Philip, J., Gold, M., Milner, A., Di Iulio, J., Miller, B., & Spruyt, O. (2006). A randomized, double-blind, crossover trial of the effect of oxygen on dyspnea in patients with advanced cancer. Journal of Pain and Symptom Management, 32(6), 541–550.

The objective of the study was to determine if patients preferred oxygen or air following 15-minute administration of both. Another aim of the study was to identify other factors that might impact the experience of dyspnea and the response to oxygen.

The study compared the response to oxygen and air in hypoxic and nonhypoxic patients. Patients were randomized to receive either oxygen or air at 4 L via nasal canula for 15 minutes. At the completion of 15 minutes, dyspnea intensity ratings and oximetry were repeated. Patients then spent 30 minutes without gas. Repeat measures were performed with a crossover to the other gas for 15 minutes. Measure of symptom intensity and oximetry were repeated, then the blinded patient and investigator designated the preferred gas.

The study reported on a sample of 51 patients. Dyspnea was related to the cancer in 47 patients (92%). In 29 of the 47 patients, cancer was the sole cause of dyspnea. In the remaining patients, dyspnea causes were from cancer complications, such as pneumonia (five patients), and cancer treatment, such as radiation pneumonitis. Fifteen patients (29%) had unrelated dyspnea causes, including 11 patients with chronic obstructive pulmonary disease.

Patients were eligible if

• They had a diagnosis of cancer
• Their dyspnea was related to the cancer
• They had a dyspnea intensity score of at least 30 mm on a 0-100 mm visual analog scale (VAS)
• They were on stable medication doses
• They had a normal cognitive status
• They were aged 18 years or older
• They had no contraindications to oxygen.

Patients were excluded if they

• Had acute respiratory distress
• Were oxygen dependent.

The study was conducted in two centers in Australia. Patients were recruited from inpatient and outpatient units.

Randomized, double-blind, crossover study

• VAS for dyspnea
• European Organization for Research and Treatment of Cancer (EORTC) QLQ C30, providing verbal ratings of dyspnea intensity
• Pulse oximetry oxygen saturation
• Qualitative descriptors of dyspnea from Dyspnea Assessment Questionnaire

Twenty-seven patients (53%) were randomized to the air first arm and 24 patients (47%) to the oxygen arm. No significant difference was seen in VAS score improvement between the two types of gases (p = 0.622). No significant difference was seen in percentage of verbal ratings of improvement after first gas (p = 0.888) and after the second gas (p= 0.767). A significant difference was seen between the two gas types in mean increase in oxygen saturation (p < 0.001, air = 0.94%, oxygen = 5.43%) No significant correlation was seen between VAS score and oxygen saturation. Twenty-one patients (41%) preferred oxygen, 15 (29%) preferred air, and 15 (29%) had no preference. No significant difference (p = 0.357) was seen in patient preference for air or oxygen. In the subgroup of 17 hypoxic patients, mean change in VAS score did not differ significantly between air and oxygen (p = 0.812, air = 15.4 mm, oxygen = 13.3 mm), but mean oxygen saturation levels increased significantly more for oxygen than for air (p = 0.005, air = 2.7%, oxygen = 10.7%).

On average, patients improved symptomatically with both air and oxygen, and no significant difference was seen between the treatments. The subgroup of 17 hypoxic patients overall did not demonstrate a significant difference between air and oxygen, despite having improved oxygen saturations when administered oxygen. No major or minor flaws were noted in the study design. The authors designated clinically significant response to oxygen to be a preference for oxygen chosen by 60% of patients. If clinically significant improvement occurred at lower increments, this study may not have been adequately powered.

Air was not considered a placebo in this trial, but in fact a placebo effect may have been associated with air administration. Another possible explanation is that no differential response to either air or oxygen may be a result of mechanoreceptos stimulated by any gas administration. Patients who were dyspneic upon exertion but not dyspneic at rest were not eligible to enter the study. Eligible patients had to record a dyspnea VAS score of at least 30 mm. There may have been a different preference and response to the gases for exertional dyspnea.

Accruing 50 patients with dyspnea to this study took five years, which underscores the clinical fragility of patients who experience dyspnea and the difficulty in conducting research in this population. This evidence contradicts the findings of Bruera et al. (1993), who demonstrated that oxygen is beneficial to and preferred by patients with hypoxia.