Authors: Paraschos Archontakis Barakakis, Adam Wolfe, Andrei Schwartz, Gabriel J. Hernandez Romero, Vipul Gidwani, Shaylika Chauhan, Shinichi Arizono, Ralph J. Panos, Spyridon Fortis
Categories: Original Research Articles, 1, 13, 18
Source: ERJ Open Research
Exertion-induced desaturation (EID) is a common complication of numerous pulmonary disorders and often treated with supplementary oxygen during exertion. We performed a systematic review and meta-analysis of randomised clinical trials (RCTs) to evaluate the efficacy of supplementary oxygen for EID in pulmonary disorders.
Medline and Embase were systematically searched from July 2022 to June 2023 following PRISMA guidelines. RCTs that met predefined inclusion criteria were included. Means and standard deviations were extracted and standardised mean differences (SMDs), the difference in means between groups divided by the standard deviation, and 95% confidence intervals were calculated. Exercise capacity was the primary outcome; exercise dyspnoea, baseline dyspnoea and quality of life were secondary objectives. The immediate, post-rehabilitation, short-term and ambulatory effects of oxygen supplementation were evaluated.
We included 15 studies in our analysis. Oxygen supplementation to treat adult EID had been investigated for COPD and idiopathic pulmonary fibrosis (IPF) only. Oxygen supplementation was superior to placebo for its immediate effect on exercise capacity for COPD (SMD 0.42, 95% CI 0.15–0.69, I^2^=3%) and IPF (SMD 0.41, 95% CI 0.08–0.75, I^2^=57%) and exercise dyspnoea for COPD (SMD −0.40, 95% CI −0.76–−0.04, I^2^=31%). Sensitivity analysis revealed similar results.
Our study revealed the efficacy of supplemental oxygen for EID and only a positive immediate effect on exercise capacity and dyspnoea, but no improvement in other short-term or long-term measures.
Exertion-induced desaturation (EID) is a complication commonly diagnosed in patients suffering from a range of chronic pulmonary disorders. Such conditions include highly prevalent ones, such as COPD, and others, such as interstitial lung diseases (ILDs) [1, 2]. EID occurs in approximately 40% of patients with normal oxygen saturation at rest and moderate-to-severe COPD [3–5].
EID is diagnosed when arterial oxygen saturation drops below 88% during exertion [1, 6]. The 6-min walk test (6MWT) is standardised by international guidelines and is the most commonly used method to evaluate patients for EID. The test is utilised in randomised clinical trials (RCTs) and day-to-day clinical practice [7].
The identification of EID is of particular importance from both diagnostic and management perspectives. EID has prognostic value and is associated with increased mortality in both COPD and ILD [8, 9]. The finding of EID is frequently followed by the prescription of ambulatory supplemental oxygen during exertion. However, although the prescription of supplemental oxygen during exertion to ameliorate EID is common, the benefit of this practice remains unclear [10] and currently represents a noticeable burden to patients and a substantial cost to healthcare systems, as demonstrated by the costs related to long-term oxygen supplementation [11].
Therefore, we performed a systematic review of the literature and a meta-analysis
This review and meta-analysis project was performed according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [12]. We registered our project with the International Prospective Register of Systematic Reviews (identifier: CRD42022372935).
Two databases (Medline and Embase) were investigated using a pre-defined search algorithm. The algorithm is included in the supplementary material. In addition to the database search, to identify further eligible studies, manual searches of the reference lists of pertinent studies and reviews were performed.
We used the following criteria to determine the inclusion of a 1) RCT study design, 2) adult study participants (>18 years old), 3) study population restricted to chronic and primary pulmonary disorders, 4) clear study definition of EID, 5) use of a pre-determined and standardised exercise test, 6) RCT study design comparing oxygen versus placebo supplementation or no supplementation, and 7) final publication available online in English.
We excluded articles for the following 1) duplicate reports, 2) studies that included participants with relatively easily reversible acute rather than chronic pulmonary disorders (e.g. bacterial pneumonia or pneumothorax) without providing separate data on our population of interest, 3) studies that included participants with chronic but not primarily pulmonary parenchymal disorders causing EID (e.g. pulmonary artery hypertension; an extensive list is available in the supplementary material) without providing separate data on our population of interest, 4) studies with a population with baseline hypoxemia at rest without providing separate data on our population of interest, 5) studies with a population without EID (e.g. exercise dyspnoea without hypoxaemia) without providing separate data on our population of interest, 6) studies investigating two or more levels of oxygen supplementation with a research goal of identifying optimal blood saturation levels, and 7) studies comparing different methods of oxygen supplementation without placebo or no supplementation.
The database and reference review started in November 2022 and was completed in June 2023 by two research teams led by A.W. and A.S. A third independent investigator (P.A.B.) was involved as needed to reach consensus.
Two independent reviewers (A.W. and A.S.) who were blinded to each other extracted data from the included studies using a pre-defined data collection form. Discrepancies were resolved by a third reviewer (P.A.B.). Extracted data included first author details, year of publication, study design, location of study, duration of trial intervention, population enrolled, number participants enrolled, mean age of participants, gender distribution of participants and pack-years of smoking.
Our selected outcomes were clinically meaningful variables. The primary outcome was exercise capacity and secondary outcomes were dyspnoea severity during exercise (exercise dyspnoea), dyspnoea during daily life (baseline dyspnoea) and quality of life (QoL) during treatment. In order to avoid introducing selection bias to our study regarding varying regimes of exercise tolerance assessment, we extracted the results of all relevant tests provided by the authors but used the one they deemed to be primary. If the authors did not designate a specific test to be primary, we used the 6MWT as default as the test proposed by guidelines for EID identification. For our secondary outcomes, we used the exact same methodology with the default options being the Borg scale for exercise dyspnoea, the Chronic Respiratory Disease Questionnaire (CRQ) – Dyspnoea for baseline dyspnoea and St. George's Respiratory Questionnaire (SGRQ) for baseline QoL. From a temporal perspective, we opted to collect data on our outcomes of interest based on relevant testing performed 1) as soon as possible and usually immediately after randomisation and with no other intervention (e.g., rehabilitation) added by the primary research team (immediate effect), 2) after a short-term rehabilitation programme defined as structured rehabilitation for a period of time of 8–12 weeks during which oxygen or placebo/no oxygen was administered with exercise (post-rehabilitation effect) and 3) after short-term ambulatory use of oxygen or placebo/no oxygen (ambulatory effect). For our research purposes, we defined “short-term” as a period between 2 weeks and 12 months after randomisation.
Two independent reviewers (A.S. and A.W.) assessed the RoB of the included studies using the revised Cochrane RoB tool for randomised trials (RoB2) [13]. Discrepancies were resolved by a third reviewer (S.C.). The identification of publication bias was performed via the creation and assessment of funnel plots for each comparison.
For our data synthesis, we utilised a series of established methodologies. First, we collected all different metrics used by the authors of each separate study for each outcome of interest (e.g., exercise duration in seconds and exercise distance in metres to measure exercise capacity). Secondly, we checked that our collected metrics had congruent vectors. For example, we reversed SGRQ scores in order for higher scores to indicate better QoL and have similar meanings to the CRQ scores. Thirdly, we primarily collected results as means and standard deviations (sds). If not available, we calculated those based on numeric data provided by the authors. If these calculations were not available or possible but a t-test numeric result or an exact p-value derived from a t-test was available, we collected those. If the authors provided medians plus range and/or quantiles, we performed calculations based on the methodology provided by Wan et al. [14]. Fourthly, in cases of data collection from a crossover trial, we used the methodology recommended by the Cochrane Handbook for Systematic Reviews of Interventions [15]. Fifthly, our primary analysis grouped studies by type of RCT (parallel or crossover) and performed pooled calculations between studies. For different pulmonary diseases (e.g. COPD and ILD), we did not perform pooled calculations as it was deemed methodologically incorrect. Finally, if a trial provided data via both a parallel and a crossover methodology, we extracted data based on the type of trial stated by the authors.
Effect estimates (e.g. exercise capacity, etc.) were calculated with the use of standardised mean differences (SMDs), defined as the difference in means between groups divided by the sd [15]. As SMDs are not usual measures of exercise capacity, we used the weighted mean sds of 6MWTs of each specific comparison to calculate and present a more intuitive mean difference in metres. A random-effects model was selected a priori because the included studies had heterogeneous study designs and baseline patient characteristics [16]. Forest plots were used to illustrate the individual study findings and the random-effects meta-analysis results. The I^2^ statistic was used to assess heterogeneity among the studies [17, 18]. A cut-off of 50% was used to indicate statistically significant heterogeneity. The Q statistic and the p-value for the Q statistic were also calculated. In addition to our main analysis, a sensitivity analysis was performed by excluding all studies with an RoB assessment of “high”. All statistical analyses were conducted with R version 4.2.1 with R studio version 2022.02.3.
The quality of the evidence was assessed by the GRADE approach and primarily based on the RoB of included RCTs, result imprecision, heterogeneity and publication bias calculations [19].
In total, 8215 records were screened and 142 full-text articles were assessed for eligibility. Of these, 15 studies met all the inclusion criteria and were advanced to qualitative and quantitative analysis [E1–E15] (corresponding references for the included studies are available in the Supplementary material). A PRISMA flow diagram with the selection process depicts this analysis (figure 1).
FIGURE 1 PRISMA flowchart.
Extensive information on the methodology of each study and on the baseline characteristics of their included populations were collected. These data are presented in tables 1 and E1.
Our investigation revealed that adult EID and its treatment with supplemental oxygen has been investigated mainly for two chronic pulmonary disorders, namely COPD and idiopathic pulmonary fibrosis (IPF). Most studies focused on COPD and a limited number on IPF. No studies of other chronic pulmonary disorders were identified.
Oxygen supplementation was delivered by nasal cannula or simple or Venturi-type masks. All methods provided low-flow oxygen. They were effective in correcting exercise desaturation, as documented in Table E1.
To determine the immediate effect of oxygen supplementation on patients with COPD, we identified a total of 232 patients in six studies that collectively demonstrated an increase in exercise capacity in the supplemental oxygen group (SMD 0.42, 95% CI 0.15–0.69, I^2^=3%, GRADE assessment of “low” as per study limitations and publication bias). The mean 6MWT difference was 54.6 m (95% CI 19.5–89.7 m) (figure 2).
FIGURE 2 Effect of oxygen versus placebo supplementation on immediate exercise capacity for patients with COPD. Corresponding references for the studies cited in the figure are available in the supplementary material. 6MWT: 6-min walk test; CPCT: constant power cycle test; DL: DerSimonian and Laird; ESWT: endurance shuttle walk test; O
2: number of patients on oxygen; Pl: number of patients on placebo; RCT: randomised control trial; RoB: risk of bias; SMD: standardised mean difference.
To determine the immediate effect of oxygen supplementation on patients with IPF, we identified a total of 396 patients in four studies who demonstrated an increase in exercise capacity for the group receiving supplemental oxygen (SMD 0.42, 95% CI 0.08–0.75, I^2^=57%, GRADE assessment of “low” as per borderline consistency and publication bias). The mean 6MWT difference was 48.2 m (95% CI 5.5–91 m) (figure 3).
FIGURE 3 Effect of oxygen versus placebo supplementation on immediate exercise capacity for patients with idiopathic pulmonary fibrosis. Corresponding references for the studies cited in the figure are available in the supplementary material. 6MWT: 6-min walk test; CPCT: constant power cycle test; DL: DerSimonian and Laird; IMCT: incremental maximal cycle test; O
2: number of patients on oxygen; Pl: number of patients on placebo; RCT: randomised control trial; RoB: risk of bias; SMD: standardised mean difference
We identified no difference between study participants with COPD receiving or not receiving supplemental oxygen during cardiopulmonary rehabilitation for post-rehabilitation exercise capacity (five studies, 231 total patients, SMD 0.15, 95% CI −0.27–0.57, I^2^=56%, GRADE assessment of “moderate” as per study limitations and borderline consistency). The mean 6MWT difference was 18.3 m (95% CI −33–69.7 m) (figure 4).
FIGURE 4 Effect of oxygen versus placebo supplementation on post-rehabilitation exercise capacity for patients with COPD. Corresponding references for the studies cited in the figure are available in the supplementary material. 6MWT: 6-min walk test; DL: DerSimonian and Laird; ESWT: endurance shuttle walk test; O
2: number of patients on oxygen; Pl: number of patients on placebo; RCT: randomised control trial; RoB: risk of bias; SMD: standardised mean difference.
To assess the short-term, ambulatory effect of oxygen supplementation during exercise on patients with COPD, we identified no difference between the groups in exercise capacity (five studies, 474 total patients, SMD 0.21, 95% CI −0.10–0.53, I^2^=59%, GRADE assessment of “moderate” as per inconsistency). The mean 6MWT difference was 23.3 m (95% CI −11.1–58.9 m) (figure 5).
FIGURE 5 Effect of oxygen versus placebo supplementation on short-term ambulatory exercise capacity for patients with COPD. Corresponding references for the studies cited in the figure are available in the supplementary material. 6MWT: 6-min walk test; A5MWT: Ambulatory 5-min walk test; DL: DerSimonian and Laird; ESWT: endurance shuttle walk test; O
2: number of patients on oxygen; Pl: number of patients on placebo; RCT: randomised control trial; RoB: risk of bias; SMD: standardised mean difference.
To determine the immediate effect of oxygen supplementation on patients with COPD, we identified a total of 200 patients in five studies who demonstrated a decrease in exercise dyspnoea in the supplemental oxygen group (SMD −0.40, 95% CI −0.76–−0.04, I^2^=31%, GRADE assessment of “low” as per study limitations and publication bias) (figure E1A).
No immediate effect benefit for either group of patients with IPF was identified in exercise dyspnoea (four studies, 396 total patients, SMD −0.17, 95% CI −0.59–0.24, I^2^=52%, GRADE assessment of “high” despite borderline consistency) (figure E1B).
We identified no difference between the study participants with COPD receiving or not receiving supplemental oxygen during cardiopulmonary rehabilitation for post-rehabilitation exercise dyspnoea (three studies, 139 total patients, SMD −0.32, 95% CI −0.66–0.01, I^2^=0%, GRADE assessment of “high”) (figure E1C).
To assess the short-term, ambulatory effect of oxygen supplementation during exercise for patients with COPD, we identified no difference in exercise dyspnoea (three studies, 382 total patients, SMD −0.12, 95% CI −0.52– 0.27, I^2^=66%, GRADE assessment of “moderate” as per borderline consistency) (figure E1D).
We identified no difference between the study participants with COPD receiving or not receiving supplemental oxygen during cardiopulmonary rehabilitation for post-rehabilitation baseline dyspnoea (three studies, 168 total patients, SMD 0.12, 95% CI −0.18–0.43, I^2^=0%, GRADE assessment of “high”) (figure E2A).
To assess the effect of short-term, ambulatory oxygen supplementation during exercise for patients with COPD, we identified no difference between study participants receiving or not receiving supplemental oxygen in baseline dyspnoea (one study).
To assess the effect of short-term, ambulatory oxygen supplementation during exercise for patients with IPF, we identified a significant difference in favour of participants receiving supplemental oxygen versus not receiving supplemental oxygen in baseline dyspnoea (one study).
We identified no difference between the study participants receiving or not receiving supplemental oxygen during cardiopulmonary rehabilitation for post-rehabilitation QoL (three studies, 166 total patients, SMD 0.03, 95% CI −0.28–0.33, I^2^=0%, GRADE assessment of “moderate” as per study limitations) (figure E3A).
To assess the effect of short-term ambulatory oxygen supplementation during exercise for patients with COPD, we identified no difference between study participants receiving or not receiving supplemental oxygen in QoL (two studies, 323 total patients, SMD −0.06, 95% CI −0.28–0.15, I^2^=0%, GRADE assessment of “high”) (figure E3B).
To assess the effect of short-term ambulatory oxygen supplementation during exercise for patients with IPF, we identified no difference between study participants receiving or not receiving supplemental oxygen in baseline dyspnoea (one study).
Our analysis excluding studies with an RoB assessment of “high risk” provided results for patients with COPD only. For immediate effect, exercise capacity (four studies, 176 total patients, SMD 0.48, 95% CI 0.18–0.78, I^2^=0%) improved in the supplemental oxygen group. The mean 6MWT difference was 49.6 m (95% CI 18.6–80.6 m). There was no difference in post-rehabilitation exercise capacity (two studies, 115 total patients, SMD 0.11, 95% CI −0.25–0.48, I^2^=0%). The mean 6MWT difference was 7.8 m (95% CI −17.8–34.1 m). There was no difference for short-term, ambulatory exercise capacity (three studies, 382 total patients, SMD 0.13, 95% CI −0.07–0.33, I^2^=0%). The mean 6MWT difference was 14.5 m (95% CI −7.8–36.7 m). For immediate effect, exercise dyspnoea (four studies, 176 total patients, SMD −0.48, 95% CI −0.88–−0.08, I^2^=35%) improved in the supplemental oxygen group. There was no difference in post-rehabilitation exercise dyspnoea (two studies, 115 total patients, SMD −0.29, 95% CI −0.65–0.08, I^2^=0%) or short-term, ambulatory exercise dyspnoea (three studies, 382 total patients, SMD −0.12, 95% CI −0.52–0.27, I^2^=66%). There was no difference in post-rehabilitation baseline dyspnoea (one study). There was no difference in post-rehabilitation QoL (one study) or short-term, ambulatory QoL (one study) (figure E4A–F).
Funnel plots were constructed for all outcomes. Based on the assessment of these figures (figure E5A–F), possible evidence of publication bias in favour of oxygen supplementation was identified in the immediate effect exercise capacity for COPD analysis, immediate effect exercise capacity for IPF analysis, immediate effect exercise capacity for COPD sensitivity analysis and immediate effect exercise dyspnoea for COPD sensitivity analysis.
Our RoB assessment was performed via the RoB2 tool for parallel and for crossover RCTs. The results of our assessment for all included studies can be found in table E2.
This study is a systematic review and meta-analysis of 15 RCTs determining the efficacy of oxygen supplementation for patients with chronic pulmonary disorders causing EID. Our results show that 1) RCTs of the treatment of adult EID with supplemental oxygen have been performed in two chronic pulmonary disorders, namely COPD and IPF, 2) data were available for immediate and longer-term effects for COPD and mainly for immediate effects for IPF, 3) oxygen supplementation increased immediate exercise capacity for COPD and IPF but not post-rehabilitation or short-term, ambulatory exercise capacity for COPD, 4) oxygen supplementation decreased immediate exercise dyspnoea for COPD and not for immediate exercise dyspnoea for IPF or post-rehabilitation or short-term, ambulatory exercise dyspnoea for COPD, 5) oxygen supplementation had no benefit for post-rehabilitation baseline dyspnoea or post-rehabilitation or short-term, ambulatory QoL for COPD, and 6) our sensitivity analyses provided results similar to our main analyses.
Our results add new data to the recommendations offered by national and international guidelines. The 2020 American Thoracic Society guidelines provide a conditional recommendation in favour of the prescription of ambulatory oxygen for patients with both COPD and ILD and isolated diagnosis of EID [20]. For COPD, the guideline authors pool two studies that used the 6MWT to assess exercise capacity, document additional studies that used other tests (e.g. 5-min walk test) and find a benefit of oxygen use. They analyse three studies that used the Borg scale to assess exercise dyspnoea and find a benefit. They finally analyse three studies with the CRQ for QoL, document additional studies using other tests (e.g. STGQ) and find a benefit. The presented evidence on exercise tolerance and exercise dyspnoea pertains to the immediate effect of oxygen and no consideration of oxygen use during pulmonary rehabilitation is included. For ILD, they pool three studies with the 6MWT for the exercise capacity assessment and find a benefit of oxygen use. They analyse three studies with the Borg scale for exercise dyspnoea assessment and find no benefit. In contrast to our study, they include a limited number of studies, pool studies without specifically identifying exercise desaturation as an inclusion criterion (e.g. they include studies with exercise dyspnoea only), studies utilising different modes of oxygen delivery and studies that compare different oxygen supplementation flow rates. Our study has narrower and more specific inclusion criteria, answers a very specific research question and pools a higher number of studies via the methodology documented above.
On the other hand, the 2015 British Thoracic Society (BTS) guidelines provide a recommendation against the routine prescription of oxygen in patients with an isolated diagnosis of EID [21]. They do not pool data but present the relevant literature available at the time and identify the short-term benefits in exercise capacity and exercise dyspnoea of ambulatory oxygen in the experimental setting and the lack of evidence on longer-term benefits. They also assess patient comfort and compliance, reviewing the burdens of ambulatory oxygen use that might lead to lower adherence. Finally, they recommend oxygen use during pulmonary rehabilitation if an exercise capacity improvement is appreciated with oxygen use for the individual. As per the publication year, the 2015 BTS guidelines do not consider primary studies that were published more recently. Our study incorporates the topics addressed in the guideline and then adds the most up-to-date research on the subject.
Except for international guidelines, primary studies and reviews/meta-analyses performed on this topic have also reached mixed conclusions. A part of this research has demonstrated some benefits of oxygen supplementation. These benefits mainly pertain to a possible improvement of exertional capacity and a possible decrease of symptomatic dyspnoea on exertion [22, 23]. There is some evidence that oxygen supplementation can potentially contribute to achieving higher exercise load during exercise sessions [24, 25]. Other research teams do not identify a benefit from oxygen supplementation, especially concerning outcomes of more longer-term nature such as QoL, disease exacerbation risk and mortality [6, 26] [E9, E14]. Finally, a third group of studies has been equivocal, with ambulatory oxygen use associated with no effect on exercise capacity or exercise dyspnoea but a possible beneficial effect on QoL [27]. Our study provides a comprehensive, up-to-date review on the treatment of EID with supplemental oxygen and provides guidance on its possible benefits. In addition to the benefits, a consideration for possible hindrances to the patient (e.g. individual or insurance cost, need for possible lifelong equipment use and strain of moving/carrying it, and possible safety hazards with concurrent smoking) is advised and thus a risk/benefit analysis and a shared decision-making process between the patient and the provider need to be performed prior to ambulatory oxygen prescription.
The mechanisms causing EID have not been definitively established. However, the consensus is that a combination of ventilation–perfusion mismatch and alveolar diffusion limitation coupled with increased oxygen demand caused by elevated muscle consumption during exercise precipitates EID. These pathophysiologic derangements have been better described in patients with COPD but also in patients with ILD, including IPF [28]. Supplemental oxygen augments systemic oxygenation to ameliorate desaturation but does not address these underlying physiologic derangements. Specifically, oxygen supplementation during exercise delays or prevents lactic acidosis provoked by the activated metabolism of working muscle fibres. It has also been associated with the belated onset or amelioration of tachypnoea. This tachypnoea can negatively affect respiratory gas-exchange efficacy by increasing dead-space ventilation and hyperinflation in restrictive and obstructive lung disorders. Finally, oxygen supplementation can have a positive impact on right-heart haemodynamics with delayed or reduced elevation of pulmonary artery pressures provoked by exercise [29–31]. Finally, a recently described pathophysiologic mechanism involves dynamic hyperinflation during exercise. Dynamic hyperinflation represents a transient increase in end-expiratory lung volume caused by the increased minute ventilation and reduced expiratory time provoked by exertion in patients with flow-limited disorders, such as COPD. Interestingly, controlling this phenomenon via a positive exhalatory pressure-generating device seems to be effective in limiting desaturation in a subgroup of individuals with COPD and EID [32–35].
Moderate methodological alignment was demonstrated in the studies included in this meta-analysis. The definition of EID used by the different research groups generally included a threshold value for pulse oximetry that varied between 88% and 92%, and some studies used the additional criterion of a decrease of saturation between 2% and 5% during exertion. Secondly, the test modalities used to diagnose EID were different, with several research teams using the 6MWT but with different types of ergometry. Thirdly, the interventions used to supplement oxygen included either a set flow rate, such as 2 L of nasal cannula, or variable flow rates to achieve a desired saturation during exertion. However, the different techniques adequately met our pre-defined criterion of low flow and were deemed both comparable to each other among the trials and to day-to-day clinical practice. Fourthly, the oxygen supplementation as performed by the researchers effectively ameliorated EID in the oxygen group, as indicated by the adequate group separation of exercise saturation measurements provided by the trials. Fifthly, the methods used in the various studies to quantify outcomes of interest to our research questions were different for exercise capacity assessment but similar for exercise and baseline dyspnoea and QoL. Finally, the duration of rehabilitation and ambulatory use periods as well as the intensity of rehabilitation regimes were also similar. In summary, the pooling of studies was deemed methodologically appropriate.
Our study demonstrates a number of strengths. Firstly, we focused on a specifically defined and very frequently encountered condition in day-to-day clinical practice. Secondly, we strictly adhered to the systematic review methodology throughout our analysis with specific attention to data extraction and numeric data processing. Thirdly, our outcomes were identified as clinically relevant for all practitioners providing care to patients with chronic pulmonary disorders. Finally, we were able to search, collect, screen and analyse a significant number of studies and thus reach a substantial participant population size.
Our study demonstrates certain limitations as well. First, the definition of EID among the included studies did not exactly align with the current, guideline-specified and internationally used definitions. Secondly, the grouping of outcomes was done in a statistically rigorous manner to pool results of different tests, but our final SMD result is not intuitive. Thirdly, our source material had moderate levels of heterogeneity from a methodology and research intervention perspective. We believe that the borderline results in study heterogeneity can be at least partially attributed to the relatively low number of included participants in each study. Fourthly, each separate comparison did not reach a pooled number of participants that would qualify the results as definitive. However, none of the included studies had a dominant effect on our comparisons and, thus, did not dictate our conclusions.
The recognition of EID is of particular clinical importance as it not infrequently leads to the prescription of supplemental oxygen. Our study reveals a significant benefit of oxygen supplementation on exercise capacity in COPD and ILD and acute exertional dyspnoea in COPD but no other outcome improvements including longer duration measures.
Our results reinforce a need for further and more robust investigations of the treatment of EID with supplemental oxygen. We would recommend a high level of adherence to RCT methodology and performance, especially concerning the randomisation and allocation concealment processes, more robust participant sample size goals, and outcomes focusing on long-term outcomes, such as mortality and disease specific or all-cause hospitalisation, while also adding patient-centred outcomes in terms of the burdens and hurdles of ambulatory oxygen use.
Our data are derived from public-domain resources. All data source material that supports the findings of this study is available on Medline and Embase.
Our data are derived from public-domain resources. All data source material that supports the findings of this study is available on Medline and Embase.