Background
Dyspnoea is a common symptom in chronic obstructive pulmonary disease (COPD). People who are hypoxaemic may be given long-term oxygen relief therapy (LTOT) to improve their life expectancy and quality of life. However, the symptomatic benefit of home oxygen therapy in mildly or non-hypoxaemic people with COPD with dyspnoea who do not meet international funding criteria for LTOT (PaO2< 55 mmHg or other special cases) is unknown.

Objectives
To determine the efficacy of oxygen versus medical air for relief of subjective dyspnoea in mildly or non-hypoxaemic people with COPD who would not otherwise qualify for home oxygen therapy. The main outcome was patient-reported dyspnoea and secondary outcome was exercise tolerance.

Search strategy
We searched the Cochrane Airways Group Register, Cochrane Central Register of Controlled Trials (CENTRAL), MEDLINE and EMBASE, to November 2009, to identify randomised controlled trials. We handsearched reference lists of included articles.

Selection criteria
We only included randomised controlled trials of oxygen versus medical air in mildly or non-hypoxaemic people with COPD. Two review authors independently assessed articles for inclusion.

Data collection and analysis
One review author completed data extraction and methodological quality assessment. A second review author then over-read evidence tables to assess for accuracy.

Main results
Twenty-eight trials on 702 patients met the criteria for inclusion; 18 trials (431 participants) were included in the meta-analysis. Oxygen reduced dyspnoea with a standardised mean difference (SMD) of -0.37 (95% confidence interval (CI) -0.50 to -0.24, P < 0.00001). We observed significant heterogeneity.

Authors' conclusions
Oxygen can relieve dyspnoea in mildly and non-hypoxaemic people with COPD who would not otherwise qualify for home oxygen therapy. Given the significant heterogeneity among the included studies, clinicians should continue to evaluate patients on an individual basis until supporting data from ongoing, large randomised controlled trials are available.