Rationale: During noninvasive ventilation (NIV) for chronic obstructive pulmonary disease (COPD) exacerbations, helium/oxygen (heliox) reduces the work of breathing and hypercapnia more than air/O2, but its impact on clinical outcomes remains unknown. Objectives: To determine whether continuous administration of heliox for 72 hours, during and in-between NIV sessions, was superior to air/O2 in reducing NIV failure (25-15%) in severe hypercapnic COPD exacerbations. Methods: This was a prospective, randomized, open-label trial in 16 intensive care units (ICUs) and 6 countries. Inclusion criteria were COPD exacerbations with PaCO2 <45 mm Hg, pH<7.35, and at least one of the following: Respiratory rate<25/min,PaO2 50mmHg, and oxygen saturation (arterial [SaO2] or measured by pulse oximetry [SpO2])<90%. A 6-month follow-up was performed. Measurements and Main Results: The primary endpoint was NIV failure (intubation or death without intubation in the ICU). The secondary endpoints were physiological parameters, duration of ventilation, duration of ICU and hospital stay, 6-month recurrence, and rehospitalization rates. The trial was stopped prematurely (445 randomized patients) because of a low global failure rate (NIV failure: Air/O2 14.5% [n = 32]; heliox 14.7% [n = 33]; P = 0.97, and time to NIV failure: Heliox group 93 hours [n = 33], air/O2 group 52 hours [n = 32]; P = 0.12). Respiratory rate, pH, PaCO2, and encephalopathy score improved significantly faster with heliox. ICU stay was comparable between the groups. In patients intubated afterNIVfailed, patientson heliox had a shorter ventilation duration (7.4±7.6 d vs. 13.6±12.6 d; P = 0.02) and a shorter ICU stay (15.8±10.9 d vs. 26.±21.0 d; P = 0.01). No difference was observed in ICU and 6-month mortality. Conclusions: Heliox improves respiratory acidosis, encephalopathy, and the respiratory rate more quickly than air/O2? but does not prevent NIV failure. Overall, the rate of NIV failure was low. Clinical trial registered with www.clinicaltrials.gov (NCT 01155310).
