This article is mentioned in:

Introduction

In the USA, chronic obstructive pulmonary disease (COPD) affects 7% individuals in the age group of 25–75 years and is the 4th leading cause of mortality in North America and Europe (1,2). This chronic progressive disease results in the gradual loss of lung function, eventually leading to hypoxemia and dyspnea, with patients complaining of chronic cough, sputum production and shortness of breath (3). Dyspnea from air-flow limitation usually restricts patients with COPD from performing strenuous physical activities with the resultant loss of strength and the endurance of skeletal muscles (4). Exercise intolerance is a characteristic of COPD and is observed in the early stages of the disease. The management of all patients with COPD, functionally disabled by breathlessness, includes pulmonary rehabilitation (PR) (5,6). An integral component of PR is physical training, which can lead to significant improvements in exercise capacity, breathlessness, fatigue and health-related quality of life outcomes of patients with COPD (7). PR benefits individuals by decreasing minute ventilation for any given external work possibly by improving the metabolic capacity of skeletal muscles (8).

Oxygen supplementation has been an ergogenic aid for patients with COPD. Long-term oxygen therapy has been shown to be beneficial for patients with chronic resting hypoxemia (9). Ambulatory oxygen has also been used for the treatment of patients who experience hypoxemia during routine daily activities, but are normoxemic at rest (10). Several clinical trials have investigated the role of supplemental oxygen during exercise training in PR programs (11–13). Patients with exercise-induced desaturation may not tolerate high-intensity exercise and may require reduced intensity training during PR, thereby limiting its effectiveness (5). It is postulated that oxygen may enhance the exercise training of patients with COPD (11). The mechanism of supplemental oxygen in increasing exercise performance has been attributed to decreased ventilatory response, possibly due to a blunted carotid body drive or delayed lactic acidosis (14). The results of randomized controlled trials (RCTs) however, have been conflicting (13,15). A Cochrane meta-analysis in 2007 reviewing 5 RCTs, reported limited evidence for oxygen supplementation during exercise training for individuals with COPD (2). There is thus a need for an updated systematic review and meta-analysis investigating the role of oxygen therapy during exercise training for patients with COPD.

Data and methods

Selection criteria

The primary objective of this systematic review was to investigate the role of supplemental oxygen in improving outcomes following exercise training in patients with COPD. This study was conducted in line with the Preferred Reporting Items for Systematic Reviews and Meta-analyses (PRISMA) statement (16) and guidelines of the Cochrane Handbook for Systematic Reviews of Intervention (17). Only RCTs evaluating the efficacy of supplementary oxygen therapy during the exercise training of patients with COPD were included. Abstracts, non-English language studies, uncontrolled studies and retrospective studies were excluded. The detailed inclusion and exclusion criteria applied for participants, interventions and outcomes are presented in Table I.

Table I. Eligibility criteria for inclusion in the systematic review.

Table I.

Eligibility criteria for inclusion in the systematic review.

	Participants	Intervention	Outcomes
Included	Adult patients with COPD with one of the following:	i) For the study group: Oxygen therapy provided	Any one of the following:
	i) Best recorded EV1/FVC ratio <0.7; ii) best recorded	via wall units/portable oxygen cylinders/Liquid	i) Maximum exercise capacity or incremental exercise
	FEV1 <80%	cannisters. For the control group: Compressed air	capacity or timed walk tests.
		or room air.	ii) Dyspnoea scores
		ii) Intervention provided during exercise training	iii) Quality of life outcomes
Excluded	Patients with COPD on long-term oxygen therapy	i) Intervention not used during exercise therapy.	Not studying any of the above-mentioned outcomes
		ii) Long-term oxygen therapy like for daily home
		activities

[i] COPD, chronic obstructive pulmonary disease; FVC, forced vital capacity; FEV1, forced expiratory volume in the first second.

Search strategy

PubMed, Scopus, Cochrane Central Register of Controlled Trials (CENTRAL) and Google scholar databases (first 100 results) were electronically searched for articles published up to May, 2019. Key words used in various combinations were as follows: Oxygen therapy [MeSH], oxygen [MeSH], supplemental oxygen [Free text], chronic obstructive pulmonary disease [MeSH], chronic obstructive lung disease [MeSH], chronic obstructive airway disease [MeSH], pulmonary rehabilitation [Free text], exercise [MeSH], dyspnea [MeSH], quality of life [MeSH] and physical endurance [MeSH]. References of included studies and review articles were analyzed for the identification of any additional studies.

Collection of data and analysis

Two independent reviewers examined potentially eligible studies for inclusion in the review. Following the removal of duplicates, studies were scrutinized by their title and abstracts. Full-texts of selected articles were then scanned for inclusion. Any difference in opinion was resolved by discussion. The following data from the included trials were extracted: Authors, publication year, study type, sample size, demographic data, interventions used, exercise protocol, outcomes assessed and study conclusions.

Quality assessment

The Cochrane Collaboration risk assessment tool for RCTs was used for assessing the risk of bias (18). Seven criteria were evaluated for each study: Random sequence generation, allocation concealment, blinding of participants and personnel, blinding of outcome assessment, incomplete outcome data, selective outcome reporting and other biases. The included studies were judged for each item and rated as ‘high risk’, ‘low risk’ or ‘unclear risk’.

Statistical analysis

Several variable exercise tests have been used for patients with COPD, such as maximal/incremental/progressive exercise tests, constant power/sub-maximal/endurance exercise tests, functional exercise tests and shuttle walk tests. Only results from the same exercise protocol were combined. Meta-analysis was conducted only if at least 3 studies reported data on the same scale. In studies where exercise outcomes were measured on oxygen and room air, only room air values were considered. Outcome data extracted were entered into Review Manager [RevMan, version 5.3; Nordic Cochrane Centre (Cochrane Collaboration), Copenhagen, Denmark, 2014] for quantitative analysis. Changes in outcome scores were used for the meta-analysis. In studies where change scores were missing, the following equation was used for calculating the change in mean and standard deviation (SD) scores: i) Mean (change) = mean (after)-mean (before); ii) SD (change) = square root {[SD2 (after)-SD2 (before)]/2}.

Considering the heterogeneity amongst studies, a random-effects model was used to calculate the pooled effect size. The mean difference (SMD) with 95% confidence interval (CI) was used for combining the data. Heterogeneity was calculated using the I2 statistic. I2 values of 25–50% represented low, values of 50–75% medium and >75% represented substantial heterogeneity.

Results

A total of 792 studies were screened by their abstracts (Fig. 1). Out of the 12 studies retrieved for full-text analysis, 5 were excluded. The reasons for exclusion were as follows: The use of non-invasive ventilation in the control group (19), the use of supplementary oxygen during home activities (20), crossover trial (21), the absence of a control group (22), and trials on pre-tested subjects showing improvement with supplemental oxygen (23). A total of 7 trials were included in this systematic review and meta-analysis (11–13,15,24–26).

Figure 1. Search results of the study.

Characteristics of the included studies

The details of studies included in the review are presented in Table II. The sample size was <20 patients per group in all studies, apart from 1 (25). Only 1 trial had a large sample of 58 patients in the study and 53 patients in the control group (25). Room air was used in the control group in 1 trial (11), while compressed/humidified air was used in the remaining studies. The exercise protocol included cycle ergometry in 3 studies (15,24,26) and treadmill walking in 1 trial (13). Physical training was carried out in 3 sessions/week in all studies apart from 1 study (11), which utilized a 5 session/week protocol. The duration of each session varied from 30 min to 1 h. In 1 trial, utilizing a 24-week PR program, outcomes were assessed at 12 and 24 weeks. To maintain homogeneity with the remaining studies, which conducted a 6–10 week PR program, outcome data of 12 weeks from this trial were included (26).

Table II. Characteristics of included studies.

Table II.

Characteristics of included studies.

		Participants		Age in years, mean (SD)
Author, year	Study type	Study	Control	Study	Control	Gases used	Type of exercise	Sessions per week	Duration of each session	Total duration of training	(Refs.)
Rooyackers et al, 1997	RCT	12	12	63 (5)	59 (13)	S: 4 lpm O2 by nasal cannula	20 min cycling (2 min exercise then 2 min rest), 5 min rowing, 5 min pulley (arm,	5	80 min with rest (51 min exercise)	10 weeks	(11)
						C: Room air	shoulder), 5 min back and abdominals, 3 min isometric strength training of arms and shoulders and legs, 3 min stair climb, 5 min chair sit/stand and slalom walk, 5 min arm weights
Garrod et al, 2000	RCT	11	11	64.3 (NR)	71.6 (NR)	S: 4 lpm O2 by nasal cannula	Upper limb at 1 kg, lower limb at no resistance, fast walk over 10 m, cycle	3	1 h	6 weeks	(12)
						C: 4 lpm compressed air	ergometer unloaded until intolerance
Wadell et al, 2001	RCT	10	10	64.6 (6.1)	68.3 (4.1)	S: 5 lpm O2 by nasal cannula	Treadmill walk	3	30 min	8 weeks	(13)
						C: 5 lpm compressed air
Emtner et al, 2003	RCT	14	15	66 (7)	67 (10)	S: 3 lpm O2 by nasal cannula	Cycle ergometry	3	45 min	7 weeks	(15)
						C: 3 lpm compressed air
Scorsone et al, 2010	RCT	10	10	67 (9)	68 (7)	S: 40% supplemental O2	Cycle ergometry	3	40 min	8 weeks	(24)
						C: Humidified air
Spielmanns et al, 2014	RCT	19	17	65 (8.7)	64 (8.4)	S: 4 lpm O2 by nasal cannula	Cycle ergometry	3	30 min	12 weeks	(26)
						C: 4 lpm compressed air
Alison et al, 2019	RCT	58	53	69 (7)	69 (8)	S: 5 lpm O2 by nasal cannula	Initially, 20 min of treadmill walking and 10 min stationary cycling, increased to	3	40 min	8 weeks	(25)
						C: 5 lpm air	20 min of treadmill walking and 20 min stationary cycling by week 3

[i] SD, Standard Deviation; RCT, randomized controlled trial; S, study group; C, control group; min, minutes; l pm, liters per min.

Outcomes and data analysis

The outcomes assessed varied across studies. Details of the outcome variables and conclusions of individual studies are presented in Table III.

Table III. Outcomes of included studies.

Table III.

Outcomes of included studies.

Author, year	Outcomes assessed	Conclusions	(Refs.)
Rooyackers	i) Maximal cycle ergometry-4 lpm O2 and room air	Pulmonary rehabilitation improved exercise	(11)
et al, 1997	ii) Constant power cycle ergometry-30% O2 and	performance and quality of life in both
	room air	groups. O2 supplementation during the
	iii) 6MWT-4 lpm O2 and room air	training did not add to the effects of training
	iv) Stair climb-up 4, plateau, down 3	on room air.
	v) Weight lift-lift between racks
	vi) Chronic respiratory Disease Questionnaire
	vii) PFT-Spirometry and transfer coefficient for
	carbon monoxide
Garrod	i) Shuttle walk test	Supplemental O2 during training does little to	(12)
et al, 2000	ii) Chronic respiratory Disease Questionnaire	enhance exercise tolerance although there is a
	iii) Hospital Anxiety and Depression Scale	small benefit in terms of dyspnea. Patients with
	iv) London Chest Activity of Daily Living Scale	severe disabling dyspnea may find symptomatic
		relief with supplemental oxygen.
Wadell	i) 6MWT on 5 lpm O2 and 5 lpm compressed	Supplemental O2 did not further improve the	(13)
et al, 2001	air (random order) with a 1 h rest in between	training effect, compared with training with air,
	ii) Arterial blood gas analysis	in patients with chronic obstructive pulmonary disease.
Emtner	i) Maximal cycle ergometry-30% O2 and	Supplemental O2 provided during high-intensity	(15)
et al, 2003	compressed air	training yields higher training intensity and
	ii) Constant power cycle ergometry-30% O2	evidence of gains in exercise tolerance in
	and compressed air	laboratory testing.
	iii) Chronic respiratory disease questionnaire
	iv) SF-36
	v) PFT-spirometry and lung volume
	vi) Arterial blood gas analysis
Scorsone	i) PFT	O2 supplementation does not contribute to	(24)
et al, 2010	ii) Maximal cycle ergometry	improved exercise performance in patients with
	iii) Constant power cycle ergometry	moderate to severe COPD.
	iv) Arterial blood gas analysis
Spielmanns	i) 6MWT-room air	O2 supplemental oxygen during the training	(26)
et al, 2014	ii) Maximal cycle ergometry-room air	program had no additional benefits in improving
	iii) SF-36	quality of life and exercise capacity in subjects
		with moderate-to-severe COPD.
Alison	i) Endurance shuttle walk test	Both O2 and Air groups significantly improved	(25)
et al, 2019	ii) Incremental shuttle walk test	exercise capacity and health related quality of life
	iii) Chronic Respiratory Disease Questionnaire	with no greater benefit from training with
	iv) Dyspnoea-12 Questionnaire	supplemental O2 than with medical air.

[i] lpm, liter per minute; 6MWT, 6-min walking distance test; SF-36, Medical Outcomes Survey Short Form 36 questionnaire; PFT, Pulmonary Function Test.

Maximal exercise capacity

Maximal exercise capacity was tested by cycle ergometry in 4 trials (11,15,24,26). The data of 55 participants in the study group and 54 in the control group were pooled for a meta-analysis. No statistically significant difference was found between the 2 groups in terms of power (random: MD = −2.38; 95% CI, −5.79 to 1.03; P=0.86, I2 = 0%) (Fig. 2). Similarly, the maximum energy expenditure (VO2 max) did not differ significantly between the supplemental oxygen and control groups (random: MD = −0.01; 95% CI, −0.06 to 0.07; P=0.45, I2 = 0%) (Fig. 3). End-of test dyspnea scores were pooled from 3 trials (11,15,24). The results of the meta-analysis indicated no significant advantage of supplemental oxygen in reducing dyspnea scores (random: MD = 0.79; 95% CI, 0.23 to 1.36; P=0.51, I2 = 0%) (Fig. 4). End-of test oxygen saturation (SpO2%) was measured by 2 studies (11,15). Both reported no advantage of supplemental oxygen.

Figure 2. Forrest plot of oxygen vs. control for power outcome of maximal exercise test.

Figure 3. Forrest plot of oxygen vs. control for VO2 outcome of maximal exercise test.

Figure 4. Forrest plot of oxygen vs. control for end-of test dyspnea scores of maximal exercise test.

Constant power exercise capacity

Three trials (11,15,24) used cycle ergometry to measure constant power exercise capacity. Data were not homogenous to be pooled for a meta-analysis. Endurance capacity, measured as an increase in exercise time, increased only in the oxygen group in 1 study (11) (the change was not significant), while in another trial (15), it improved significantly in both groups. Isotime VO2 was recorded by 2 trials (15,24). Both found no significant difference between the 2 groups. One trial (15) found end-of test dyspnea scores to be significantly improved in the oxygen group as compared to the control group. End-of test oxygen saturation (SpO2%), measured by 2 trials (11,15), did not differ significantly between the 2 groups.

Functional exercise capacity

Three studies (11,13,26) measured functional exercise capacity using 6-min walk tests (6MWT). Pooled data of 41 participants in the study group and 39 in the control group indicated no significant differences between the 2 groups (random: MD = −14.93; 95% CI, −32.64 to 2.78; P=0.93, I2 = 0%) (Fig. 5). Two trials (11,13) recorded end-of test dyspnea scores and end-of test oxygen saturation and found no significant difference between the supplemental oxygen and control groups.

Figure 5. Forrest plot of oxygen vs. control for 6-min walk tests.

Shuttle walk tests

Incremental shuttle walk tests were used as an outcome measure in 2 trials (12,25). While 1 study (25) found no significant difference between groups, the other study recorded a significant reduction in end-of test dyspnea scores in the oxygen group (12). Data from the endurance shuttle walk test from 1 trial (25) demonstrated significant benefits in both groups with no significant in-between group difference.

Quality of life

Health-related quality of life was measured using the Chronic Respiratory Disease Questionnaire (CRQ) in 4 studies (11,12,15,25). Data of 95 participants in the study group and 91 participants in the control group were analyzed. The results of meta-analysis of total scores indicated no difference between the 2 groups (random: MD = −0.09; 95% CI, −0.16 to −0.01; P=0.59, I2 = 0%) (Fig. 6).

Figure 6. Forrest plot of oxygen vs. control for quality of life outcome.

Quality of included studies

The authors' judgment of risk of bias of included trials is presented in Fig. 7. A total of 6 of the 7 studies described an adequate randomization method (11–13,15,25,26). Adequate information on allocation concealment was reported in 3 studies (11,12,25). The blinding of participants and outcome assessment was clearly reported in 4 trials (15,24–26). One trial had a significant number of drop-outs from the study (26).

Figure 7. Risk of bias summary.

Discussion

The primary objective of this systematic review and meta-analysis was to analyze the beneficial effect of supplemental oxygen during exercise training in patients with COPD. Overall, the results of this review can be categorized into 3 sub-headings: Exercise capacity, dyspnea scores and health-related quality of life.

Exercise tolerance is an important predictor of morbidity and mortality in COPD patients. A number of exercise testing protocols were utilized by the included studies. While maximal exercise testing measures the participants' response to a gradually increasing workload, a constant power exercise test assesses the participants systemic response to a constant metabolic demand (27). Since maximal exercise capacity is seldom reached during daily activities, the metabolic load of day-to-day activities is more suitably assessed using the constant power exercise test or other functional exercise tests like 6 or 12 MWT and shuttle walk tests (28). These investigations are also more sensitive to changes following intervention for exercise training (2). Another disadvantage of a maximal exercise test is the high level of motivation required on behalf of the participant to achieve the actual maximal limit. Secondly, approximately 6–10% of day-to-day variation in VO2 max can be observed in patients with COPD (29). Maximal exercise test using cycle ergometry was utilized by 4 of the 7 included trials. On pooling of the data, supplemental oxygen during exercise training was not found to be beneficial in improving VO2 max and power output.

In the absence of homogenous data, quantitative analysis was not carried out of the 3 trials reporting constant power exercise tests. The results of endurance capacity, measured as an increase in exercise time wer ambiguous. Emtner et al (15), while reporting a significant improvement in exercise time in both groups, reported better outcomes in the supplemental oxygen group. The authors suggested that the results may have been underestimated as 16 tests in the oxygen-trained group were terminated at 30 min versus only 7 tests in the control group. Rooyackers et al (11) found an increase in the cycling time only in the experimental group with no change in the control group; however, the change not was statistically significant. Outcome measurements at specific time-points (isotime responses) during constant power exercise tests are effort-independent and allow for a more robust comparison of intervention effects (30). While our review indicates no significant change in VO2 max and cardiovascular parameters (heart rate and blood pressure) with supplemental oxygen therapy, Emtner et al (15) documented a slower and deeper pattern of breathing in the oxygen-trained group.

Non-laboratory functional exercise tests, such as 6MWT have been validated to assess the functional status of patients with COPD (31). Not only are they easy to perform and inexpensive, but standardization is also easy for measuring the effect of interventions. The results from the meta-analysis of 6MWT suggest no benefits with the use of supplemental oxygen during exercise training of patients with COPD. Similar results were found by Garrod et al (12) and Alison et al (25) during shuttle walk tests. Although some authors (13) have reported better performance of the oxygen group during training sessions, the outcome assessment on room air failed to demonstrate any benefit despite prolonged training. One possible explanation suggested is that control groups train under slight hypoxemia as compared to the oxygen groups. The ‘slight hypoxemia’ causes physiological stress, which is required for improving physical capacity (13).

Dyspnea is an important symptom that limits exercise tolerance in patients with COPD. The actual role of oxygen in improving dyspnea is complex and not yet well understood (32). Dyspnea was recorded as an outcome variable in majority trials of this review. However, meta-analysis for dyspnea scores measured on the Borg scale was conducted only for the maximal exercise test, which demonstrated no benefit of supplemental oxygen therapy. Garrod et al (12) demonstrated a reduction in end-of test dyspnea scores after shuttle walk tests but without any significant change of other variables like exercise capacity and health-related quality of life outcomes. The isolated dyspnea results may be attributed to the small sample size of the study.

PR has been shown to significantly improve quality of life outcomes in patients with COPD (7). PR programs include an exercise module and an education/psychological support module. While the exact contribution of each module in improving quality of life is unknown, a recent review suggested that self-management educational programs can significantly improve the quality of life of COPD patients (33). Health-related quality of life outcome was measured via CRQ in 4 studies (11,12,15,25) included in the review. Our analysis failed to demonstrate any beneficial effect of supplemental oxygen in improving the quality of life of patients with COPD.

Some limitations of our study need to be elaborated. The strength of any systematic review and meta-analysis depends upon the homogeneity and quality of the included studies. The lack of homogeneity of the included trials is the largest drawback of this study. There was a wide variation in the methodology, intervention, exercise protocol and outcome measurements, which precluded a quantitative analysis for all variables. The inclusion criteria also varied between studies. While majority trials were carried out on participants who desaturate on exercise, studies by Emtner et al (15) and Spielmanns et al (26) focused only on patients with COPD who were normoxemic at rest and exercise. A sub-group analysis analyzing the effect of oxygen on each group was not possible considering the heterogeneity. Secondly, not all studies reported adequate allocation concealment with blinding of participants and outcome assessment. A lack of rigorous methodology could have affected results. Thirdly, the sample size of majority studies was small with less than 20 participants per group.

Nevertheless, this study is an important update over the previous review on this topic (2). Our results indicate that supplemental oxygen during the exercise training of patients with COPD does not improve exercise capacity, dyspnea scores and quality of life. However, the quality of the evidence is weak. Multi-center RCTs with homogenous methodology and intervention are required to provide stronger evidence on this subject.

Acknowledgements

Not applicable.

Funding

No funding was received.

Availability of data and materials

All data generated or analyzed during this study are included in this published article or are available from the corresponding author on reasonable request.

Authors' contributions

YL and FG designed and searched literature. YL prepared the manuscript. FG edited the manuscript. Both authors have read and approved the final manuscript.

Ethics approval and consent to participate

Not applicable.

Patient consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

References