Abstract Background High-flow nasal cannula (HFNC) can improve ventilatory function in patients with acute COPD exacerbation. However, its effect on clinical outcomes remains uncertain. Methods This randomized controlled trial was conducted from July 2017 to December 2020 in 16 tertiary hospitals in China. Patients with acute COPD exacerbation with mild hypercapnia (pH ≥ 7.35 and arterial partial pressure of carbon dioxide > 45 mmHg) were randomly assigned to either HFNC or conventional oxygen therapy. The primary outcome was the proportion of patients who met the criteria for intubation during hospitalization. Secondary outcomes included treatment failure (intolerance and need for non-invasive or invasive ventilation), length of hospital stay, hospital cost, mortality, and readmission at day 90. Results Among 337 randomized patients (median age, 70.0 years; 280 men [83.1%]; median pH 7.399; arterial partial pressure of carbon dioxide 51 mmHg), 330 completed the trial. 4/158 patients on HFNC and 1/172 patient on conventional oxygen therapy met the criteria for intubation (P = 0.198). Patients progressed to NPPV in both groups were comparable (15 [9.5%] in the HFNC group vs. 22 [12.8%] in the conventional oxygen therapy group; P = 0.343). Compared with conventional oxygen therapy, HFNC yielded a significantly longer median length of hospital stay (9.0 [interquartile range, 7.0–13.0] vs. 8.0 [interquartile range, 7.0–11.0] days) and a higher median hospital cost (approximately $2298 [interquartile range, $1613–$3782] vs. $2005 [interquartile range, $1439–$2968]). There were no significant differences in other secondary outcomes between groups. Conclusions In this multi-center randomized controlled study, HFNC compared to conventional oxygen therapy did not reduce need for intubation among acute COPD exacerbation patients with mild hypercapnia. The future studies should focus on patients with acute COPD exacerbation with respiratory acidosis (pH < 7.35). However, because the primary outcome rate was well below expected, the study was underpowered to show a meaningful difference between the two treatment groups. Trial registration: NCT03003559 . Registered on December 28, 2016.
