WHAT IS ALREADY KNOWN ON THIS TOPICWHAT THIS STUDY ADDS

Using the Swedish nationwide, population-based DISCOVERY cohort, LTOT is associated with reduced rates of both total and hospitalised acute exacerbations and all-cause hospitalisations in patients with chronic obstructive pulmonary disease (COPD) with and without baseline hypercapnia. Similar observations were noted in patients with interstitial lung disease (ILD) and pulmonary hypertension (PH) who had over 12 months of follow-up.

HOW THIS STUDY MIGHT AFFECT RESEARCH, PRACTICE OR POLICY

Our results provide supporting evidence on favourable impact of LTOT for treating resting hypoxaemia in COPD in reducing acute exacerbations and all-cause hospitalisations, which are important health outcomes in these populations. Furthermore, use of LTOT does not appear to adversely increase the risk of acute exacerbations or hospital burden in patients with hypercapnic COPD. In patients with ILD and PH, there may be beneficial effects in those with expected prognosis of ≥12 months.

INTRODUCTION

Chronic respiratory diseases are a major health burden worldwide, affecting over 7% of the global population with approximately three million deaths each year.1 2 Resting hypoxaemia is a key clinical manifestation of patients with advanced chronic respiratory diseases. The prevalence of chronic resting hypoxaemia varies across different disease groups, but is consistently associated with worse prognosis and increased symptom burden.3 4

Long-term oxygen therapy (LTOT) is recommended for patients with chronic severe resting hypoxaemia, which is prescribed for home use of at least 15 hours per day.5 Two pivotal but small randomised controlled trials from the 1980s demonstrated survival benefits of LTOT in patients with chronic obstructive pulmonary disease (COPD) and severe resting hypoxaemia,6 7 which is now extrapolated to the management of people with other chronic diseases, including interstitial lung disease (ILD) and pulmonary hypertension (PH).8–12 Despite widespread use in these and other diseases, the effects of LTOT on acute exacerbation and healthcare utilisation remain uncertain.

The aim of this study was to evaluate the impact of LTOT initiation on acute exacerbations of the underlying chronic respiratory disease, hospitalisations and outpatient visits among patients with COPD, ILD, and PH. LTOT was hypothesised to be associated with reduced total and hospitalised acute exacerbations, annualised rates and cumulative lengths of stay of all-cause hospitalisations, and annualised rates of all-cause outpatient visits in patients with COPD, ILD and PH. We further hypothesised that among patients with COPD, there would be a similar reduction in acute exacerbations and hospital burden following the initiation of LTOT in both the presence and absence of hypercapnia.

MethodsStudy population

This longitudinal study was conducted using the nationwide, population-based DISCOVERY cohort consisting of the Swedish National Quality Registry of Respiratory Failure (Swedevox) with governmental and other Swedish national registries (online supplemental e-Methods 1).13 We included consecutive patients aged ≥18 years who were initiated on LTOT between 2000 and 2018 for physician-diagnosed COPD, ILD or PH with a follow-up duration of at least 3 months. There was no restriction on the underlying cause of COPD, ILD or PH. The guideline recommended prescribing criteria for LTOT included resting arterial blood gas (ABG) on room air showing partial pressure of oxygen ≤55 mm Hg or 56 to 59 mm Hg with evidence of polycythaemia, cor pulmonale, and/or PH.11 The COPD cohort was further subdivided into hypercapnic (≥45 mm Hg) and non-hypercapnic (<45 mm Hg) groups using baseline arterial partial pressure of carbon dioxide (breathing air). For patients with multiple episodes of LTOT initiation, data from the latest episode were used.

Outcome evaluation

Hospital burden was evaluated in the year prior to and up to 12 months following the initiation of LTOT. Outcomes included total and hospitalised acute exacerbations of the underlying disease responsible for LTOT initiation, all-cause hospitalisations and all-cause outpatient visits. Non-hospitalised acute exacerbations of the underlying disease were determined based on the medication prescription records on the National Prescribed Drugs Registry13 as follows for evaluation of total acute exacerbations:

COPD and ILD: initiation of a prednisolone-equivalent dose of ≥20 mg/day or increased dose by ≥2 times compared with the preceding month for up to 2 weeks of use;

PH: initiation or increased dose of furosemide for up to 2 weeks, with or without concurrent initiation of thiazide or potassium-sparing diuretics.

Medication prescriptions that were linked to hospitalisations were excluded in the definition of a non-hospitalised acute exacerbation. Of note, the evaluation of non-hospitalised acute exacerbations was from July 2005 onwards, when the National Prescribed Drugs Registry was commenced. Hospitalised acute exacerbations were identified using ICD codes from the Swedish National Patient Registry (online supplemental e-Table 1). For acute exacerbation-related and all-cause hospitalisations, the lengths of inpatient stay were also collected.

Additional data

Additional extracted data from the Swedevox Registry included patient demographics, smoking status, LTOT initiation date and indication, baseline ABG and lung function, disease stability status and WHO performance status at LTOT initiation, and concurrent use of ambulatory oxygen therapy and long-term mechanical ventilation. The disease stability status at LTOT initiation was classified as chronic stable disease or postacute illness. These data have previously been validated for accuracy.14

Statistical analysis

Statistical analyses were performed using Stata (V.17, StataCorp, USA). Summary statistics are presented as frequency (%) for categorical data and mean±SD or median (IQR) for continuous data, depending on the distribution.

Primary analyses

To account for varying follow-up durations, events of acute exacerbations, hospitalisations and outpatient visits were annualised for comparison before and after LTOT initiation and then modelled using the multilevel negative binomial regression models with random intercept at individual level. This approach was chosen due to overdispersion of outcome variables (online supplemental e-Methods 2). Baseline assessments were evaluated at the time of LTOT initiation for the preceding 12 months. After the initiation of LTOT, outcomes were evaluated for up to 12 months of follow-up. For patients who ceased LTOT within 12 months, the outcomes were divided by the length of follow-up in days and multiplied by 365. The proportions of patients with ≥1 acute exacerbation, hospitalisation and outpatient visit before and after LTOT initiation were compared using the McNemar test. Adjusted analyses were performed by including prespecified covariates: age, sex, body mass index, WHO performance status, the year of LTOT initiation and length of follow-up, with the addition of the presence of hypercapnia for the COPD cohort. Concomitant use of non-invasive ventilation was also adjusted for in the models evaluating the COPD cohort with and without hypercapnia. Complete case analyses were performed with no imputation for missing data.

Subgroup and sensitivity analyses

Subgroup analyses were performed according to disease stability status at LTOT initiation (chronic stable disease vs postacute illness), adherence to LTOT eligibility (fulfilling and not fulfilling the prescribing criteria), and the year of LTOT initiation (2000 to 2009 vs 2010 to 2017). Sensitivity analyses were conducted for patients with 12 months of follow-up.

ResultsPatient characteristics

A total of 10 134 patients with COPD, 2507 with ILD, and 850 with PH were included (figure 1). Patients of different disease cohorts had comparable age, severity of hypoxaemia, disease stability status, and WHO performance status at LTOT initiation (table 1). The COPD cohort included a greater percentage of ever-smokers, had worse spirometric measurements and was more likely to have hypercapnia at LTOT initiation than the ILD and PH cohorts. The ILD cohort included a smaller percentage of females, compared with those with COPD and PH. Patients with less than 3 months of follow-up (online supplemental e-Table 2) and who had hypercapnic COPD (online supplemental e-Table 3) were more likely to have postacute illness at LTOT initiation, compared with the overall cohort and patients without hypercapnic COPD, respectively. Over the years, there was increasing proportion of LTOT initiation for ILD and PH over the years (online supplemental e-Figure 1). Between 2% and 6% of patients for different disease cohorts did not meet the prescribing eligibility criteria for LTOT (table 1). There were 7430 (73%), 1442 (58%) and 550 (65%) patients with COPD, ILD and PH, respectively, with 12 months of follow-up. The median follow-up durations were 1.8 years for patients with COPD, 1.1 years for patients with ILD and 1.3 years for patients with PH, with 1-year mortality rates post-LTOT initiation of 25%, 44% and 32%, respectively.

Figure 1

Study flow diagram. COPD, chronic obstructive pulmonary disease; ILD, interstitial lung disease; PH, pulmonary hypertension.

Table 1

Baseline patient characteristics

Acute exacerbation of underlying disease

Annualised rates of total and hospitalised acute exacerbations of the underlying disease in patients with COPD declined after LTOT initiation, with increment for those with ILD and PH (table 2 and online supplemental e-Table 4). There was a decline in annualised lengths of hospital stay for acute exacerbations following LTOT initiation in the COPD cohort only. However, the proportion of patients with ≥1 total and hospitalised acute exacerbations was reduced following LTOT initiation in all cohorts (online supplemental e-Table 5). Adjusted analyses of annualised rates of total and hospitalised acute exacerbations, as well as lengths of hospital stay for acute exacerbations, before and after LTOT initiation showed largely similar findings for the three disease cohorts (table 3 and online supplemental e-Table 6).

Table 2

Incidence rate ratios of unadjusted annualised rates of acute exacerbation of underlying disease and all-cause hospital burden before and after the initiation of LTOT for patients with COPD, ILD and PH

Table 3

Incidence rate ratios of adjusted annualised rates of acute exacerbation of underlying disease and all-cause hospital burden before and after the initiation of LTOT for patients with COPD, ILD and PH

All-cause hospitalisation

Annualised rates of all-cause hospitalisations and lengths of stay decreased following LTOT initiation in patients with COPD, but increased in those with ILD (table 2 and online supplemental e-Table 4). In patients with PH, there was increased annualised rates of all-cause hospitalisations but not the lengths of stay. The number of patients with ≥1 all-cause hospitalisation for all cohorts reduced during the year after the initiation of LTOT (online supplemental e-Table 5). In adjusted analyses, similar annualised rates and lengths of stay for all-cause hospitalisations before and after LTOT initiation for all cohorts were observed (table 3 and online supplemental e-Table 6).

Hypercapnic and non-hypercapnic COPD groups

Among the COPD cohort, both hypercapnic and non-hypercapnic groups had a decline in annualised rates of total acute COPD exacerbations, as well as annualised rates of and lengths of stay for hospitalised acute exacerbations following LTOT initiation (online supplemental e-Table 7). The numbers of patients with ≥1 total and hospitalised acute exacerbation of COPD were reduced during the year following the initiation of LTOT for both groups (online supplemental e-Table 8). Following LTOT initiation, both groups also had a decline in annualised rates of and lengths of stay for all-cause hospitalisations, with increased annualised rates of all-cause outpatient visits (online supplemental e-Table 7). On adjusted analyses, similar observations of reduced acute exacerbations and all-cause hospitalisations with increased all-cause outpatient visits following LTOT initiation for both groups (online supplemental e-Table 9).

Subgroup and sensitivity analyses

Subgroup analyses according to the disease stability status at LTOT initiation, adherence to LTOT eligibility, the year of LTOT initiation and concurrent use of ambulatory oxygen therapy are presented in online supplemental e-Tables 10 and 11. Similar observations were seen for different subgroups analysed for all disease cohorts, with few exceptions primarily in those who did not meet LTOT eligibility criteria. Sensitivity analyses for patients with 12-month follow-up after LTOT initiation revealed similar results compared with the primary analyses for patients with COPD (online supplemental e-Tables 12 and 13). In patients with ILD and PH, there were reduced annualised rates of acute exacerbations and all-cause hospitalisations.

Discussion

While LTOT has been used for treating severe resting hypoxaemia for decades, high-level evidence of its therapeutic effects from randomised controlled trials remains sparse given the ethical implications of withholding treatment. Using nationwide, population-based data, the main finding of this study is that initiation of LTOT was associated with reduced disease-specific acute exacerbations and all-cause hospitalisations in patients with COPD who had severe resting hypoxaemia, as well as in those with ILD and PH who had follow-up of ≥12 months. Similar therapeutic benefits were observed in patients with hypercapnic and non-hypercapnic COPD. This was contrasted with an increase in all-cause outpatient visits across all cohorts following LTOT initiation. Reduced proportions of patients with ≥1 acute exacerbation and all-cause hospitalisation were consistently observed across the disease cohorts, which could be influenced by decreased follow-up durations due to the increased attrition rates in those with ILD and PH.

Over 540 million people are affected by chronic respiratory diseases, with COPD and ILD being among the most prevalent three diseases.1 Up to 70 million people are estimated to have PH.15 16 Acute exacerbations are major events in COPD, ILD and PH with substantial risk of mortality, particularly in patients with greater disease severity.17–20 In-hospital mortality for hospitalised acute exacerbations is over 4% in COPD,17 over 50% in ILD,18 19 and between 14 and 100% in PH.20 Acute exacerbations also cause substantial morbidity with worsened symptom burden and quality of life in patients with COPD, ILD and PH.17 18 20 Our study is the first to show that the use of LTOT is associated with a reduction in both total and hospitalised acute exacerbations in patients with COPD, as well as a subset of patients with ILD and PH with ≥12 months of follow-up. This suggests potential benefits of LTOT beyond mortality, including significant cost savings from the reduction in hospital admissions and length of hospital stay, as well as likely improved symptoms and quality of life as a result of reduced total and severe acute exacerbations.

Healthcare utilisation is doubled in patients with COPD and ILD compared with age-matched and sex-matched controls.21 22 Pulmonary hypertension consists of five groups of different aetiology with varying healthcare utilisation that is directly associated with disease severity.23 24 Our findings confirm previous studies of reduced all-cause hospitalisations with LTOT in smaller COPD populations,25–27 with additional evidence of similar effects in selected patients with ILD and PH. Increments in all-cause outpatient visits were noted across different disease cohorts. This may reflect the need for additional support with LTOT and outpatient disease management in this population. In a previous survey of patients from the Swedevox Registry, over 70% experienced adverse events related to LTOT use, with common ones being reduced physical activity, dry mouth, nasal congestion or dryness, and increased fatigue.28 Of note, there is no requirement for routine outpatient visits following LTOT initiation in Sweden. This may also indicate that LTOT allows managing acute exacerbation as outpatients in some patients, or opportunities to address different health issues during outpatient visits for LTOT management. LTOT has been reported to account for between 5% and 53% of total healthcare costs in patients with COPD.29 30 This variation is due to differences in COPD disease severity of the study populations, local healthcare costs and the cost categories included in the evaluation. Nevertheless, hospitalisations are the primary driver of healthcare costs,31 including in patients with COPD,21 ILD32 and PH.23 24 The shift in the use of healthcare utilisation from hospitalisation to LTOT use and outpatient visits is likely favourable for the total healthcare costs. In-depth evaluations of factors for increased all-cause outpatient visits and cost-effectiveness are needed to allow the delivery of optimal clinical care in patients using LTOT.

Chronic hypercapnia is present in up to 30% of patients with severe COPD.33–35 Oxygen-induced hypercapnia is a complex phenomenon that is caused by ventilation-perfusion mismatching, changes in minute ventilation, and the Halden effect with a rightward shift of the carbon dioxide dissociation curve.36 37 Uncontrolled supplemental oxygen therapy during COPD exacerbations is associated with increased mortality.38 39 After LTOT initiation, both hypercapnic and non-hypercapnic COPD cohorts had similar reduction in total and hospitalised acute exacerbations, as well as all-cause hospitalisations and outpatient visits. The benefits appear to be more pronounced in patients with hypercapnia on adjusted analyses. These findings support the safety of LTOT use in patients with mild-to-moderate hypercapnic COPD. Nevertheless, careful assessment and monitoring are encouraged to ensure the absence of worsening hypercapnia or development of respiratory acidosis with LTOT in this population.40

Treatment effects of LTOT in patients who remain hypoxaemic postacute illness or without severe resting hypoxaemia are uncertain. Both groups had reduced acute exacerbation-related and all-cause hospitalisations following LTOT initiation in patients with COPD, but not in those with ILD and PH. Hypoxaemia following acute illness or hospitalisation is often limited to only a few days to a few weeks, resolving in up to 80% of patients at subsequent assessment between 30 and 90 days.41 42 In the LOTT trial of patients with COPD and moderate resting hypoxaemia, LTOT did not affect survival, hospitalisations and acute exacerbations.43 Importantly, there are potential adverse effects with LTOT beyond the associated cost, including physical and psychological impacts and patient safety considerations such as falls and burn injuries.5 40 Hence, reassessments following LTOT initiation postacute illness are necessary, while prescribing LTOT in patients without severe resting hypoxaemia should generally be discouraged and carefully considered in individual cases.

A strength of this study is the use of national population data with a large number of patients for the three disease cohorts. There were no exclusion criteria related to health insurance status or geographical location. There are some limitations. Definitions for non-hospitalised and hospitalised acute exacerbations were based on medication prescription records and ICD codes, respectively, which can cause diagnostic inaccuracy with either underestimation or overestimation due to misattribution of medication indication or hospital diagnosis. However, these biases should affect data for both before and after LTOT initiation equally, and should not affect the numbers of all-cause hospitalisations and outpatient visits. Annualised rates of events of interest were calculated and estimated for patients with a follow-up duration of less than 12 months. Of note, sensitivity analyses using patients with 12-month follow-up revealed similar findings. Patients with limited follow-up of less than 3 months were excluded, who may have different patterns of acute exacerbations and healthcare utilisation. Concomitant disease-specific treatments that may affect health outcomes could not be determined in this study. Subgroup analyses based on two separate enrolment periods with likely different contemporary treatments did not identify significant differences. Causes for outpatient visits, the number of general practitioner visits, and the cost for community or home care support could not be assessed, which warrants evaluation in future studies.

Conclusion

This large nationwide cohort study provides supporting evidence on favourable impact of LTOT for treatment of resting hypoxaemia in COPD, as well as selected cases of ILD and PH, in reducing acute exacerbations and all-cause hospitalisations, with an increase in outpatient visits that likely represents greater ability to manage acute exacerbations in an outpatient setting when patients are using LTOT. Use of LTOT does not appear to adversely increase the risk of acute exacerbations or hospital burden in patients with hypercapnic COPD. These findings suggest potential improvement in health outcomes and healthcare costs with LTOT in patients with advanced chronic respiratory diseases. Cost-benefit analysis is needed to fully elucidate the value of LTOT to inform clinical practice and guidelines.

Data availability statement

Data are available upon reasonable request.

Ethics statementsPatient consent for publication

Not applicable.

Ethics approval

This study involves human participants and was approved by Swedish Ethical Review Board (DNr: 2018/51). Participants gave informed consent to participate in the study before taking part.

Acknowledgments

YHK receives fellowship support from the National Health and Medical Research Council Investigator Grant (ID: 2008255). ZA was supported by the Swedish Heart-Lung Foundation (ID: 20200295). ME was supported by an unrestricted grant from the Swedish Research Council (ID: 2019-02081).