment found, our aim is to report timely to clinicians its entity and heterogeneity, consistent with the complex pathophysiology of COVID-19 [14]. Long-term follow-up (i.e. at 12 months) is required (ongoing) and research protocols with tools not yet routinely available (D_LNO) or for highly specialised centres (forced oscillatory technique) to be developed.

• Received March 18, 2021.
• Accepted May 17, 2021.

• Copyright ©The authors 2021