Abstract
Introduction
The physiological basis for dyspnea, a hallmark of post-COVID syndrome (PCS), remains poorly understood.
Methods
In this analysis of the prospective, multicenter, population-based, longitudinal COVIDOM study, we studied 936 previously healthy adults assessed ≥6 months after a mostly mild, PCR-confirmed SARS-CoV-2 infection. Participants underwent comprehensive pulmonary function testing including spirometry, body plethysmography, diffusing capacity for carbon monoxide, and airwave oscillometry. Dyspnea was assessed by questionnaires (mMRC ≥1/MDP-A1 domain ≥1). We performed cross-sectional and longitudinal analyses for lung function in relation to both dyspnea and a previously defined PCS severity score (PCS-S).
Results
Between 11/2020 and 05/2023, we examined 936 previously healthy COVIDOM participants (median age 37 [IQR 28–51], 56% female). Dyspnea prevalence increased significantly with PCS severity (low PCS-S: 19.3%; intermediate PCS-S: 53.8%; high PCS-S: 81.8%; p < 0.001). Women suffered more frequently from dyspnea and PCS. Small airway dysfunction (SAD), as indicated by abnormal R5–20 Hz or AX5 Hz measures, tended to be more frequent in participants with high PCS severity and dyspnea compared to those with low PCS and no dyspnea (37% vs. 25%, p = 0.058) with corresponding R5–20 Hz of 0.03 [0.01–0.07] vs. 0.01 [0–0.03] kPa·L−1·s−1 (p < 0.01). Longitudinally, however, none of the baseline or follow-up lung function parameters, including measures of SAD, differed between participants with persistent dyspnea and those who became asymptomatic.
Conclusion
Oscillometry-derived R5–R20 Hz differed significantly between dyspneic PCS patients and controls. The high frequency of SAD and the absence of longitudinal improvement might indicate the potential clinical relevance of SAD assessment, despite its only numeric differences between PCS severity groups.
Keywords: Post-COVID syndrome, Small airway dysfunction, Epidemiology, SARS-CoV-2 infection, Dyspnea, Airway physiology
Introduction
Acute COVID-19 is characterized by various respiratory symptoms, including breathlessness, exercise intolerance, chest pain, and cough [1–3]. The persistence of these symptoms beyond 12 weeks and the absence of an alternative diagnosis define post-COVID syndrome (PCS) [4]. While risk factors for severe acute COVID-19 were identified early on during the pandemic, causes of the development and the persistence of PCS remain poorly understood [5].
In the population-based, prospective, multicenter COVIDOM study, we previously established a PCS severity score and identified baseline factors predicting a high PCS score, namely, severity of acute COVID-19 and personal resilience [1]. Based on these findings, we identified two different phenotypes of PCS and developed two novel predictor-specific PCS scores [6]. The resilience-specific PCS-R score comprises fatigue, neurological ailments, and sleep disturbances, while the acute severity-specific PCS-S score comprises exercise intolerance, joint or muscle pain, chemosensory deficits, persisting signs of infection, and also fatigue. In the COVIDOM cohort of >3,500 participants, the majority is of working age and only 5% experienced severe acute COVID-19 requiring hospital treatment [1]. In line with other reports, women suffered from severe PCS more frequently, but the influence of pre-existing diseases on the persistence of PCS symptoms in general, and respiratory symptoms in particular, has remained elusive. In particular, PCS in previously healthy individuals after a mild course of acute COVID-19 is a major health concern as it has negative secondary effects on working ability and functioning in everyday life.
Lung function testing is consistently recommended for the diagnostic evaluation of PCS patients with respiratory symptoms [7]. Population-based data from Denmark suggest an annual decline of forced expiratory volume in the first second (FEV1) and of forced vital capacity (FVC) that is accelerated after mild COVID-19 compared to uninfected individuals, even when taking pre-pandemic lung function status into account [8, 9]. However, other studies found no excessive lung function decline after mild-to-moderate COVID-19 [10, 11]. Although forced spirometry (FS) is a widely accepted measure in routine clinical practice, it might miss subtle changes in the peripheral small airways [12]. Therefore, more sensitive measures such as airwave oscillometry (AOS), body plethysmography (BP), and diffusing capacity of the lung for carbon monoxide (DLCO) appear more adequate to objectify respiratory symptoms in PCS patients. In a cohort study of mostly healthy young men that underwent pulmonary function testing as early as 11 days following mild or asymptomatic SARS-CoV-2 infection, comparison of pre- and postinfectious FS, BP, and DLCO showed no significant changes in lung function measures [13]. However, the study population of 98% males with a mean age of 35 years is not representative of the general population and the time point of lung function assessment within the acute phase of infection does not allow to draw conclusions on association with PCS.
We hence measured FS, AOS, BP, and DLCO in >3,400 COVIDOM participants with PCR-proven SARS-CoV-2 infection at least 6 months after mostly mild acute COVID-19. The analysis presented below will focus upon a specific subset of these individuals, namely, those without significant pre-existing chronic diseases. This exercise had two main objectives: first, to determine which lung function measurements correlate best with the respiratory symptoms of PCS, and second, to track the longitudinal course of these measurements.
Materials and Methods
COVIDOM is a prospective, population-based cohort study of participants with PCR-proven SARS-CoV-2 infection, recruited in three different regions in Germany: Kiel, Würzburg, and Berlin. The study was initiated in October 2020 as part of the National Pandemic Cohort Network (NAPKON-POP) to investigate the long-term health effects of COVID-19 in Germany. The main goal of COVIDOM was to establish a representative sample from the general German population. Invitations to participate in the COVIDOM study were sent out and managed by cooperating local healthcare authorities, who are legally obliged to document all SARS-CoV-2 infections in the district under their control (§7 Federal Infection Protection Law). A total of 3,664 participants underwent a detailed baseline onsite assessment at the respective site: Kiel: 2,556, Würzburg: 608, Berlin: 470. Study visits took place between November 15, 2020, and May 17, 2023, and were scheduled at least 6 months after the initial diagnosis of SARS-CoV-2 infection. Longitudinal assessments were scheduled annually after the initial visit, requiring a second study visit that was scheduled roughly 18–24 months postinfection. Based on a previously developed PCS severity score [1], all participants with severe PCS were invited for onsite follow-up. While a matched subgroup of participants with low or intermediate PCS severity was also invited for onsite follow-up, all other COVIDOM participants were subjected to a detailed online/telephone follow-up questionnaire. Details on the study procedures have been published previously [14].
Ethic Committee
The COVIDOM study has been conducted in accordance with current guidelines and legal requirements. The study was approved by the Local Ethics Committees of the universities of Kiel (No. D 537/20) and Würzburg (No. 236/20_z). According to the professional code of the Berlin Medical Association, approval by the Kiel Ethics Committee was also valid for the Berlin study site. All participants provided written informed consent prior to their inclusion.
Symptom Assessment and Medical History
The COVIDOM study procedures included a detailed patient history, structured interview, and patient-reported outcome measures as well as an in-depth clinical examination and biomaterial collection [1, 14]. Respiratory symptoms have been assessed through the modified Medical Research Council (mMRC) dyspnea score [15] and the Multidimensional Dyspnea Profile (MDP) [16]. For the current analyses, dyspnea is defined as an mMRC score ≥1 or as the A1 domain of the MDP (i.e., general unpleasantness of breathing) scoring ≥1. In addition to mMRC and MDP, dyspnea was also assessed by way of the PROMIS dyspnea questionnaire in individuals with an MDP score ≥1 [17]. In addition to the A1 domain (general discomfort of breathing), the MDP consists of a sensory domain, addressing the presence and intensity of (I) physical breathing effort, (II) air hunger, (III) tightness, (IV) mental breathing effort, and (IV) hyperpnea. The affective domain investigates emotions that might be associated with the presence of breathing difficulties, namely, (I) depression, (II) anxiousness, (III) anger, (IV) frustration, and (V) fear. Health-related quality of life has been assessed by EQ-5D-5L. Medical history including potential existence of pre-existing diseases was assessed in a structured interview. Most of the interview could be completed online prior to the study visit. Data on medication and diseases were checked by a study physician during the onsite visit.
Post-COVID Syndrome Score (PCS Score)
In a previous analysis of the COVIDOM study, we defined a novel PCS severity score that was based on the presence or absence of 12 binary symptom indicators [1]. The score was developed using k-means clustering, resulting in three distinct hierarchical groups, with higher groups displaying more or more important symptoms. PCS scores from 0 to 10.75 were defined as absent or mild PCS, PCS scores from >10.75 to 26.25 were defined as moderate PCS, and a PCS score >26.25 was defined as severe PCS (for more details, see [1]).
Exclusion of Participants with Chronic Diseases
For the current analyses, we excluded all participants with relevant pre-existing chronic disease that might impact everyday wellbeing. In particular, we excluded any cardiovascular, respiratory, neurologic, psychiatric, rheumatologic, renal, or haemato-oncologic diseases, as well as HIV, tuberculosis, and individuals following organ transplant. The only exception was hay fever because this is mostly a mild and intermittent condition. By excluding participants as described, we deliberately excluded one phenotype of PCS patients, namely, that with worsening of pre-existing chronic disease [18]. We chose this focus on previously healthy participants to eliminate potential effects of pre-existing diseases. Previously, we found that neurological and psychiatric diseases, in particular, could be possible risk factors for a certain type of PCS syndrome [1, 6].
Lung Function
Comprehensive lung function assessment, including FS, BP, AOS, and DLCO, was performed according to current standards of the European Respiratory Society (ERS). Lung function impairment was defined using the “lower and upper limits of normal” (LLN, ULN) of the Global Lung Function Initiative (GLI) reference values for spirometry, plethysmography, and gas uptake [19–21]. Normal oscillometry reference values and z-scores were defined according to Ostveen et al. [22]. Oscillometry measures were reported as z-scores, if applicable.
Definition of lung function impairment was based on the following measures: FEV1, FVC (both from FS); residual volume (RV), total lung capacity (TLC), and their ratio (RV/TLC) (all from BP); DLCO, DLCO/alveolar volume (i.e., KCO) (all from DLCO); respiratory resistance at 5 Hz (R5 Hz), difference between R5 Hz and R20 Hz (R5–20 Hz), respiratory reactance at 5 Hz (X5 Hz) (all from AOS). The following criteria were used to define specific lung function impairments:
Airflow obstruction: FEV1/FVC < LLN
Hyperinflation: RV/TLC > ULN and RV >ULN
Restriction: TLC < LLN
Impaired diffusion: DLCO <LLN or KCO < LLN
And small airway dysfunction (SAD): area of reactance (AX) > ULN or R5–20 Hz >0.07 kPa·L⁻¹·s⁻¹ [23, 24]
If possible, lung function measures were reported as z-scores to account for age, sex, and height. Absolute measures were reported only for R5–20 Hz because there are no valid reference equations for z-score calculation available for this parameter.
Values differing by more than three interquartile ranges from the upper and lower quartile were excluded as outliers. If the R5–R20 Hz difference resulted in a negative value, the respective value was set to zero for further analyses.
Statistics
All statistical analyses were carried out with R version 4.3.3 (2024-02-29) on the x86_64-pc-linux-gnu (64 bit) platform. Fisher’s exact test was used to compare categorical variables between different groups. Continuous variables were analyzed with either non-parametric or parametric methods, depending on their distribution. For non-normally distributed data, a Wilcoxon rank-sum test was used for two-group comparisons and a Kruskal-Wallis test for comparisons involving more than two groups. When data followed a normal distribution, comparisons between two groups were performed by way of a t test or a one-way ANOVA, for more than two groups. Longitudinal assessments of within-subject changes were based on paired t tests.
Since the present study was exploratory, no primary outcome was defined in advance and no adjustments for multiple testing were made. Differences in lung function and other health-related variables were assessed between predefined subgroups of previously healthy participants. These analyses included comparisons based on PCS score categories, between individuals with low PCS score and no dyspnea (as defined above) and those with high PCS score and manifest dyspnea, and between participants exhibiting different patterns of lung function impairment. Within-subject analyses were conducted to compare baseline and first follow-up measurements for participants for whom data were available from both time points.
Results
The initial dataset included 2,379 COVIDOM participants from Kiel, 545 from Würzburg, and 446 from Berlin, for a total of 3,370 participants. After excluding participants with pre-existing chronic disease or with implausible lung function data, the final cohort included 936 participants (Fig. 1). Detailed clinical and lung function information on participants with pre-existing chronic diseases are provided in online supplementary Table 1 (for all online suppl. material, see https://doi.org/10.1159/000549966).
Fig. 1.
Stepwise exclusion of participants with chronic disease pre-existing prior to SARS-CoV-2 infection.
Within the subcohort of previously healthy participants, the PCS score was low for 449 (47.9%), intermediate for 389 (41.6%), and high for 98 (10.5%) individuals. Most SARS-CoV-2 infections occurred in 2020, i.e., before the widespread introduction of COVID-19 vaccination programs. Since the Omicron variant first spread in Germany in January 2022, this variant is unlikely to be present in the cohort.
Basic demographic and lung function measures, stratified by PCS score group, are summarized in Table 1. Participants with a high PCS score were characterized by older age, higher BMI, a greater proportion of female sex, and a higher frequency of dyspnea compared to those with a low PCS score (Table 1). The overlap of various lung function impairments is illustrated in Figure 2, both for the entire cohort and by PCS score group. Oscillometry-defined SAD was the most frequent single lung function impairment across PCS score groups. Notably, the various impairments observed, including SAD, airflow obstruction, hyperinflation, restriction, and impaired diffusion, showed minimal overlap with one another (Fig. 2). The distribution of lung function impairments did not vary markedly with PCS severity (Fig. 2; Table 1). The only lung function parameters that differed significantly between PCS score groups were FRC, TLC, and R5–R20 Hz. Of note, mean values of these parameters were between the upper and lower limits of normal for all PCS score groups, and the proportion of patients with restriction (TLC < LLN) or SAD (AX > ULN or R5–20 Hz >0.07 kPa·L⁻¹·s⁻¹) was similar across groups (Table 1).
Table 1.
Basic characteristics and frequency of lung function impairments in different PCS score groups
| | PCS low (≤10.75), N = 449 (47.9%) | PCS intermediate (>10.75 and ≤26.25), N = 389 (41.6%) | PCS high (>26.25), N = 98 (10.5%) | p value |
|---|---|---|---|---|
| Age, median [IQR], years | 35 [28; 49] | 38 [29; 50] | 44 [30; 54] | 0.009 |
| Female/male ratio, n (%) | 218 (48.6)/231 (51.5) | 241 (62.0)/148 (38.1) | 67 (68.4)/31 (31.6) | <0.001 |
| BMI, mean (SD), kg/m2 | 25.2 (±4.3) | 26.0 (±5.0) | 26.4 (±5.6) | 0.104 |
| Current smoking, n (%) | 72 (16.7) | 54 (14.5) | 12 (12.5) | 0.525 |
| PCS score, mean (SD) | 2.8 (±3.4) | 18.7 (±4.3) | 34.3 (±6.3) | <0.001 |
| MDP-A1 global dyspnea score, median [IQR] | 0 [0; 0] | 0 [0; 2] | 2 [1; 4] | <0.001 |
| mMRC dyspnea score, median [IQR] | 0 [0; 0] | 0 [0; 0.5] | 1 [0; 1] | <0.001 |
| No dyspnea | 361 (94.75) | 254 (74.93) | 44 (48.89) | |
| mMRC score 1 | 18 (4.72) | 82 (24.19) | 30 (33.33) | |
| mMRC score 2 | 2 (0.52) | 3 (0.88) | 14 (15.56) | |
| mMRC score 3 | 0 (0) | 0 (0) | 1 (1.11) | |
| mMRC score 4 | 0 (0) | 0 (0) | 1 (1.11) | |
| Dyspnea (MDP-A1 >1 and mMRC >0), n (%) | 72 (19.3) | 185 (53.8) | 72 (81.8) | <0.001 |
| Year of infection, n (%) | | | | 0.363 |
| 2020 | 328 (74.) | 264 (68.6) | 66 (71.0) | |
| 2021 | 110 (24.8) | 118 (30.7) | 27 (29.0) | |
| 2022 | 5 (1.1) | 3 (0.8) | 0 | |
| Vaccination prior to infection, n (%) | 3 (0.8) | 9 (2.7) | 2 (2.2) | 0.107 |
| FEV1/FVC (z-score), n (%) | −0.36 (1.05) | −0.28 (0.92) | −0.25 (1.04) | 0.166 |
| Airflow obstruction, n (%) | 45 (10.0) | 22 (5.8) | 9 (9.7) | 0.065 |
| FRC (z-score) | 0.28 (±0.93) | 0.09 (±0.96) | 0.06 (±0.95) | 0.003 |
| RV (z-score) | 0.55 (±1) | 0.43 (±0.98) | 0.48 (0.91) | 0.182 |
| TLC (z-score) | −0,27 (±1.09) | −0,38 (±0.97) | −0.52 (±1.01) | 0.044 |
| RV/TLC (z-score) | 0.92 (±1.14) | 0.83 (±1.13) | 0.99 (±1.05) | 0.930 |
| Restriction, n (%) | 44 (10.2) | 29 (7.8) | 13 (14.0) | 0.148 |
| Hyperinflation, n (%) | 43 (10.0) | 23 (6.1) | 6 (6.5) | 0.124 |
| DLCO (z-score) | −0.32 (±0.94) | −0.36 (±0.92) | −0.48 (±0.94) | 0.157 |
| KCO (z-score) | −0.31 (±0.99) | −0.25 (±0.94) | −0.16 (±0.99) | 0.162 |
| Impaired diffusion, n (%) | 49 (12.3) | 46 (13.2) | 8 (9.4) | 0.684 |
| R5–20 Hz, median [IQR], kPa·L−1·s−1 | 0.01 [0; 0.04] | 0.01 [0; 0.04] | 0.03 [0.01; 0.05] | 0.002 |
| R5 Hz (z-score), median [IQR], n (%) | 0.31 (1.03) | 0.27 (1.05) | 0.23 (0.86) | 0.917 |
| AX5 Hz (z-score), median [IQR], n (%) | 0.92 (1.02) | 0.89 (1.06) | 0.96 (1.12) | 0.857 |
| Small airway dysfunction, n (%) | 115 (40.1) | 98 (28.0) | 26 (31.0) | 0.844 |
The data are given as mean (SD), median [IQR], or frequency (%) as appropriate. Spirometric, body plethysmographic, and gas uptake data are presented as z-scores with the reference values from the Global Lung Function Initiative (GLI, see methods). For airwave oscillometry, R5 Hz and AX5 Hz are also reported as z-scores, while R5–20 Hz is reported as absolute measure due to a lack of valid z-scores (see methods). Lung function impairments are defined as follows: airflow obstruction, FEV1/FVC <LLN; restriction, TLC <LLN; hyperinflation, RV/TLC >ULN and RV >ULN; impaired diffusion, TLCO <LLN, or KCO <LLN; small airway dysfunction, R5–20 Hz >0.07 kPa·L⁻¹·s⁻¹ or AX5 Hz >ULN.
Fig. 2.
Venn diagram of lung function impairments across all participants (a), participants in the low PCS score group (b), participants in the intermediate PCS score group (c), and participants in the high PCS score group (d).
Dyspnea in Relation to PCS Score and Lung Function
The prevalence of dyspnea (MDP-A1 ≥1 or mMRC ≥1) differed significantly between PCS score groups, rising from 19.3% in the low via 53.8% in the intermediate to 81.8% in the high PCS score group (p < 0.001). To investigate the basis of this phenomenon further, two contrasting subgroups of participants were compared, namely, those with dyspnea and a high PCS score (dyspnea/high PCS, n = 53) on the one hand and those without dyspnea and a low PCS score (no dyspnea/low PCS, n = 63) on the other hand. Of note, the dyspnea/high PCS group had a significantly higher proportion of female participants and was older and had higher BMI (Table 2).
Table 2.
Comparison of lung function parameters in previously healthy participants with dyspnea and severe PCS vs. no dyspnea and weak/absent PCS
| | Low PCS + no dyspnea, N = 301 | High PCS + present dyspnea, N = 72 | p value |
|---|---|---|---|
| Age, median [IQR], years | 34 [27; 48] | 44 [32; 55] | 0.001 |
| Female/male ratio, n (%) | 139 (45)/162(54) | 50 (69)/22 (31) | <0.001 |
| BMI, mean (SD), kg/m2 | 24.7 (±3,8) | 26.9 (±5,8) | 0.008 |
| FEV1/FVC (z-score), mean (SD) | −0.32 (±1.08) | −0.23 (±1.18) | 0.534 |
| Airflow obstruction, n (%) | 30 (10.1) | 9 (13.24) | 0.513 |
| FRC (z-score) | 0.32 (±0.92) | 0.05 (±0.98) | 0.030 |
| RV (z-score) | 0.57 (±1.04) | 0.43 (±0.98) | 0.321 |
| TLC (z-score) | −0.23 (±1.14) | −0.47 (±0.96) | 0.178 |
| RV/TLC (z-score) | 0.91 (±1.18) | 0.92 (±1.11) | 0.989 |
| Restriction, n (%) | 26 (8.63) | 7 (10.29) | 0.815 |
| Hyperinflation, n (%) | 34 (11.76) | 6 (8.82) | 0.669 |
| DLCO (z-score) | −0.28 (±0.93) | −0.52 (±0.99) | 0.064 |
| KCO (z-score) | −0.32 (±0.96) | −0.24 (±0.85) | 0.593 |
| Impaired diffusion, n (%) | 32 (11.81) | 5 (7.94) | 0.505 |
| R5–20 Hz, median [IQR], kPa·L−1·s−1 | 0.01 [0; 0.03] | 0.03 [0.01; 0.07] | <0.0001 |
| R5 Hz (z-score), n (%) | 0.31 (1.02) | 0.20 (0.85) | 0.717 |
| AX5 Hz (z-score), n (%) | 0.85 (0.95) | 0.96 (1.19) | 0.521 |
| SAD, n (%) | 69 (24.91) | 23 (37.1) | 0.058 |
Data are presented as mean (SD), median [IQR], or frequency (%) as appropriate. For definition of parameters, see subheading of Table 1.
The dyspnea/high PCS group had FRC values that were closer to the normal values, as defined by z-scores, but had significantly worse R5–R20 Hz values, than the no dyspnea/low PCS group. The difference in frequency of SAD (25% vs. 37%) was of borderline statistical significance (p = 0.058; Table 2). While SAD showed at least a weak association with dyspnea in PCS patients, no notable correlation was observed between individual dyspnea questionnaire items and single lung function parameters (all r < 0.3) (online suppl. Fig. 1).
Longitudinal Changes in Lung Function and Dyspnea
Of the COVIDOM participants with a completed onsite follow-up, 286 were previously healthy individuals according to the study criteria. Longitudinal analysis of the entire cohort revealed significant changes in different lung function parameters from baseline to follow-up (Table 3). Specifically, there was a significant decrease in FEV1/FVC and KCO z-score, combined with a significant increase in peripheral airway resistance (R5–R20) and AX, all meaning worsening of the participant’s condition. By contrast, measures of hyperinflation (RV and RV/TLC) showed significant reduction, indicating improvement. These longitudinal changes were consistent across all PCS score groups and were independent of dyspnea status (see online suppl. Tables 2–4).
Table 3.
Longitudinal comparison of lung functions
| | Baseline (n = 265) | Follow-up 1 (n = 265) | p value |
|---|---|---|---|
| PCS score group baseline, n (%) | | | – |
| 1 | 96 (36.23) | | |
| 2 | 102 (38.49) | | |
| 3 | 67 (25.28) | | |
| Reinfection since baseline, n (%) | | 113 (44.5) | – |
| FEV1/FVC (z-score), mean (SD) | −0.26 (0.99) | −0.57 (0.87) | <0.0001 |
| FRC (z-score), mean (SD) | 0.12 (0.85) | 0.05 (0.9) | 0.053 |
| RV (z-score), mean (SD) | 0.45 (0.81) | −0.05 (0.87) | <0.0001 |
| TLC (z-score), mean (SD) | −0.32 (0.94) | −0.31 (0.92) | 0.865 |
| RV/TLC (z-score), mean (SD) | 0.81 (0.92) | 0.14 (0.87) | <0.0001 |
| DLCO (z-score), mean (SD) | −0.33 (0.92) | −0.35 (1.0) | 0.427 |
| KCO (z-score), mean (SD) | −0.27 (0.94) | −0.36 (0.99) | 0.009 |
| R5–20 Hz, median [IQR], kPa·L−1·s−1 | 0.02 [0; 0.04] | 0.03 [0; 0.05] | 0.006 |
| R5 Hz (z-score), mean (SD) | 0.17 (1.02) | 0.27 (0.95) | 0.017 |
| AX5 Hz (z-score), mean (SD) | 0.85 (1.04) | 1.01 (0.93) | 0.002 |
Data are presented as mean (SD), median [IQR], or frequency (%) as appropriate. Participants were excluded if at least one of the parameters was missing. The total of 265 participants represents the frequency of individuals that were scheduled for an onsite follow-up, had complete lung function data, and were previously healthy according to the criteria of the study.
The longitudinal course of dyspnea was next analyzed across the two previously defined subgroups (dyspnea/high PCS vs. no dyspnea/low PCS). At follow-up, 66.7% of the high-PCS/dyspnea group reported persistent dyspnea. In the low PCS/no dyspnea group, only 11.5% developed new-onset dyspnea, while 88.5% remained asymptomatic. Despite these different symptom trajectories, a comparison between participants with persistent dyspnea (from the high PCS group) and with persistent absence of dyspnea (from the low PCS group) revealed no significant differences in any of the lung function parameters studied, neither at baseline nor at follow-up (Table 4).
Table 4.
Comparison of lung function in participants with improved vs. persistent dyspnea at follow-up 1
| | Absent dyspnea (n = 54) | Persistent dyspnea (n = 64) | p value |
|---|---|---|---|
| Age, median [IQR], years | 37 [27; 51] | 44 [34.75; 52] | 0.029 |
| Female/male ratio, n (%) | 35 (64.81)/19 (35.19) | 42 (65.62)/22 (34.38) | 1 |
| BMI, mean (SD), kg/m2 | 25.84 (5.8) | 27.42 (5.64) | 0.150 |
| Reinfection since baseline, n (%) | 23 (42.6) | 24 (40.0) | 0.850 |
| Airflow obstruction, n (%) | 5 (9.26) | 5 (8.2) | 1 |
| Restriction, n (%) | 4 (7.41) | 5 (8.33) | 1 |
| Hyperinflation, n (%) | 4 (7.41) | 1 (1.64) | 0.185 |
| Impaired diffusion, n (%) | 6 (12.5) | 4 (7.14) | 0.507 |
| SAD, n (%) | 18 (37.5) | 18 (32.73) | 0.681 |
| Baseline | |||
| FEV1/FVC (z-score), mean (SD) | −0.28 (0.96) | −0.18 (1.06) | 0.623 |
| FRC (z-score), mean (SD) | 0.19 (0.87) | −0.03 (0.87) | 0.178 |
| RV (z-score), mean (SD) | 0.56 (1.02) | 0.39 (0.80) | 0.305 |
| TLC (z-score), mean (SD) | −0.35 (0.99) | −0.36 (0.83) | 0.917 |
| RV/TLC (z-score), mean (SD) | 0.98 (1.09) | 0.75 (0.88) | 0.228 |
| DLCO (z-score), mean (SD) | −0.36 (0.98) | −0.21 (0.79) | 0.397 |
| KCO (z-score), mean (SD) | −0.16 (0.93) | −0.16 (0.82) | 0.981 |
| R5–20 Hz, median [IQR], kPa·L−1·s−1 | 0.02 [0; 0.05] | 0.03 [0.01; 0.06] | 0.507 |
| R5 Hz (z-score), mean (SD) | 0.2 (1.06) | 0.28 (0.95) | 0.700 |
| AX5 Hz (z-score), mean (SD) | 1.12 (1.34) | 0.83 (1.04) | 0.214 |
| Follow-up 1 | |||
| FEV1/FVC (z-score), mean (SD) | −0.71 (0.98) | −0.59 (0.85) | 0.462 |
| FRC (z-score), mean (SD) | 0.05 (0.87) | −0.09 (0.96) | 0.398 |
| RV (z-score), mean (SD) | 0.01 (0.75) | −0.05 (0.99) | 0.711 |
| TLC (z-score), mean (SD) | −0.38 (0.86) | −0.36 (0.97) | 0.892 |
| RV/TLC (z-score), mean (SD) | 0.24 (0.75) | 0.15 (0.92) | 0.598 |
| DLCO (z-score), mean (SD) | −0.24 (1.01) | −0.34 (0.95) | 0.578 |
| KCO (z-score), mean (SD) | −0.18 (0.99) | −0.3 (0.84) | 0.507 |
| R5–20 Hz, median [IQR], kPa·L⁻¹·s⁻¹ | 0.03 [0.01; 0.06] | 0.04 [0; 0.07] | 0.914 |
| R5 Hz (z-score), mean (SD) | 0.43 (1.09) | 0.31 (0.93) | 0.567 |
| AX5 Hz (z-score), mean (SD) | 1.04 (1.07) | 1.13 (0.87) | 0.658 |
| Change over time | |||
| Δ FEV1/FVC (z-score), mean (SD) | 0.43 (0.68) | 0.45 (0.81) | 0.910 |
| Δ FRC (z-score), mean (SD) | 0.15 (0.63) | 0.08 (0.91) | 0.665 |
| Δ RV (z-score), mean (SD) | 0.55 (0.95) | 0.48 (1.18) | 0.520 |
| Δ TLC (z-score), mean (SD) | 0.03 (0.65) | 0.04 (0.77) | 0.972 |
| Δ RV/TLC (z-score), mean (SD) | 0.74 (1.01) | 0.61 (1.15) | 0.521 |
| Δ DLCO (z-score), mean (SD) | −0.07 (0.64) | 0.15 (0.81) | 0.165 |
| Δ KCO (z-score), mean (SD) | 0.02 (0.54) | 0.14 (0.69) | 0.370 |
| Δ R5–20 Hz, mean (SD), kPa·L⁻¹·s⁻¹ | 0 (0.05) | 0 (0.05) | 0.571 |
| Δ R5 Hz (z-score), mean (SD) | −0.31 (0.72) | −0.06 (0.95) | 0.175 |
| Δ AX5 Hz (z-score), mean (SD) | −0.09 (0.87) | −0.23 (1.04) | 0.502 |
Data are presented as mean (SD), median [IQR], or frequency (%) as appropriate. Participants were excluded if at least one of the parameters was missing.
Discussion
In our cohort of previously healthy participants of the multicenter, population-based COVIDOM study, infected during 2020–2022, we made several key observations. Dyspnea prevalence increased significantly with PCS severity, and highly symptomatic participants were more often women. SAD, as identified by oscillometry, was the most frequent impairment (35%) and the strongest functional correlate of severe PCS. Elevated peripheral airway resistance (R5–20 Hz) further distinguished highly symptomatic from low-symptom individuals. Longitudinally, although airflow obstruction, diffusion capacity, and SAD worsened, these changes did not correlate with dyspnea persistence. Furthermore, no lung function parameters reliably discriminated persistent from improving dyspnea.
Mild alterations of lung volumes have been observed in two previous German studies on mostly mild cases of COVID-19 (≈10% hospitalization). In the Hamburg City Health Study, TLC and specific airway resistance were lower in post-COVID participants vs. controls, regardless of symptoms [25]. Trinkmann et al. [26] found lower VC, FEV1, and TLCO in persistently symptomatic patients and speculated that underlying respiratory muscle weakness or subtle airway abnormalities might be detectable only with oscillometry. Our study constitutes the first large-scale analysis in >900 individuals without a history of pre-existing disease, employing a full diagnostic array of lung function assessments, encompassing AOS. While TLC and FRC z-scores differed between PCS groups, average values remained normal. Interestingly, FRC was closer to the reference value in the severe PCS score group and slightly higher in the low PCS score group, while TLC was below the reference value in the severe PCS score group. Lower FRC in severe PCS may reflect impaired muscle strength, altered mechanics, or dysfunctional breathing. Chaotic breathing, known to cause exertional dyspnea and a possible target of physiotherapy, could underlie such findings [27].
Previous AOS studies in PCS yielded divergent results. Lopes et al. [28] and Lu et al. [29] demonstrated AOS sensitivity beyond spirometry, whereas two Finnish cohorts reported no SAD [30, 31]. Heterogeneity in definitions, severity, and sample sizes likely explain discrepancies. Kjellberg et al. [32] recently confirmed SAD relevance but highlighted ventilation inhomogeneity by multiple breath washout as even more sensitive and clinically meaningful technique. In our study, SAD prevalence was similar across PCS severity groups and in general rather high. While between-group frequency differences did not reach significance, SAD occurred largely independently of other impairments such as restriction, obstruction, impaired diffusion, or hyperinflation, warranting further exploration. Associations with BMI, known to influence peripheral obstruction, must also be considered, especially as R5–20 Hz lacks standardized z-scores and as other AOS parameters such as respiratory impedance failed to show similarly stringent effects [33].
The MDP captures sensory and affective dimensions of dyspnea [34]. In our study, neither unpleasantness nor sensory or affective domains correlated with lung function. Still, dyspnea in severe PCS coincided with higher peripheral resistance values, supporting AOS as a potentially more sensitive tool than spirometry, BP, or diffusion capacity in detecting clinically relevant abnormalities. A longitudinal study by Iversen et al. [8] with pre-pandemic spirometry found mild COVID-19 disease severity associated with excess FVC decline, but not significant FEV1 decline. No corresponding TLC changes were observed in our cohort, though we noted normalization of an elevated RV/TLC ratio. Lewis et al. [10] found no long-term decline in lung function when considering pre-pandemic levels. The BAMSE birth cohort similarly reported no effects of mild-to-moderate COVID-19 on young adults [11]. While our cohort lacks pre-pandemic values, longitudinal deterioration of FEV1/FVC and accompanied impairments of oscillometric resistance suggests mechanical changes. Future analyses of new follow-up visits by the COVIDOM cohort will have to clarify, whether the deterioration of FEV1/FVC, peripheral obstruction and airway impedance persists. The robustness of AOS, based on tidal breathing without forced maneuvers, increases the plausibility and the practical meaning of our results.
Regarding perception of dyspnea, we found no predictive lung function parameters distinguishing resolving vs. persisting dyspnea. This supports the view that subjectively perceived dyspnea is difficult to objectify. Lo et al. [35] reported stronger associations between SAD, hyperinflation, and dyspnea, though their referred, comorbidity-enriched cohort differs from ours. Notably, in their study, improvements of SP, BP, and DLCO started from a low level (i.e., 50% of participants with impaired DLCO) and were not paralleled by oscillometric recovery, nor did dyspnea prevalence decrease.
Our cohort also exhibited sex-related differences: women were over-represented among highly symptomatic, dyspneic participants. This aligns with prior evidence that women report persistent symptoms more frequently, independent of physiological correlates [36, 37]. Sex-related differences in immune response, symptom perception, and psychosocial vulnerability may contribute to this pattern [38].
Strengths of our study include its population-based design, large sample, and restriction to individuals without comorbidity, improving generalizability. Rigorous, standardized lung function testing across sites further supports validity. Importantly, our subgroup reflects those most affected by PCS-related occupational and societal consequences, despite prior good health.
Limitations include absence of pre-pandemic lung function data, restricting causal inference. Participants were infected mostly pre-Omicron and largely unvaccinated; findings may not extrapolate to later variants. Possible selection bias exists, as severely affected individuals might be under- or over-represented depending on access or tolerance for long study visits.
Conclusion
In this large population-based cohort, oscillometry-derived R5–20 Hz was the sole lung function impairment distinguishing dyspnea in PCS. While prevalence of SAD did not differ significantly between PCS severity groups, its high frequency (≈35%) and longitudinal worsening underscore its potential relevance. Altered oscillometric resistance and FRC patterns may reflect abnormal breathing behaviors, highlighting physiotherapy as a potential treatment target.
Acknowledgments
We thank the COVIDOM study teams at Kiel, Würzburg, and Berlin for study logistics, contact management, and data acquisition. We also thank the NAPKON infrastructure units for making COVIDOM possible, including the Interaction Core Unit (ICU), Epidemiological Core Unit (ECU), and Biosample Core Unit (BCU), as well as the NAPKON Steering Committee and the NAPKON Use and Access Committee (see the NAPKON study group list for a full disclosure of NAPKON and COVIDOM collaborators who contributed to the project). There was no medical writer involved.
Statement of Ethics
The COVIDOM study received approval from the Local Ethics Committees of the universities in Kiel (Ref. No. D 537/20) and Würzburg (Ref. No. 236/20_z). In accordance with the professional code of the Berlin Medical Association, the approval granted by the Kiel Ethics Committee also extended to the Berlin study site. The study is registered at www.clinicaltrials.com (NCT04679584) and www.drks.de (DRKS00023742). It was conducted in accordance with all applicable legal and ethical standards. All participants provided written informed consent prior to their inclusion.
Conflict of Interest Statement
With respect to the present manuscript, all authors declare that they have no conflicts of interest. Several authors declare receipt of grants, royalties/licenses, fees for consultation, honoraria for lectures or presentations, payment for expert testimony, support for attending meetings, and other financial activities during the past 36 months that are not in conflict with the present work.
Funding Sources
The COVIDOM study is part of the National Pandemic Cohort Network (NAPKON). NAPKON is funded by COVID-19-related grants from the German Federal Ministry for Education and Research (BMBF), administrated by the Network University Medicine (NUM; NAPKON support code: 01KX2021). EEG received funding by the DFG Clinician Scientist Program in Evolutionary Medicine “CSEM” (project No. 413490537). Parts of the infrastructure of the Kiel and Würzburg study sites were funded by the federal states of Schleswig-Holstein and Bavaria. The funding sources were not involved in the collection, analysis, and interpretation of the data.
Author Contributions
Conceptualization: T.B., W.L., L.K., C.S., K.F.R., M.W., J.J.V., S.Stö., T.K., P.U.H., M.K., and S.Sch. Formal analysis: A.-K.R., T.B., and M.A. Funding acquisition: T.B., W.L., M.W., J.J.V., T.K., P.U.H., M.K., and S.Sch. Investigation: T.B., L.M.R., A.V., W.M., S.Stö., J.H., A.S., E.E.G., A.V., E.H., and S.B.-L. Methodology: T.B., W.L., L.M.K., K.F.R., M.W., J.J.V., S.Stö., T.K., P.H., M.K., S.Sch., J.H., S.B.-L., E.E.G., and S.H. Project administration: T.B., W.L., S.B.-L., A.S., L.K., M.W., J.J.V., T.K., P.U.H., M.K., S.Sch., J.H., C.N., S.H., and A.V. Resources: T.B., W.L., M.W., J.J.V., T.K., P.H., M.K., S.Sch., and J.H. Software: A.-K.R. and E.H. Supervision: T.B., M.A., and M.K. Validation: T.B., L.R., W.L., A.S., L.K., J.F., P.U.H., M.K., S.Sch., A.-K.R., C.N., and A.V. Visualization: A.-K.R. and T.B. Writing – original draft: M.A., T.B., and A.-K.R. Writing – review and editing: T.B., A.-K.R., L.M.R., S.B.-L., A.V., C.N., A.S., S.Stö, P.U.H., T.Z., M.W., L.M.K., T.K., E.H., S.H., J.J.V., W.L., M.K., K.F.R., E.E.G., S.Sch., J.H., and M.A. The data underlying the present work were verified by T.B., W.L., P.U.H., M.K., and S.Sch. All authors had full access to all data, approved the final version of the manuscript, and agreed to be accountable for all aspects of the work.
Funding Statement
The COVIDOM study is part of the National Pandemic Cohort Network (NAPKON). NAPKON is funded by COVID-19-related grants from the German Federal Ministry for Education and Research (BMBF), administrated by the Network University Medicine (NUM; NAPKON support code: 01KX2021). EEG received funding by the DFG Clinician Scientist Program in Evolutionary Medicine “CSEM” (project No. 413490537). Parts of the infrastructure of the Kiel and Würzburg study sites were funded by the federal states of Schleswig-Holstein and Bavaria. The funding sources were not involved in the collection, analysis, and interpretation of the data.
Data Availability Statement
All data from this study are available upon request through the NAPKON Data Use and Access Committee. For details on the NAPKON data governance and to submit a research proposal, please visit https://proskive.napkon.de (accessed July 11, 2025).
Supplementary Material.
References
- 1. Bahmer T, Borzikowsky C, Lieb W, Horn A, Krist L, Fricke J, et al. Severity, predictors and clinical correlates of Post-COVID syndrome (PCS) in Germany: a prospective, multi-centre, population-based cohort study. eClinicalMedicine. 2022;51:101549. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Whitaker M, Elliott J, Bodinier B, Barclay W, Ward H, Cooke G, et al. Variant-specific symptoms of COVID-19 in a study of 1,542,510 adults in England. Nat Commun. 2022;13(1):6856. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Wiersinga WJ, Rhodes A, Cheng AC, Peacock SJ, Prescott HC. Pathophysiology, transmission, diagnosis, and treatment of coronavirus disease 2019 (COVID-19): a review. JAMA. 2020;324(8):782–93. [DOI] [PubMed] [Google Scholar]
- 4. Greenhalgh T, Sivan M, Perlowski A, Nikolich JŽ. Long COVID: a clinical update. Lancet. 2024;404(10453):707–24. [DOI] [PubMed] [Google Scholar]
- 5. Tsampasian V, Elghazaly H, Chattopadhyay R, Debski M, Naing TKP, Garg P, et al. Risk factors associated with Post−COVID-19 condition: a systematic review and meta-analysis. JAMA Intern Med. 2023;183(6):566–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Ballhausen-Lübcker S, Ruß A-K, Lieb W, Schäfer A, Krist L, Fricke J, et al. Subdomains of Post-COVID syndrome (PCS): a population-based study. BMC Infect Dis. 2025;25(1):1072. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7. Koczulla AR, Ankermann T, Behrends U, Berlit P, Berner R, Böing S, et al. S1-Leitlinie Long-/Post-COVID. Pneumologie. 2022;76(12):855–907. [DOI] [PubMed] [Google Scholar]
- 8. Iversen KK, Ronit A, Ahlström MG, Nordestgaard BG, Afzal S, Benfield T. Lung function trajectories in mild COVID-19 with 2-year Follow-up. J Infect Dis. 2024;229(6):1750–8. [DOI] [PubMed] [Google Scholar]
- 9. Iversen KK, Afzal S, Ahlström MG, Nordestgaard BG, Schneider UV, Nielsen L, et al. Lung function decline in relation to COVID-19 in the general population: a matched cohort study with prepandemic assessment of lung function. J Infect Dis. 2022;225(8):1308–16. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Lewis KL, Helgeson SA, Tatari MM, Mallea JM, Baig HZ, Patel NM. COVID-19 and the effects on pulmonary function following infection: a retrospective analysis. EClinicalMedicine. 2021;39:101079. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11. Mogensen I, Hallberg J, Björkander S, Du L, Zuo F, Hammarström L, et al. Lung function before and after COVID-19 in young adults: a population-based study. J Allergy Clin Immunol Glob. 2022;1(2):37–42. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12. Trinkmann F, Watz H, Herth FJF. Why do we still cling to spirometry for assessing small airway function? Eur Respir J. 2020;56(1):2001071. [DOI] [PubMed] [Google Scholar]
- 13. Taib RR, Kozlov Y, Ekshtein A, Gordon B, Wand O, Ben-Ari O. A comparison of pulmonary function pre and post mild SARS-CoV-2 infection among healthy adults. BMC Pulm Med. 2025;25(1):163. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14. Horn A, Krist L, Lieb W, et al. Long-term health sequelae and quality of life at least 6 months after infection with SARS-CoV-2: design and rationale of the COVIDOM-Study as part of the NAPKON population-based cohort platform (POP). Infect [Internet]. 2021. [cited 2021 Oct 14]. Available from: https://link.springer.com [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Mahler DA, Wells CK. Evaluation of clinical methods for rating dyspnea. Chest. 1988;93(3):580–6. [DOI] [PubMed] [Google Scholar]
- 16. Meek PM, Banzett R, Parsall MB, Gracely RH, Schwartzstein RM, Lansing R. Reliability and validity of the multidimensional dyspnea profile. Chest. 2012;141(6):1546–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Choi SW, Victorson DE, Yount S, Anton S, Cella D. Development of a conceptual framework and calibrated item banks to measure patient-reported dyspnea severity and related functional limitations. Value Health. 2011;14(2):291–306. [DOI] [PubMed] [Google Scholar]
- 18. Koczulla AR, Ankermann T, Behrends U, Berlit P, Böing S, Brinkmann F, et al. [S1 Guideline Post-COVID/Long-COVID]. Pneumologie. 2021;75(11):869–900. [DOI] [PubMed] [Google Scholar]
- 19. Stanojevic S, Graham BL, Cooper BG, Thompson BR, Carter KW, Francis RW, et al. Official ERS technical standards: Global Lung Function Initiative reference values for the carbon monoxide transfer factor for Caucasians. Eur Respir J. 2017;50(3):1700010. [DOI] [PubMed] [Google Scholar]
- 20. Quanjer PH, Stanojevic S, Cole TJ, Baur X, Hall GL, Culver BH, et al. Multi-ethnic reference values for spirometry for the 3–95-yr age range: the global lung function 2012 equations. Eur Respir J. 2012;40(6):1324–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21. Hall GL, Filipow N, Ruppel G, Okitika T, Thompson B, Kirkby J, et al. Official ERS technical standard: Global Lung Function Initiative reference values for static lung volumes in individuals of European ancestry. Eur Respir J. 2021;57(3):2000289. [DOI] [PubMed] [Google Scholar]
- 22. Oostveen E, MacLeod D, Lorino H, Farré R, Hantos Z, Desager K, et al. The forced oscillation technique in clinical practice: methodology, recommendations and future developments. Eur Respir J. 2003;22(6):1026–41. [DOI] [PubMed] [Google Scholar]
- 23. Oppenheimer BW, Goldring RM, Herberg ME, Hofer IS, Reyfman PA, Liautaud S, et al. Distal airway function in symptomatic subjects with normal spirometry following world trade center dust exposure. Chest. 2007;132(4):1275–82. [DOI] [PubMed] [Google Scholar]
- 24. Abdo M, Watz H, Trinkmann F, et al. Oscillometry-defined small airway dysfunction in tobacco-exposed adults with impaired or preserved airflow. Am J Respir Crit Care Med. 2025. rccm.202501-0028OC. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Petersen EL, Goßling A, Adam G, et al. Multi-organ assessment in mainly non-hospitalized individuals after SARS-CoV-2 infection: the Hamburg City Health Study COVID programme. Eur Heart J. 2022:ehab914. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26. Trinkmann F, Müller M, Reif A, Kahn N, Kreuter M, Trudzinski F, et al. Residual symptoms and lower lung function in patients recovering from SARS-CoV-2 infection. Eur Respir J. 2021;57(2):2003002. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 27. Samaranayake CB, Warren C, Rhamie S, Haji G, Wort SJ, Price LC, et al. Chaotic breathing in post-COVID-19 breathlessness: a key feature of dysfunctional breathing can be characterised objectively by approximate entropy. ERJ Open Res. 2023;9(4):00117-2023–02023. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28. Lopes AJ, Litrento PF, Provenzano BC, Carneiro AS, Monnerat LB, da Cal MS, et al. Small airway dysfunction on impulse oscillometry and pathological signs on lung ultrasound are frequent in post-COVID-19 patients with persistent respiratory symptoms. PLoS One. 2021;16(11):e0260679. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29. Lu C, Lin S, Chen C. Impulse oscillometry in patients with persistent post‐COVID‐19 symptoms: a retrospective study. Health Sci Rep. 2024;7(6):e2191. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30. Tamminen P, Kerimov D, Viskari H, Aittoniemi J, Syrjänen J, Lehtimäki L. Lung function during and after acute respiratory infection in COVID-19 positive and negative outpatients. Eur Respir J. 2022;59(3):2102837. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31. Lindahl A, Reijula J, Malmberg LP, Aro M, Vasankari T, Mäkelä MJ. Small airway function in Finnish COVID-19 survivors. Respir Res. 2021;22(1):237. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32. Kjellberg S, Holm A, Berguerand N, Sandén H, Schiöler L, Olsén MF, et al. Impaired function in the lung periphery following COVID-19 is associated with lingering breathing difficulties. Physiol Rep. 2024;12(2):e15918. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33. Ringbaek T, Froelund L, Mortensen J, El Ali HH. The influence of obesity on IOS parameters in asthma, COPD, and other lung diseases: analyzed by random forest. BMC Pulm Med. 2025;25(1):218. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34. Banzett RB, O’Donnell CR, Guilfoyle TE, Parshall MB, Schwartzstein RM, Meek PM, et al. Multidimensional Dyspnea Profile: an instrument for clinical and laboratory research. Eur Respir J. 2015;45(6):1681–91. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35. Lo P-C, Feng J-Y, Hsiao Y-H, Su KC, Chou KT, Chen YM, et al. Long COVID symptoms after 8-month recovery: persistent static lung hyperinflation associated with small airway dysfunction. Respir Res. 2024;25(1):209. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36. Sudre CH, Murray B, Varsavsky T, et al. Attributes and predictors of long COVID. Nat Med. 2021. [Internet] [cited 2021 Mar 28]; Available from: http://www.nature.com/articles/s41591-021-01292-y [Google Scholar]
- 37. Fernández-de-Las-Peñas C, Martín-Guerrero JD, Pellicer-Valero ÓJ, et al. Female sex is a risk factor associated with long-term Post-COVID related-symptoms but not with COVID-19 symptoms: the LONG-COVID-EXP-CM multicenter study. J Clin Med. 2022;11:413. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38. Bai F, Tomasoni D, Falcinella C, Barbanotti D, Castoldi R, Mulè G, et al. Female gender is associated with long COVID syndrome: a prospective cohort study. Clin Microbiol Infect. 2022;28(4):611.e9–611.e16. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
All data from this study are available upon request through the NAPKON Data Use and Access Committee. For details on the NAPKON data governance and to submit a research proposal, please visit https://proskive.napkon.de (accessed July 11, 2025).
