This prospective observational study is the second peer reviewed published study conducted in Germany and, to date, the largest to analyze the safety, feasibility, and efficacy of a three-week inpatient PR treatment in patients after COVID-19. All patients admitted due to COVID-19 between April 28th, 2020 and January 08th, 2021 were asked to participate in the study, of whom 90% participated. The baseline values of our patient population are broadly comparable to those of Glöckl et al. [43]. Thus, our patient cohort seems to be representative of post-COVID-19 PR patients in Germany.
Almost all assessed outcomes improved significantly in the whole group at the end of PR, with moderate to large effect sizes. While negative outcomes, such as dyspnea, depression, and fatigue, were alleviated, measurements of performance and wellbeing, such as lung capacity, exercise capacity, and QoL, increased. We detected large effect sizes (Cohen’s d [42] d > 0.8) for the intensity of exertional dyspnea (primary outcome), physical submaximal exercise capacity, QoL, fatigue, and depression. Regarding effect size, the two strongest improvements were found for the 6MWD (d = 1.36) and the VAS scale of the EQ-5D-5L questionnaire (d = 1.088), indicating that at the end of PR, patients achieved a significant improvement in physical performance and QoL. Furthermore, moderate effects (0.5 ≤ d < 0.8) were found for resting dyspnea and dyspnea in daily life in the last 7 days as measured by the mMRC. Cohen's d > 0.4 was found not only for important objectively measured lung function parameters (VC, FEV1, PaO2) but also for a decrease in anxiety. Significant but rather small effect sizes (0.2 < d < 0.4) were found for cough, sputum, pain, other lung function parameters (TLC, TLCO, PI max), and results of the laboratory blood tests (D-dimers, CRP).
There was a high burden of comorbidities. In only 2 cases, no secondary diagnoses were mentioned. This figure is higher than what is reported in the literature for the overall group of COVID-19 patients [44, 45] but is in line with cross-sectional studies in post-COVID-19 rehabilitation settings [46–48]. This difference could be explained by the fact that pre-existing comorbidities negatively influence the course of COVID-19, and patients with a more severe course are more likely to undergo post-COVID-19 rehabilitation. The most frequently mentioned comorbidities were cardiovascular diseases and obesity. This finding is in line with existing data [45]. In addition, orthopedic and neuromuscular comorbidities were frequent, and some ICU-associated, psychological, and internal diseases were observed. These comorbidities require the multimodal, multiprofessional, and interdisciplinary approach of PR. Previous studies regarding rehabilitation following COVID-19 have reported similar frequencies of comorbidities [48–51]. This finding supports the necessity of a multimodal and interdisciplinary therapeutic approach for post-COVID-19 rehabilitation.
Effectiveness of PR
Comparing the three subgroups, for some outcomes, stronger effects were found in groups A and B. For example, exertional dyspnea improved in groups A and B, with large effect sizes (d = 0.922 and d = 0.845, respectively). Group C (mild after interval) also improved significantly but with a rather low to moderate effect size (d = 0.485). Similar trends were found for some lung function parameters (VC, FEV1, TLC, PaO2). Furthermore, patients in group C rated both the efficacy of the PR and the personal improvement lower than the other two subgroups. In conclusion, these results indicate that patients with severe forms of COVID-19 show greater improvements over the course of PR, especially if the beginning of treatment occurs soon after the acute phase. These results are in line with the findings obtained by Al Chikhanie et al. [52]. However, it is important to emphasize that significant improvements were also found in group C, with some strong (6MWD, EQ-5D-5L-VAS) and moderate effect sizes (PHQ-9). Cohen's d values > 0.4 were also found in group C regarding the intensity of exertional dyspnea, dyspnea in daily life, and impairments in QoL. This indicates that despite a lower effect size, patients with an initially mild course of the disease still benefit from PR even after a long duration, as described by Glöckl et al.
Primary outcome: dyspnea
Exertional dyspnea was mentioned as the “most important symptom” by the majority of patients in all three groups (groups A, B, and C in 65%, 50%, and 40%, respectively). Regarding the intensity of exertional dyspnea, as measured by NRS, all 3 subgroups benefited significantly, with large (groups A and B) or moderate (group C) effect sizes. These improvements were in the same range as those in our recent study on PR for asthma patients [53]. However, the decrease in NRS scores was not associated with improvements in lung function or 6MWD, but there were significant correlations with the reduction in fatigue and especially anxiety (r = 0.348). Possibly, the self-confidence in one's own performance regained through training during PR and the resulting reduction in the fear of exertion contributes to a reduction in exertional dyspnea.
Although most lung function parameters (VC, TLC, FEV1, TLCO, PaO2) in group C were normal on average and were higher than those in groups A and B, group C scored their resting dyspnea higher at the start of PR and still scored higher at the end of PR. The more pronounced fatigue and the tendentially lower training effects of group C might have had an influence in this regard.
Regarding the mMRC dyspnea scale, which records shortness of breath in everyday life in the last week, there was a significant improvement, with a moderate effect size. This is worth noting because the mMRC dyspnea scale is considered to have low sensitivity to change. The improvement in the mMRC dyspnea scale is descriptively above the mean improvement seen in a large study on PR for COPD patients [53]. However, similar to the reduction in NRS values, the decrease in mMRC scores was not correlated with an increase in lung function but with a decrease in fatigue and anxiety.
Very few post-COVID-19 rehabilitation studies have reported results on dyspnea scores. A dyspnea assessment using the mMRC dyspnea scale was reported for a subgroup in the study by Glöckl et al.[43](N = 26 patients after severe to critical COVID-19). In this study, the mean mMRC at the start of PR was 2. This finding is comparable to the findings in groups A and B, as is the change after PR. Furthermore, the decrease in patients reporting an mMRC score of ≥ 2 is comparable (77–54% in Glöckl et al.; 79–54% in our study). In the early rehabilitation group of Curci et al. [48] (N = 41 still severely impaired patients, directly transferred from an ICU to the COVID-19 rehabilitation unit of "Policlinico San Marco" Hospital, Zingonia, Bergamo, Italy), which is comparable to our population only to a very limited extent, 90.2% of patients documented an mMRC level of 4. All of the patients improved by at least 1 level; the corresponding figures in our group A were 43.9% at rehabilitation onset, of whom 83.3% improved by at least 1 level at discharge. Thus, the changes in mMRC values we detected over the course of PR are comparable to those detected in other studies.
Physical capacity, other clinical symptoms, and other objectively measured parameters
A reduced exercise capacity compared to that in the healthy reference population was observed in all patients at T1. The largest deviations were found in group A. Such values, however, should be interpreted with caution, since there are various, sometimes conflicting, normative value calculations for the 6MWD of healthy individuals [30, 54–56]. With certain caveats, a walking distance of approximately 550 meters for women and 600 meters for men can be considered normal for our population. The observed 6MWD values are in the range of COPD rehabilitants in Germany [53, 57] and are slightly above the values for patients suffering from interstitial lung diseases [58]. However, the values may vary with the time interval from the acute phase of the disease and decrease with disease severity, as was shown for post-COVID-19 rehabilitants by Glöckl et al.[43]. This may explain the fact that the 6MWDs of group B and group C were higher than those reported for COVID-19 patients directly after an acute infection [46, 47, 59]. In all subgroups, the 6MWD improved, with large effect sizes, in groups A, B and C by 131 m (34.8%), 96 m (20.9%), and 74 m (15.4%), respectively. The changes in groups A and B are descriptively greater than the improvements achieved in our own PR studies concerning COPD or asthma patients [57, 60]. The improvement in patients in group C, who started PR a mean of 143 days after the acute phase and whose pulmonary function parameters were normal except for a reduced PImax value, was well above the MCID of patients with COPD [61–63] or idiopathic pulmonary fibrosis [64]. There were no significant correlations between the improvement in the 6MWD and lung function parameters except for a correlation with the improvement in PImax. Thus, we assume that the improved exercise capacity might not be a consequence of improved lung function but rather the result of adaptive mechanisms of the cardiovascular system and the musculature due to exercise training.
Lung function tests
The lung function pattern, which was more restrictive at baseline, improved in the overall group with a moderate effect size, with significant improvements in groups A and B. Similar lung function improvements were also seen in some other post-COVID-19 PR studies [43, 46, 52]. Maximal inspiratory pressure also improved in the total group, with a small effect size, with the most severe limitation and the smallest nonsignificant improvement seen in group C at T1. However, the changes in VC, TLC, and FEV1 did not correlate with improvements in exertional dyspnea or physical capacity.
Quality of life, fatigue, depression, and anxiety
In line with previous research [7, 10], we detected not only physical but also mental impairments and subjectively experienced impairments in health-related QoL in the sample at T1. In all subgroups, patients reported impairments in health-related QoL and symptoms of fatigue, depression, and anxiety. Over the course of PR, these parameters improved significantly, with mostly large effect sizes. Patients reported lower levels of impairment in health-related QoL and a decrease in fatigue, depression, and anxiety. In comparison to our own results for depression and anxiety in rehabilitants with asthma or COPD at the beginning and at the end of PR [60, 65], the results can be considered comparable. Interestingly, both measures of QoL were significantly correlated with a reduction in fatigue. Furthermore, the increase in the EQ-5D-5L-VAS showed a moderate correlation with the reduction in anxiety. However, there were no significant correlations between either of the two measures of QoL and the decrease in depression. Considering these numbers and the fact that one-third of the patients reported either sadness regarding their own health status or faintness/lack of energy as their most burdening symptom at T1, we assume that fatigue and anxiety are of particular importance in the subjectively experienced burden of disease regarding COVID-19. Noticeably, improvements in fatigue, anxiety, and QoL were the only variables that showed statistically significant correlations with reductions in exertional dyspnea. Even though these results align with each other, it is currently hard to fully interpret the findings because of a lack of studies that examine the role of anxiety and/or fatigue in PR following COVID-19. Therefore, we highly recommend a closer examination in this regard to further explain our findings.
Regarding the overall subjectively experienced current health status, our data revealed a significant improvement, with large effect sizes for all groups. Three-quarters of the patients achieved an improvement in the VAS score above the MCID of 8 points.
Given the significant reductions in depression and particularly anxiety in all the subgroups, we endorse the conclusion of Demeco et al. [66], who recommended post-COVID-19 rehabilitation programs not only for physical reasons but also for psychological reasons.
Regarding fatigue, we found a significant difference between group C and groups A and B. Group C reported the highest values of fatigue at T1 and achieved no significant changes over the course of PR. These results suggest that patients with initially rather mild courses of disease and leading fatigue symptoms may need more specific therapeutic approaches for the treatment of persisting fatigue symptoms. It is, however, important to point out the small sample size of 15 patients with fatigue scores for both measurements. Therefore, these results must be interpreted with adequate caution.
Safety and feasibility of PR in patients after COVID-19.
The rehabilitation program proved to be safe; in particular, no complications occurred, and all but two patients were able to complete PR. Moreover, the program proved to be feasible, since all patients were able to perform more than 90% of the prescribed therapies.