To the Editor:

Coronavirus disease (COVID-19) can impair gas exchange (1), but the risk of post-infectious oxygen-dependent chronic respiratory failure is unknown.

We performed a population-based, nationwide study of cumulative incidence, risk factors and clinical course of long-term oxygen therapy (LTOT) after COVID-19, using data from the SCIFI-PEARL (Swedish Covid-19 Investigation for Future Insights – a Population Epidemiology Approach using Register Linkage) study (2). We included all people in Sweden aged ⩾16 years with a laboratory-confirmed severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) infection from January 1, 2020, until August 31, 2021, with no LTOT before the COVID-19 diagnosis. Data on LTOT start any time after COVID-19 until September 30, 2021, were obtained from the Swedish National Registry of Respiratory Failure (Swedevox) (3). In Sweden, home oxygen is prescribed strictly according to the American Thoracic Society (ATS)/European Respiratory Society (ERS) criteria at specific respiratory units, using stationary and portable oxygen concentrators (3).

Risk factors for starting LTOT after COVID-19 were assessed using multivariable logistic regression. Severity of COVID-19 was categorized as mild (no hospitalization), severe (hospitalized without need of ICU), or critical (ICU care), age as 16–49, 50–59, 60–69, and ⩾70 years, and education as primary (compulsory school of 9 years), secondary (2–3 years beyond compulsory school), and tertiary education (university studies). COVID-19 comorbidity risk groups according to the National Board of Health and Welfare (4) included chronic heart, lung, and renal disease, type 2 diabetes with complications, obesity, and hypertension. Chronic lung diseases included chronic obstructive pulmonary disease (COPD; 85%), interstitial lung disease, cystic fibrosis, or chronic respiratory failure from other causes (4). Cumulative incidence of LTOT starts was compared among four periods: January–July 2020 (first wave); August–December 2020 (original SARS-CoV-2 variant of concern [VOC] with standardized COVID-19 treatments including corticosteroids and ventilation strategies [5]); January–March 2021 (alpha variant dominance); and April–September 2021 (delta variant and established mass vaccination in Sweden). In two merged periods—January–December 2020 and January–September 2021, including new VOCs and mass vaccination—differences in cumulative incidence and factors associated with LTOT were analyzed using chi-square tests, interaction analysis by period, and multivariable logistic regression.

In patients starting LTOT after COVID-19, frequencies of death, LTOT discontinuation due to improvement, and ongoing LTOT up to September 30, 2021, were calculated.

During the entire observation period of 21 months, 271 of 992,968 individuals with COVID-19 started LTOT. The overall cumulative incidence was 27 (95% confidence interval [CI], 24–31) per 100,000, decreasing over time from 81 (January–July 2020) to 28 (August–December 2020), 25 (January–March 2021), and 13 (April–September 2021). The difference between 2020 and 2021 was statistically significant: 38 (32–44) versus 20 (16–23) per 100,000, P < 0.0001, as was effect modification by first or second period on the associations of hospitalized COVID-19 infection and chronic respiratory disease, respectively, with LTOT start (data not shown).

Median time from COVID-19 confirmation to LTOT start was 46 (interquartile range [IQR], 30–83) days. Overall and in both time periods, the strongest independent risk factors for LTOT were severe or critical COVID-19, older age, and preexisting chronic respiratory disease (Table 1 and Figure 1). Lower educational level and female sex were associated with higher LTOT risk for the entire observation period, with similar point estimates in both periods (Table 1). In an analysis restricted to hospitalized patients, risk factors for LTOT after COVID-19 (odds ratio; 95% CI) in both merged periods were chronic respiratory disease (5.84; 4.01–8.52 and 8.11; 5.11–12.9) and higher age (60–69 years: 6.12; 2.36–15.9 and 1.71; 1.32–5.60; ⩾70 years: 5.00; 1.95–12.8 and 2.10; 1.02–4.33). During a median follow-up of 150 (IQR, 89–230) days, 181 (67%) remained on LTOT, 42 (15%) died, and 48 (18%) patients discontinued LTOT. In the latter, 67% had no previous respiratory disease, median (IQR) time on LTOT was 89 (50–158) days, and 54% discontinued LTOT within 3 months.

Figure 1. Incidence of long-term oxygen therapy (LTOT) after coronavirus disease (COVID-19). Cumulative incidence of LTOT starts presented as number per 100,000 patients distributed by severity of COVID-19, in an unselected national sample of all patients with laboratory-confirmed COVID-19 in the Swedish population aged 16 years and above, identified January 1, 2020, until August 31, 2021. The bar clusters refer to the periods of January–December 2020, covering the phase of the original variant of COVID-19, and January–September 2021, including new severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) variants of concern and mass vaccination. Chronic respiratory disease (yes/no) and age groups.

Download Figure | Download Powerpoint

[More] [Minimize]

This is the first population-based nationwide study, to our knowledge, evaluating incidence and risk factors for LTOT in patients with laboratory-confirmed COVID-19. In summary, the proportion developing oxygen-dependent chronic respiratory failure after COVID-19 is considerably lower than reported from selected populations (6, 7). The decrease of LTOT incidence during the study period suggests protective effect from improved treatment and vaccination (5).

Chronic respiratory disease is an expected risk factor for LTOT after COVID-19, as it increases risk for chronic hypoxic respiratory failure (3) and increases mortality in COVID-19 (8). The overall number of LTOT starts in Sweden did not increase among 2019 (n = 1,084), 2020 (n = 1,037), and 2021 until last September (n = 830). However, 65 of 160 people (41%) starting LTOT after COVID-19 in 2020 had underlying chronic respiratory disease. We speculate that a substantial proportion of patients with chronic respiratory disease would have been started on LTOT anyway but received LTOT earlier owing to COVID-19. The association of higher age with risk of LTOT after COVID-19 is consistent with the progressive decline of lung function (9). The association of lower education with LTOT is consistent with the association of lower socioeconomic status with worse COVID-19 outcomes (10). The finding that LTOT was slightly more common in women confirms our previous study in which 62% of patients receiving LTOT generally are women (3). The finding that 18% of LTOT patients were able to quit oxygen treatment supports previously reported improvement of lung function in patients followed until 12 months after a COVID-19 infection (1).

Study strengths are the inclusion of all laboratory-confirmed COVID-19 cases in the entire adult population of Sweden, high data validity, and coverage of Swedevox exceeding 90% of patients starting LTOT nationwide (3). A potential limitation is that our study period could not include the most recent VOC, omicron. However, it covers both the initial phase and the establishment of treatment strategies, mass vaccination, and several new VOCs. In addition, follow-up after omicron would be too limited to date.

We conclude that the risk of developing oxygen-dependent chronic respiratory failure after COVID-19 is generally low. The strongest risk factors are severe or critical COVID-19, higher age, and preexisting chronic respiratory disease. During our follow-up, a substantial proportion were able to quit LTOT. Future research should focus on longer-term follow-up of post–COVID-19 oxygen-dependent chronic respiratory failure.

References

Section:

ChooseTop of pageReferences <<CITING ARTICLES