Preoxygenation with high-flow nasal oxygen at various flow rates in elective surgical patients: a prospective, randomised, single-blind clinical trial

Citation: Sjoblom A, Hoffman F, Hedberg M, Forsberg I-M, Jonsson Fagerlund M. Preoxygenation with high-flow nasal oxygen at various flow rates in elective surgical patients: a prospective, randomised, single-blind clinical trial. Br J Anaesth. 2025;[Epub ahead of print]. doi:10.1016/j.bja.2025.11.017.

Study at a glance:

- Design and setting: Prospective, randomised, single-blind 3-arm physiologic RCT in adults (18–84 yr, BMI <35 kg m−2, ASA ≤3) undergoing elective colorectal, urological or gynaecological surgery at a single Swedish university hospital (Karolinska); 75 randomised, 72 analysed after 3 exclusions in the 45 L min−1 arm (1 new Mobitz II AV block before induction, 2 protocol violations). Standardised IV induction (fast-acting opioid, propofol, rocuronium 1 mg kg−1), videolaryngoscopic intubation, and apnoeic oxygenation without mask ventilation until SpO2 93%.

- Interventions: 3 min HFNO preoxygenation at FiO2 1.0 in 10° reverse Trendelenburg via Optiflow+ nasal cannula and Hamilton ventilator circuit:

• 45 L min−1 HFNO (n=22 analysed)

• 70 L min−1 HFNO (n=25)

• 95 L min−1 HFNO (n=25; one patient down-titrated to 70 L min−1 for discomfort but kept in 95 L min−1 group).

All groups had invasive arterial monitoring and serial ABGs before/during preoxygenation, at apnoea start, during apnoea and at termination; PaCO2 and PaO2 trajectories and PaCO2–ETCO2 gradient were recorded.

- Primary outcome (safe apnoea time to SpO2 93%): Median apnoea times were similar across flows—45 L min−1: 472 s (IQR 402–670); 70 L min−1: 523 s (420–664); 95 L min−1: 483 s (375–605); global P=0.59. Most patients in all arms tolerated ≥360 s of apnoea without SpO2 <93% (86%, 96%, 80%; P=0.22). Thus, in relatively healthy, non-obese elective surgical patients, higher HFNO flows (70 or 95 L min−1) did not meaningfully prolong safe apnoea time compared with 45 L min−1 (low-certainty evidence, RoB 2 overall “some concerns” due to per-protocol analysis and unbalanced post-randomisation exclusions).

- Key secondary and exploratory findings: PaO2 increased similarly during preoxygenation across groups, but at apnoea start PaO2 was higher with 95 vs 45 L min−1 (median 73.5 vs 65.1 kPa; P=0.001) without translating into longer safe apnoea time; PaCO2 at apnoea termination was similar (~8 kPa) regardless of flow. Preoxygenation-related discomfort (VAS 0–10) differed markedly: 45 L min−1 was best tolerated (median 1 [0–2]) versus 70 L min−1 5 (2–5) and 95 L min−1 3 (1–5); global P=0.002, with 45 L min−1 significantly less uncomfortable than both higher flows. Mouth closure during preoxygenation was high in all arms (closed throughout: 100%, 84%, 84%; P=0.37), and an exploratory analysis showed that failure to keep the mouth closed did not significantly shorten safe apnoea time. Serial ABGs demonstrated a nonlinear PaCO2 rise during apnoea (largest increase in the first minute) and a mean PaCO2–ETCO2 gradient of ~1.4 kPa at apnoea termination.

- Applicability and implications: Small, single-centre, physiologic OR study in non-obese, ASA 1–3 elective patients; does not address obese, high-risk, emergency, ICU or rapid-sequence settings, nor compare HFNO with conventional facemask preoxygenation. Within this population, 45 L min−1 HFNO for 3 min appears to provide apnoeic oxygenation (safe apnoea time) comparable to 70–95 L min−1, with substantially less patient discomfort; use of higher flows may be reserved for specific scenarios where a modest PaO2 gain is desired, recognising the comfort trade-off and the current low–moderate certainty of evidence.