Comparison of respiratory support with high-flow nasal oxygen versus conventional nasal cannula during sedated endoscopy for acute nonvariceal upper gastrointestinal bleeding: A single-center retrospective study

Hypoxemia is the most common adverse event during sedated gastrointestinal endoscopy. High-flow nasal oxygen (HFNO) reduces hypoxemia in elective endoscopy, but data on its utility in patients undergoing urgent endoscopy for acute nonvariceal upper gastrointestinal (UGI) bleeding are scarce. We compared HFNO with the conventional nasal cannula. This single-center retrospective cohort study included adult patients undergoing endoscopy for acute nonvariceal UGI bleeding with anesthesiologist-supervised propofol sedation within 24 h of presentation. Patients were grouped by oxygenation strategy: HFNO at 20–70 L/min with an inspired oxygen fraction (FiO 2 ) of 1.0, or conventional nasal cannula at 2–4 L/min (estimated FiO 2 0.28–0.37). The primary outcome was intraprocedural desaturation (peripheral oxygen saturation [SpO 2 ] ≤ 92% sustained for at least 10 s); secondary outcomes included the lowest intraprocedural SpO 2 , rescue airway maneuvers, post-procedural arterial blood gas values, and total propofol dose. Multivariable logistic regression adjusted for baseline variables differing between the groups. Of 114 patients, 67 received HFNO and 47 a conventional nasal cannula. The HFNO group was older (median 76 vs. 67 years; p = 0.005) with lower baseline SpO 2 (94% vs. 96%; p = 0.001) and higher American Society of Anesthesiologists (ASA) physical status ( p = 0.014). Desaturation occurred in 3 (4.5%) versus 19 (40.4%) patients and rescue airway maneuvers in 4 (6.0%) versus 24 (51.1%) ( p < 0.001 for both). After adjustment for age and baseline SpO 2 , HFNO remained associated with lower odds of desaturation (adjusted OR 0.036, 95% CI 0.008–0.173) and rescue airway maneuvers (adjusted OR 0.036, 95% CI 0.009–0.145). Post-procedural PaO 2 was higher with HFNO (143 vs. 94 mmHg; p < 0.001). In patients undergoing sedated endoscopy for acute nonvariceal UGI bleeding, the HFNO strategy was associated with fewer desaturation events and rescue airway maneuvers than the conventional nasal cannula strategy, whereas post-procedural arterial carbon dioxide (PaCO 2 ) did not differ. As the HFNO group also received a markedly higher FiO 2 , these findings reflect a comparison of two oxygenation strategies rather than an isolated device effect; this imbalance is the principal limitation. Randomized trials using equivalent FiO 2 are required.