CPR Simulation Shows Little Oxygen Build-Up at Defibrillation Sites

Key Takeaways

• In simulated CPR at typical defibrillation sites, oxygen levels stayed close to room air, with 95% of readings below 24%.
• Moving oxygen-delivery devices more than 1 m away did not lower measured oxygen concentrations.
• Endotracheal tube and laryngeal mask setups, and an oxygen-demand valve rather than a reservoir bag, were associated with slightly lower oxygen fractions.
• A poorly fitting mask produced localized neck leakage while chest-region values stayed below the commonly cited enrichment threshold.

Defibrillation during cardiopulmonary resuscitation (CPR) forces teams to balance uninterrupted oxygen delivery against concern that oxygen could collect near the chest during shock preparation. Investigators examined that assumption by measuring ambient oxygen in closed-room and ambulance simulations.

Using a high-fidelity Ambu mannequin with a closed airway system to simulate exhalation, investigators conducted a controlled manikin study of ambient oxygen enrichment during CPR that sampled oxygen every 10 seconds at three locations corresponding to typical defibrillation sites. Across 20 scenarios and 65 runs, they recorded 6,555 measurements in a closed room and a stationary ambulance with doors and windows closed. A standardized shockable-rhythm sequence spanned basic life support (BLS) and advanced life support (ALS), while oxygen-delivery device, airway management, ventilation approach, source distance during defibrillation, CPR mode, and upper-garment handling were deliberately varied.

Oxygen levels remained near ambient, with an overall median oxygen concentration of 21.6% (IQR 21.3% to 22.1%). Investigators saw no marked oxygen spikes during ventilation, no pronounced drop after oxygen sources were removed, and no measurable reduction when oxygen-delivery devices were moved more than 1 m away. Endotracheal tube and laryngeal mask scenarios, and use of an oxygen-demand valve rather than a reservoir bag at 15 L/min, were associated with slightly lower oxygen fractions. When a mask fit poorly, enrichment was localized near the neck rather than standard chest pad positions.

Interpretation is bounded by the fact that this was a manikin study under deliberately low-turbulence conditions rather than active human resuscitation. Oxygen was sampled at fixed measurement points instead of being mapped throughout the surrounding space, and the sensors were capped at 30%, which limits characterization of any higher local peaks. No direct ignition or combustion testing was performed, so any inference about defibrillation fire risk is indirect and limited to measured oxygen concentrations. Patient anatomy, body heat, airway moisture, device leakage, and the airflow generated during real resuscitation could all alter how oxygen disperses around the chest and head.

The authors concluded that, under the tested simulation conditions, clinically relevant oxygen accumulation at defibrillation sites was not observed and routine removal of oxygen sources did not measurably change the local environment

Clinician Questions

How closely did the CPR simulation match actual defibrillation workflow?

The scenarios followed a structured shockable-rhythm basic life support and advanced life support sequence, with initial CPR, rhythm analysis, charging during ongoing compressions, shock delivery, and repeated cycles through a fourth defibrillation. Each scenario was carried out by two ERC-certified ALS providers in either a closed room or a stationary ambulance, but the workflow remained standardized rather than subject to the variability of an actual arrest.

What parts of real patient resuscitation are not captured by this mannequin model?

No human participants were studied, and the mannequin model could not reproduce patient anatomy, a compliant moving chest wall, body heat, perspiration, airway moisture, or variable leakage around oxygen-delivery devices. The experiments were intentionally low-turbulence and relied on fixed sampling points, so airflow and oxygen dispersion during real resuscitation could differ.

Why doesn’t measuring ambient oxygen near defibrillation pads fully answer the fire-risk question?

The investigators measured local oxygen concentration as a proxy rather than recording actual arcing, ignition, or combustion during defibrillation. They also treated 24% as a precautionary oxygen-enrichment threshold rather than a hard ignition cutoff, so the findings define the oxygen environment under simulation conditions without directly measuring defibrillation-fire events.