Neonatal automated oxygen titration algorithms: response characteristics and performance

Neonatal respiratory support devices now have algorithms for automated adjustment of oxygen concentration (fraction of inspired oxygen (FiO₂)) in response to oxygen saturation (SpO₂) disturbances, but their performance has not been systematically examined. We investigated the response of six available devices and algorithms to standardised and simulated SpO₂ disturbances.

Bench-top study in neonatal centres.

Devices under study received SpO₂ input from a vital signs simulator at 1 Hz; device response was recorded as measured FiO₂ in the respiratory circuit. Response to protracted hypoxaemia (SpO₂=80%) and hyperoxaemia (SpO₂=98%) was examined in several test conditions. In closed-loop testing, measured FiO₂ was combined with abstracted patient data in an oxygenation simulator, generating novel SpO₂ values input to the device. Device capability in keeping simulated SpO₂ in target range (91-95%) was examined, as was FiO₂ adjustment frequency and propensity for FiO₂ oscillation.

The FiO₂ trajectories in response to SpO₂ disturbances differed considerably between devices. Maximum rate of sustained FiO₂ escalation (ΔFiO₂/min) during hypoxaemia was Avea: 0.30; Fabian: 0.20; Leoni: 0.20; SLE: 0.10; Sophie: 0.062; Vapotherm: 0.30. Corresponding values for FiO₂ reduction during hyperoxaemia were 1.20, 0.20, 0.20, 0.13, 0.53 and 1.32. In closed-loop testing, average SpO₂ target range time proportion was 51%, 52%, 55%, 77%, 64% and 53% for the six devices, respectively, with 1.9-5.1 automated FiO₂ adjustments/min, in some cases with a prominent oscillatory pattern.

Automated oxygen control devices exhibit considerable variation in their FiO₂ response to SpO₂ disturbances, with important consequences for SpO₂ targeting potential and the pattern of FiO₂ adjustment.