Acute respiratory distress syndrome (ARDS) significantly contributes to COVID-19 mortality, leading to respiratory failure and multi-organ dysfunction.1,2 Despite improved understanding of COVID-19 ARDS pathophysiology,3 survivors still worry about potential long-term respiratory issues, yet information is limited. Physiological differences, particularly in oxygenation, pulmonary mechanics, and treatment response, exist between COVID-19 ARDS and other ARDS variants.4 Therefore, prior insights into non-COVID-19 ARDS long-term outcomes5 may not be directly applicable to COVID-19 patients. Similarly, data on sequelae from severe COVID-19 pneumonia cannot be directly extrapolated to ARDS, since these represent distinct spectrums of COVID-19, characterized by notable differences in lung damage and pathogenic mechanisms.6

To date, disability assessment in survivors of COVID-19-related ARDS has predominantly relied on the six-minute walk test7–11 in studies characterized by small sample sizes, brief follow-up periods, and varying severity levels. Consequently, there is limited information available on the impact of COVID-19 ARDS on the diverse pathophysiological pathways implicated in exercise intolerance, necessitating a more comprehensive evaluation using a cardiopulmonary exercise test.12 Moreover, there is insufficient data on the determinants of medium-term disability and its long-term influence on residual lung function.

Hence, our objective was to investigate the extent of medium-term disability among survivors of severe COVID-19-associated ARDS, while delineating the pathophysiological changes contributing to their exercise intolerance and exploring its usefulness in estimating the risk of persistence of long-term functional impairment. Through a comprehensive assessment that incorporates clinical, radiological, and functional parameters, we also aimed to identify the primary risk factors associated with these changes and analyze their long-term impact on patients’ pulmonary function.

Figure 1 shows the patient recruitment flowchart, and Table 1, along with Tables S1 and S2 in the Additional file, summarize the main characteristics of the 108 patients finally selected.

Fig. 1.

Flow chart.

(0.14MB).

Clinical, morphological, and functional assessment at six months after ICU discharge

Frequency and characteristics of respiratory symptoms and health-related quality of life are presented in Fig. S1 and Table S3. Among the 100 patients with available CT scans six months post-ICU discharge, only 10 showed complete resolution of lung abnormalities (Table S4). Specifically, 10 patients exhibited unequivocal signs of established fibrosis, 73 presented fibrotic-like abnormalities and 7 displayed non-fibrotic abnormalities. Additionally, patchy perfusion defects were identified in 25 out of 70 patients with interpretable iodine maps (35.7%). Persistent established fibrosis or fibrotic-type abnormalities were associated with the number of previous comorbidities and nosocomial infection during ICU stay (Table S5).

Regarding conventional lung function (Table S6), restrictive impairment was the most prevalent functional disorder (57%), with impaired gas transfer observed in 30 patients (28%) (Fig. S2, Table S7). A detailed examination of gas exchange components through DLNO (Table S8) revealed membrane component involvement in almost all patients with low DLCO, which also remained reduced in a substantial percentage of patients with normal DLCO (Fig. 2). Lymphopenia at ICU admission emerged as the only independent risk factor associated with restrictive impairment, while the absence of obesity, hypothyroidism and days of invasive mechanical ventilation were independently associated with the presence of gas transfer impairment (Table S9). Additionally, male sex, age and the number of previous comorbidities were independent determinants of membrane component impairment (Table S9).

Fig. 2.

Frequency and distribution of lung function disorders.

(0.18MB).

Gas exchange impairment was associated with the presence of dyspnea (OR 2.53, 95%CI 1.07-6.01, P=0.035) or any respiratory symptom (OR 5.50, 95%CI 1.21-25.07, P=0.028), with dyspnea severity higher in patients with restrictive disorder, gas exchange impairment, or low membrane component involvement (Fig. S3). Furthermore, a CT pattern with unequivocal signs of established fibrosis was associated with restrictive impairment, whereas an increase in high attenuation volume was associated with both restrictive disorder and gas transfer impairment (Table S10).

The main finding of our study is the demonstration that nearly half of the survivors of severe ARDS secondary to COVID-19 exhibit moderate to severe functional disability in the medium term. This impairment is more common in women, individuals with a history of hypothyroidism, and in the presence of concurrent reductions in residual functional capacity, increased lung parenchyma attenuation, and decreased diffusion membrane component. Additionally, both progressive cardiorespiratory exercise testing and analysis of the diffusion membrane component provide additional value in predicting long-term functional deterioration among those survivors who, at six months post-ICU discharge, exhibit reduced TLC or DLCO values. Thus, both the diffusion membrane component and exercise tolerance, along with parameters related to dead space and gas exchange at peak exercise, are independently associated with the persistence of a restrictive disorder associated with gas exchange impairment at 24 months post-discharge, suggestive of lung fibrosis development.

Considering the notable prevalence of post-intensive care syndrome observed in the months after discharge from COVID-19-associated ARDS,24 coupled with the documented progressive enhancement in exercise tolerance throughout the convalescent phase,5 we opted to conduct the initial assessment six months post-discharge. This timeframe aligns with previous investigations assessing exercise tolerance via the walking test among survivors of COVID-19-related ARDS,25,26 reflecting a consensus in the field regarding the optimal timing for such evaluations.

The incidence of restrictive impairment in our patients was slightly higher than in other COVID-19 ARDS series (where lung volumes weren’t assessed by plethysmography),27–29 or ARDS of other origin.30 This might be due to the severity of our cohort, since restrictive impairment is more common in COVID-19 patients admitted to the ICU requiring invasive ventilation compared to those managed with high-flow nasal oxygen or non-invasive ventilation.9

Surprisingly, only 28% of our patients showed signs of gas exchange impairment, a notably lower rate compared to survivors of severe COVID-19 pneumonia31 or other COVID-19 ARDS.32 Previous studies often defined gas exchange impairment based on varying DLCO percentages, rather than its lower limit of normal (LLN). When LLN criteria were applied, studies indicated a prevalence of gas exchange impairment in COVID-19 severe pneumonia patients ranging from 17% to 34%.33,34 Our findings clarify that the impaired gas exchange observed in these patients primarily stems from alveolar-capillary membrane damage rather than vascular issues. Intriguingly, recent data from non-critical patients recovering from severe COVID-19 pneumonia revealed a correlation between reduced DLCO at 3- and 12-months post-discharge and membrane factor involvement.34,35

We also explore the dissociation between morphological and clinical-functional sequalae. Whereas a restrictive or gas exchange impairment was related with higher dyspnea levels, no differences were observed between respiratory symptoms and the existence or not of parenchymal alterations. Furthermore, unequivocal signs of established fibrosis were only associated with restrictive impairment, not reduced DLCO. These findings partly agree with previous studies showing no association between imaging abnormalities and reduced lung function or symptoms in severe COVID-19 cases.28,31 This observation could be explained by the limited extent of the lesions, typically not exceeding 25% of the lung parenchyma, as seen in other studies.29 Hence, quantitatively assessing lung parenchymal attenuation might provide a more sensitive measure of lung function, as shown by its correlation with both restriction and gas transfer issues in our patients.

Our study assessed survivors of severe COVID-19-related ARDS using a progressive exercise test, revealing that nearly half of them suffer from moderate to severe functional disability in the medium term. This distinguishes them from ARDS survivors of other causes who typically experience more severe and enduring declines in function, primarily due to persistent muscle and wasting.36 Notably, the functional limitations observed in our patients seem related to respiratory mechanics impairment, indicated by reduced tidal volume and less pronounced EELV reduction, along with impaired gas exchange shown by elevated carbon dioxide ventilatory equivalent and heightened symptomatic perception during exercise. Additionally, cardiac limitations may contribute, as indicated by differences in peak oxygen pulse and cardiovascular response to exercise.

Furthermore, it is also interesting to delineate the clinical and functional factors associated with decreased exercise tolerance. The presence of a restrictive disorder, characterized by reduced functional residual capacity, impaired gas exchange membrane component, and heightened lung parenchymal density are independently associated with moderate-to-severe disability. The contribution of hypothyroidism may extend beyond its effect on DLCO, encompassing its recognized influence on peripheral muscle function as well.37 Additionally, the association with female sex may be attributed to its established status as a risk factor for post-COVID-19 pulmonary fibrosis.38

In any case, in our opinion, the most significant finding of our study lies in the potential utility of progressive exercise testing and the pulmonary diffusion membrane component in identifying the risk of persistent functional sequelae of COVID-19-ARDS, which characterize the development of pulmonary fibrosis. In survivors of severe COVID-19, it has been described that the mid-term reduction in the most commonly assessed functional parameters (TLC and DLCO) may partially reverse in the long term,29 raises questions about the potential reversibility of post-COVID-19 pulmonary fibrosis mechanisms.39 In our patients, values at 6 months post-ICU discharge of the pulmonary diffusion membrane component, peak oxygen uptake, and two additional exercise response variables (dead space and gas exchange) are independent risk factors for identifying those patients with mid-term TLC or DLCO abnormalities who, at 24 months post-discharge, will exhibit a functional alteration pattern suggestive of pulmonary fibrosis, characterized by a combination of restrictive disorder and gas exchange impairment.

We acknowledge several limitations in our study. Firstly, it is a single-center, observational study. Second, like many other studies, we were unable to assess the premorbid clinical status, making it impossible to rule out the presence of impairments before admission. Third, our patient cohort was collected during two different periods of the pandemic, which could introduce variations in treatments for acute viral infection and differences in the viral variants responsible for lung pneumonia. However, all patients received similar post-COVID-19 management to ensure a consistent evaluation of residual sequelae. Fourth, follow-up assessment was only conducted in patients with functional disturbances 6 months after ICU discharge due to the ethical constraints on repeating procedures in normal subjects. Additionally, CT scans were not repeated during follow-up to avoid excessive radiation. Fifth, there was no control group because it was impossible to simultaneously recruit a similar group of patients with severe ARDS of another origin and given the limitations of using historic control subjects. Sixth, the still small number of studies assessing disability in survivors of ARDS secondary to COVID-19 using cardiopulmonary exercise testing has not allowed us to validate the predictive model described in an external cohort.

In conclusion, this study provides a comprehensive evaluation of the respiratory sequelae affecting survivors of severe ARDS secondary to COVID-19, assessing respiratory symptoms, quality of life, parenchymal and pulmonary circulation alterations, pulmonary function, and exercise tolerance, all in accordance with current standards. Our findings shed light on the extent of medium-term functional disability among survivors, potential underlying mechanisms and determinants, and its utility in identifying persistent functional alterations in the long term.

This study was supported by grants from Sociedad Madrileña de Neumología y Cirugía Torácica (NEUMOMADRID), Spain PE20/0001, to F. García-Río, Comunidad de Madrid, Spain SPACE 2-CV-COVID-CM, to F. García-Río and Instituto de Salud Carlos III-Fondos FEDER, Spain COV20/00207, to C. Cubillos-Zapata.

FG-R had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. JMA and FG-R conceived the study idea. RC, RG, JMA, CC-Z and FG-R contributed to the study conception and design. RC was the project manager. RC, RG, MT-V, SG-T, EA, ED-G, EM-C, MG-G, IT, MN-F, AF-V, MF-V, JMA, CC-Z and FG-R handled recruitment, data collection, deliverance of intervention, and assessment of all participants. FG-R, RC, MF-V and JMA managed the data and statistical analysis. FG-R, RC, RG and JMA were responsible for the study management and coordination. RC and FG-R drafted the first version of the manuscript. All authors contributed to interpretation of the data, revision of the manuscript, and approved the final manuscript and submission of the article.

The authors declare not to have any conflicts of interest that may be considered to influence directly or indirectly in the content of the manuscript.

All the individual patient data collected during the study will be shared. The data will be made available within 12 months of publication. All available data can be obtained by contacting the corresponding author (fgr01m@gmail.com). It will be necessary to provide a detailed protocol for the proposed study, to provide the approval of an ethics committee, to supply information about the funding and resources one has to carry out the study, and to consider inviting the original authors to participate in the re-analysis.

None of the materials have been partially or wholly produced with the aid of any artificial intelligence software or tool.