Uronis, H.E., Currow, D.C., McCrory, D.C., Samsa, G.P., & Abernethy, A.P. (2008). Oxygen for relief of dyspnoea in mildly- or non-hypoxaemic patients with cancer: A systematic review and meta-analysis. British Journal of Cancer, 98(2), 294-299.

The objective was to identify articles that evaluate the efficacy of oxygen therapy and medical air for the improvement of dyspnea in patients with cancer.

Databases searched were MEDLINE and EMBASE (1966-December 2006).

Search keywords were MeSH terms including dyspnea (dyspnoea) , oxygen, and palliative care and text words including oxygen, dyspnea, breathlessness, oxygen, and inhalation therapy.

Randomized controlled trials comparing oxygen and medical air in patients with cancer suffering from refractory dyspnea who do not qualify for home oxygen therapy were included.

Studies were excluded if

• Study subjects had a mean PaO₂ less than 55 mmHg or more than 50% of subjects had oxygen saturation less than 88% by pulse oximetry
• Study subjects were already receiving home oxygen therapy
• Study intervention was not oxygen versus placebo
• Method of oxygen delivery was something other than nasal cannula, mouthpiece, or mask
• No dyspnea outcomes were reported.

Studies also were excluded for \"other\" reason (e.g., study articles were not editorial or review article).

A total of 203 citations was reviewed with 54 full-text articles examined. None of the selected studies had evidence of calculations to claim adequate power to answer the questions defined. All five studies included in the analysis were blinded, randomized, controlled crossover trials. Each study was assigned a Jadad score in quality evaluation. External validity was assessed for subject description, detailed intervention description, and adequately reported dyspnea outcomes. Dyspnea ratings as assessed by the Modified Borg’s 0-10 numerical rating scale (NRS), 100 mm visual analog scale (VAS), or 300 mm VAS were converted to standardized mean differences (SMDs). When data pertaining to paired analyses were missing in two periods of crossover trials, standard errors were interpreted. P-values were used to estimate correlations between repeated outcomes when available and when unavailable, the lowest estimate from other studies was used. Meta-analysis was performed only on studies from which means and variances from for dyspnea measurements could be determined from published reports, and effect sizes were reported as SMD with 95% confidence intervals. Statistical significance was determined by a P < 0.05. If additional data were needed, authors were contacted.

Four studies focused on comparing oxygen versus medical air for dyspnea relief, while the fifth compared Heliox28 (an agent that contains 72% helium and 28% oxygen) with oxygen and medical air. In three studies, oxygen was delivered by nasal cannula, and in the other two by face mask. Doses of oxygen ranged from 3-5 L/min. Oxygen was administered at rest in three studies and during a six-minute walk test (6MWT) in two studies.

One hundred thirty-four were included in meta-analysis (148 were analyzed). Studies had a median of 33 participants, with a mean of 29.6 and range of 14-51. The median participant age was 65 years; 39% were female, and no data related to race or ethnicity were available. Subjects presented with the following malignancies: lung cancer (65%) or unspecified cancer with metastasis to lung (15%), breast cancer (5%), colon (3%), and other (i.e., lymphoma, melanoma, sarcoma, carcinoid, skin, bladder, and head and neck) (7%). Baseline oxygen saturation for four of the five was reported. Baseline dyspnea at rest, as provided by three of the studies, was 0 (modified Borg), 5 mm (NRS), and 59 mm by 100 mm (VAS).

Oxygen therapy was shown ineffective in the relief of dyspnea in mildly or non-hypoxemic patients with cancer (SMD = -0.09, 95% CI -0.22 to 0.04; P = 0.16). Sensitivity analysis on the three included studies for which patient data were available was stable. Conflicting results were reported in the two studies that compared the effect of oxygen therapy on exercise tolerance (via 6MWT), where one study (Ahmedzai et al., 2004) indicated a statistically significant increase in distance with oxygen (174.6 m; SD = 11.2) use over medical air (128.8 m; P < 0.01, SD = 10.3), while the other (Bruera et al., 2003) did not indicate a difference between use of oxygen (331.6 m; SD = 54.9) and medical air (330.7 m; SD = 57.9). Of the four out of five studies that provided data on still-blinded patient preference for oxygen versus medical air, two studies reported a statistically significant still-blinded patient preference for oxygen over medical air. Four out of five studies had poor quality of reporting and inadequate discussion of randomization and blinding methodology.

Oxygen therapy did not appear to relieve the overall sensation of dyspnea in patients with cancer who do not qualify for long-term home oxygen therapy.

Study limitations include small patient sample size (n = 148), and 65% of patients were diagnosed with lung cancer (which limits the generalizability of data from populations with a wider range of malignancies with other factors contributing to dyspnea).

Further research and evaluation still is indicated. Despite conflicting findings among studies related to patient preference and oxygen effect on endurance and the overall lack of support for oxygen therapy in palliation of refractory dyspnea in patients with cancer, the subjective/psychological nature of dyspnea indicates that oxygen therapy may provide a comforting sense of dyspneic relief in some populations if served as a supplement to other, more effective interventions.