Palliative use of non-invasive ventilation in end-of-life patients with solid tumours: a randomised feasibility trial.

Background Despite best-possible medical management, many patients with end-stage cancer experience breathlessness, especially towards the end of their lives. We assessed the acceptability and effectiveness of non-invasive mechanical ventilation (NIV) versus oxygen therapy in decreasing dyspnoea and the amount of opiates needed. Methods In this randomised feasibility study, we recruited patients from seven centres in Italy, Spain, and Taiwan, who had solid tumours and acute respiratory failure and had a life expectancy of less than 6 months. We randomly allocated patients to receive either NIV (using the Pressure Support mode and scheduled on patients' request and mask comfort) or oxygen therapy (using a Venturi or a reservoir mask). We used a computer-generated sequence for randomisation, stratified on the basis of patients' hypercapnic status (PaCO2 >45 mm Hg or PaCO2 <= 45 mm Hg), and assigned treatment allocation using opaque, sealed envelopes. Patients in both groups were given sufficient subcutaneous morphine to reduce their dyspnoea score by at least one point on the Borg scale. Our primary endpoints were to assess the acceptability of NIV used solely as a palliative measure and to assess its effectiveness in reducing dyspnoea and the amount of opiates needed compared with oxygen therapy. Analysis was done by intention to treat. This study is registered with ClinicalTrials.gov, number NCT00533143. Findings We recruited patients between Jan 15, 2008, and March 9, 2011. Of 234 patients eligible for recruitment, we randomly allocated 200 (85%) to treatment: 99 to NIV and 101 to oxygen. 11 (11%) patients in the NIV group discontinued treatment; no patients in the oxygen group discontinued treatment. Dyspnoea decreased more rapidly in the NIV group compared with the oxygen group (average change in Borg scale -0.58, 95% CI -0.92 to -0.23, p=0.0012), with most benefit seen after the first hour of treatment and in hypercapnic patients. The total dose of morphine during the first 48 h was lower in the NIV group than it was in the oxygen group (26.9 mg [37.3] for NIV vs 59.4 mg [SD 67.1] for oxygen; mean difference -32.4 mg, 95% CI -47.5 to -17.4). Adverse events leading to NIV discontinuation were mainly related to mask intolerance and anxiety. Morphine was suspended because of severe vomiting and nausea (one patient in each group), sudden respiratory arrest (one patient in the NIV group), and myocardial infarction (one patient in the oxygen group). Interpretation Our findings suggest that NIV is more effective compared with oxygen in reducing dyspnoea and decreasing the doses of morphine needed in patients with end-stage cancer. Further studies are needed to confirm our findings and to assess the effectiveness of NIV on other outcomes such as survival. The use of NIV is, however, restricted to centres with NIV equipment, our findings are not generalisable to all cancer or palliative care units.