Lung Function Trajectories in Mild COVID-19 With 2-year Follow-up

Abstract

Background

The long-term pulmonary sequelae of mild coronavirus disease 2019 (COVID-19) remains unknown. In this study, we aimed to characterize lung function trajectories in individuals with mild COVID-19 from preinfection to 2 years postinfection.

Methods

We reinvited participants 2 years after infection from our matched cohort study of the Copenhagen General Population who had initially been examined 5.4 months after infection. We repeated lung tests and questionnaires. Linear mixed models were used to estimate dynamics in lung volumes in individuals with COVID-19 patients versus uninfected controls over two intervals: from pre-infection to 6 months postinfection and 6 months postinfection to 2 years postinfection.

Results

52 individuals (48.6%) attended the 2-year examination at median 1.9 years (interquartile range, 1.8–2.4) after COVID-19, all with mild infection. Individuals with COVID-19 had an adjusted excess decline in forced expiratory volume in 1 second (FEV₁) of 13.0 mL per year (95% confidence interval [CI], −23.5 to −2.5; P = .02) from before infection to 6 months after infection compared to uninfected controls. From 6 to 24 months after infection, they had an excess decline of 7.5 mL per year (95% CI, −25.6–9.6; P = .40). A similar pattern was observed for forced vital capacity (FVC). Participants had a mean increase in diffusing capacity for carbon monoxide (DL_co) of 3.33 (SD 7.97) between the 6- and 24-month examination.

Conclusions

Our results indicate that mild COVID-19 infection affects lung function at the time of infection with limited recovery 2 years after infection.

Pulmonary sequelae following severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection have been widely reported, leading to an increased understanding of lung function trajectories at short to mid-term follow-up [1–7]. Prospective cohort studies and meta-analyses investigating lung function and lung abnormalities at 1-year follow-up report impaired diffusion capacity and restrictive pulmonary dysfunction as the most frequent pulmonary consequences of coronavirus disease 2019 (COVID-19). The prevalence of these findings varies between 8% and 31% at 1-year follow-up in survivors hospitalized with COVID-19 [8–11]. However, it remains unknown if the initial lung impairment regresses or progresses. Reports from SARS and Middle East respiratory syndrome (MERS) suggest that up to a third of admitted patients experienced pulmonary dysfunction, and in particular restrictive dysfunction, for up to 3 years following infection [12]. Therefore, there is a need for long-term follow-up studies of patients with COVID-19 to evaluate the prognosis and the potential recovery of the initial pulmonary impairment.

Furthermore, it has been established that the impact of COVID-19 on lung function is not limited to severe cases. Studies of mild COVID-19 have demonstrated an apparent decline in dynamic volumes and carbon monoxide diffusing capacity (DL_co) in 16%–36% of patients evaluated from 3 to 5 months after infection [13, 14]. Our previous study on pulmonary function decline following mild COVID-19 showed an excess decline in dynamic lung volumes compared to uninfected individuals [15]. Taken together, this indicates that even mild infection causes pulmonary parenchymal damage. However, the extent and duration of this decline remains unclear. In this study, we aimed to investigate the long-term impact of mild COVID-19 on lung function and respiratory symptoms by extending our previous analyses with an additional follow-up visit 2 years after infection [15].

METHODS

Study Population and Approvals

Participants from the Copenhagen General Populations Study (CGPS) COVID-19 substudy on lung function and respiratory symptoms were invited to attend a second follow-up examination at approximately 2 years after infection. Results from the early follow-up (median 5.4 months after infection) on lung function have previously been reported [15]. The study was approved by the regional ethics committee (H-KF-01-144/0 and H-20072296) and written and oral consent were obtained from all participants. The study report follows the STROBE guidelines for cohort studies [16].

All invited individuals had performed baseline (prior to infection) lung function testing and a questionnaire on respiratory symptoms as a part of CGPS between January 2013 and November 2020 as previously described [17]. Examinations at 6 and 24 months after COVID-19 were both performed at Copenhagen University Hospital-Hvidovre Hospital, the 6-month visit was conducted from March to June of 2021 and the 24-month visit during September and October of 2022. Participants had tested positive for SARS-CoV-2 infection between 9 March 2020 and 16 January 2021.

Data Collection and Lung Function

Information on sex (male/female), age (years), body mass index (BMI; kg/m₂), and smoking status (classified as current/former/never smokers) at baseline was obtained from the CGPS, along with information on cumulative tobacco consumption reported as pack years (number of cigarettes per day divided by 20 × years). The questionnaire on self-reported respiratory morbidity and symptoms was administered at all 3 visits and reviewed by the investigator together with the participant. Dyspnea was defined according to the modified Medical Research Council scale, with a score of ≥2 defined as dyspnea [18]. Sputum was defined as a daily productive cough for more than 3 months each year.

SARS-CoV-2 polymerase chain reaction (PCR) test results were obtained from microbiology departments serving the area of greater Copenhagen [15]. At 6 months after infection, information on symptoms and duration of symptoms related to COVID-19 was obtained from a specific questionnaire. Information on respiratory support were obtained from electronic medical records if participants were admitted to hospital in relation to acute infection.

Spirometry was performed at all 3 visits and diffusing capacity was measured at 6 and 24 months after infection. Baseline spirometry was conducted using a Portable EasyOne Plus ultrasonic spirometer (ndd Medical) [19]. At 6 and 24 months after infection, dynamic volumes, DL_co, and total lung capacity were measured using a portable single-breath diffusing capacity device (EasyOne Pro; ndd Medical Technologies). DL_CO was measured using the single-breath maneuver technique with carbon monoxide as a tracer gas. Spirometry and DL_CO tests were performed according to published standards and as previously described [15, 20]. The predicted values for forced expiratory volume in 1 second (FEV₁), forced vital capacity (FVC), FEV₁/FVC, and lower limit of normal (LLN) were calculated based on the reference equations provided by the Global Lung Function Initiative [20].

Statistical Methods

Normality of all variables was checked using histograms and QQ plots. Continuous variables were reported as mean with standard deviation (SD) or median with interquartile range (IQR) and for comparison t test, paired t test, or nonparametric tests were used as appropriate. Categorical variables were reported as absolute values and percentages and compared using McNemar test for comparison of proportions.

Controls were individuals from the GCPS with spirometry data available prior to the pandemic and without a positive SARS-CoV-2 PCR test result. For each individual up to 5 controls from the GCPS were identified as previously described [15]. Individuals with COVID-19 were matched based on age at baseline and at 6 months after infection (±2.5 years), as well as sex and smoking status at baseline.

Our primary outcome was a comparison of the mean decline (slope) in FEV₁ per year between individuals with COVID-19 and uninfected controls from baseline to 6 months after infection and from 6 to 24 months after infection. Our secondary outcome was a comparison of the mean decline (slope) in FVC per year. An additional secondary outcome was the change in respiratory symptoms in individuals with COVID-19 from baseline to 24 months after infection. We used a linear mixed model adjusted for age, sex, height, smoking status, and years at baseline. An interaction link between COVID-19 and time between examination was used to assess if individuals with COVID-19 had an increased decline. Results from the regression analysis were reported as change in mL with 95% confidence intervals (CI) and P values. A 2-sided P value of .05 was considered statistically significant. Analyses were performed using R statistical software version 4.0.1.

RESULTS

Clinical Characteristics

Of 107 individuals who attended the 6-month examination, 58 (54.2%) accepted our invitation to the attend the 24-month examination (Figure 1). Six individuals had been hospitalized in relation to infection and were excluded from all tables and analyses. Individuals who attended the 6- and 24-month examinations after infection were comparable with the original study cohort (Supplementary Table 1). Time between baseline and 6 months after infection was median 5.7 years (IQR, 3.8–8.7) and there was a median of 1.9 years (IQR, 1.8–2.4) from infection to the 24-month examination. At baseline, the 52 infected individuals and the 219 controls had comparable characteristics (Table 1). The mean age of individuals with COVID-19 at baseline was 56.2 years (SD 7.7), approximately half were female, 42.3% were never smokers, and the mean cumulative tobacco consumption at baseline was 8.9 pack-years (SD 12.1) (Table 1). Among the 52 individuals with COVID-19, 3 reported respiratory comorbidities such as asthma, and 4 (5.2%) had a spirometry result below the LLN.

Between the 6- and 24-month examination, 18 individuals (34.6%) were reinfected with COVID-19, but none required hospitalization. All had completed a primary series of COVID-19 vaccination at the time of reinfection.

Change in Lung Function Volumes and Respiratory Symptoms

In individuals with COVID-19, we observed a mean annual decline of 46.1 mL per year (SD 41.0) for FEV₁ and 76.7 mL per year (IQR, 41.6–97.3) for FVC from baseline to 6 months after infection. From 6 to 24 months after infection, median annual decline for FEV₁ was 13.1 mL per year (IQR, −66.0 to 54.4) and for FVC it was 0.3 mL per year (IQR, −23.1 to 27.5). Table 2 presents change in lung volumes. Figure 2 shows the dynamics in respiratory symptoms from baseline to 24 months after infection. Two individuals (3.4%) reported dyspnea before infection, which increased to 7 participants (13.5%, P = .10) at 6 months after infection and to 8 (15.4%, P = .78) at 24 months after infection. Additionally, 6 (11.5%) individuals reported dyspnea for the first time at 6 months after infection and continued to report this 24 months after infection.

Change in Diffusing Capacity for Carbon Monoxide

DL_co was assessed at 6 and 24 months after infection. At 6 months after infection, the mean DL_co was 90.0% of predicted (SD 15.9), which increased to 94.9% of predicted (SD 18.4) at the 24-month examination (P < .01). At 6 months after infection, 10 participants (25.9%) had diffusion impairment, which decreased to 9 (22.4%) at the 24-month examination. See Table 2 for more details on the DL_co measurements.

Association of COVID-19 With Longitudinal Change in Lung Function Volumes

The longitudinal dynamics of FEV₁ and FVC in individuals with COVID-19 and uninfected controls are shown in Figure 3. From our mixed model, we observed that from baseline to approximately 6 months after infection, individuals with COVID-19 had an excess decline of 13.0 mL per year (95% CI, −23.5 to −2.5; P = .02; Table 3). From 6 months after infection to 24 months, they had an excess decline of 7.5 mL per year (95% CI, −25.6–9.6; P = .40) compared to uninfected controls. In terms of FVC, individuals with COVID-19 had an excess decline of 33.1 mL per year from baseline to 6 months follow-up (95% CI, −48.5 to −17.8; P < .01) compared to 16.1 mL (95% CI, −39.1–5.0; P = .15) in the second time period. Throughout the entire study duration, from baseline to 24 months after infection, individuals with COVID-19 had an excess decline in FEV₁ of 3.1 mL per year (95% CI, −32.9–5.5; P = .57) and in FVC of 19.8 mL per year (95% CI, −32.9 to −5.5; P < .01).

From our linear mixed model, we observed that COVID-19 alone, had an impact of −214.5 mL (95% CI, −398.1 to −33.4; P = .04) on FEV₁ and −300.6 mL (95% CI, −658.9 to −180.3; P < .01) on FVC from baseline to the 6-month visit. Subsequently, from 6 months postinfection to the 24-month visit, the impact on FEV₁ was −145.4 mL (95% CI, −349.3 to 26.2; P = .1) and on FVC −247.1 mL (95% CI, −565.3 to −104.7; P = .02). See Supplementary Table 2 for a full list of effect estimates from our linear mixed model.

DISCUSSION

This study presents results on long-term follow-up assessment of lung function in individuals with COVID-19. By comparing longitudinal changes in individuals with mild COVID-19 and uninfected controls, our results indicate that mild COVID-19 leads to a significant excess decline in lung function during the first 6 months after infection and a slower decline thereafter, indicating lasting decline with limited recovery. However, we observed that the acceleration of the decline slowed and eventually followed the expected annual age-associated decline approximately 6 months postinfection.

A limited number of studies have investigated the long-term lung function trajectories after COVID-19 in hospitalized and nonhospitalized individuals 2 years after acute infection. The studies report conflicting results. One followed 288 hospitalized individuals at 6 months, 1 year, and 2 years [21]. Of these, 45 participants did not receive oxygen upon admission and therefore may be comparable to our study population. Interestingly, they found an unadjusted increase in FVC from admission to 6 months after infection of 10 mL, suggesting an initial recovery phase in the first 6 months. However, from the 12- to 24-month follow-up, they observed an unadjusted decline of 50 mL interpreted as an age-related decline. Another study on 144 hospitalized individuals reports a small but significant unadjusted increase in both FEV₁ and FVC from 6 to 24 months across all levels of severity [22]. It is worth noting that both studies had no information on lung function volumes prior to infection, and the reported lung function volumes were not adjusted for factors that impact lung function, features that may explain the discrepancy between the studies. Moreover, the studies did not include unexposed controls, which may limit the ability to assess the effect of COVID-19. In contrast, our study included matched controls and observed an adjusted decline of 13 mL per year and 33 mL in FEV₁ and FVC from prior to infection to 6 months, and seem to resume age-related decline from 6 to 24 months after infection. Our results suggest that COVID-19 results in lasting lung impairment with limited recovery.

In our study cohort, there was an increase in predicted DL_CO of 3.9% over the approximately 18 months between measurements, suggesting a potential improvement, which is consistent with studies conducted in hospitalized individuals [22–25]. A 2-year follow-up study utilizing computed tomography (CT) chest scans and DL_co measurements at 6, 12, and 24 months reported an increase in DL_co from 80% to 84% between the 6- and 24-month examination [22]. Notably, this study observed the most significant improvement in individuals with nonfibrotic interstitial abnormalities, such as ground glass opacities, and reported gradual improvement over time up to 2 years, concluding that the observed improvement in DL_co was mainly due to the resolution of these radiologic abnormalities. In this context, we have previously shown that individuals from this study cohort showed radiologic lesions at the first follow-up at 5.4 months [26]. The improvement in DLco reported here might be due resolution of these abnormalities, although we do not have chest CT scans at the 24-month examination to verify this. Conversely, the study by Zhang et al investigating DL_CO trajectories reported conflicting results [21]. They observed a median absolute decrease in percent predicted DL_CO of 2.2 across all levels of disease severity in hospitalized individuals from admission to the 2-year follow-up period [21]. However, it is important to note that the participants in Zhang et al's study were more severely ill, which could account for the different trajectories observed.

Despite demonstrating a mean increase in DL_co from 6 to 24 months after infection, a corresponding decrease in dyspnea was not observed. At 24 months after infection, only 1 of 7 individuals who reported dyspnea for the first time 6 months after infection did not experience dyspnea. Furthermore, it is notable that participants reported dyspnea prior to infection, emphasizing both the variability in respiratory symptoms and importance of preinfection information. Previous research at mid- to long-term follow-up has demonstrated a decrease in respiratory symptoms over time simultaneously with an improvement in DL_co [27, 28]. These conflicting results could be due to the limited number of participants reporting dyspnea. Furthermore, the association of DL_co and respiratory symptoms remains elusive and future research with larger cohorts and more comprehensive assessments of respiratory symptoms is warranted to gain a clearer understanding of the relationship between DL_co and dyspnea.

Our study has several strengths. First, it is a prospective cohort study with lung function measured at baseline and followed up for 2 years. This gave us a unique opportunity for a more detailed description of the before and long-term lung function trajectories in relation to infection. Second, matched uninfected controls were included, which made it possible to estimate whether any decline in lung function was due to natural age-related decline or related to infection. Last, the study focused on nonhospitalized individuals, making it more generalizable as most individuals experience mild infection. However, there were also several limitations to the study. In terms of generalizability, a limitation is that the entire study cohort was Caucasian, thereby constraining the applicability of the results to a more diverse population. Another limitation is that almost 50% of participants were lost to follow-up from the 6- to 24-month examination, which may have introduced nonresponse bias. However, sensitivity analysis indicated that the full cohort had the same decline from prior infection to 6 months after infection, suggesting that the sample is representative of the full cohort making this a minor concern. Another limitation is the study size, which limits the ability to perform subgroup analyses among hospitalized patients, individuals with lung disease, or those with varying smoking history and symptom duration. Such analyses could have provided deeper insights into potential diverse trajectories and the identification of factors contributing to a more rapid decline in individuals with mild infection. Additionally, there was no information on diffusion capacity of the lungs (DL_co) and total lung capacity prior to infection, which made it difficult to estimate temporal changes, but this is a defect in nearly all respiratory disease studies.

In conclusion, our study found evidence of long-term lung function impairment among individuals who had recovered from mild COVID-19. The decline in dynamic lung volumes was greater in affected individuals compared to matched controls and persisted up to 2 years after the initial infection. Further research is needed to fully understand the implications of these findings, particularly as the global burden of COVID-19 continues.

Supplementary Data

Supplementary materials are available at The Journal of Infectious Diseases online (http://jid.oxfordjournals.org/). Supplementary materials consist of data provided by the author that are published to benefit the reader. The posted materials are not copyedited. The contents of all supplementary data are the sole responsibility of the authors. Questions or messages regarding errors should be addressed to the author.

Notes

Author contributions. A. R., S. A., and T. B. conceived the idea and planned the study. K. I. collected the data and performed the statistical analysis. K. I., A. R., and T. B. wrote the draft manuscript. All authors contributed to subsequent writing, critical revisions, and approved the final version of the manuscript. T. B. and B. G. N. obtained funding and provided administrative, technical, and material support.

Acknowledgments. We thank all individuals with COVID-19 for their participation.

Financial support. This work was supported by the Research Council of Rigshospitalet; AP Møller og Hustru Chastine McKinney Møllers Fond; and the Brodie Foundation.

References

Author notes