Comparison of two automated oxygen control algorithms in mechanically ventilated neonates: an open label, randomised, crossover trial

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Objective To compare rule-based non-fuzzy algorithm (Predictive Intelligent Control of Oxygenation (PRICO)) with adaptive algorithm (Closed-loop Inspired Oxygen (CLiO 2 )) for automated oxygen control in mechanically ventilated neonates. Design Open label, randomised, crossover trial. Setting Level III neonatal intensive care unit. Patients Neonates ≤44 weeks post-menstrual age receiving invasive or non-invasive ventilation and fraction of inspired oxygen (FiO ₂ ) ≥25%. Interventions Participants randomised to one of two sequences of auto-FiO2 (CLiO 2 then PRICO or PRICO then CLiO 2 ). Each algorithm in the sequence was applied for 24 hours before crossing over to an alternate algorithm. Target saturation was 91–95%. Masimo pulse-oximeters were used for saturation measurement. Main outcome Primary outcome was time-weighted proportion of time spent within target saturation range. Secondary outcomes were time spent below and above target saturation range. Results 32 neonates were enrolled; 29 completed both intervention periods and contributed to paired data. Mean time-weighted proportion within target range was 76% in the CLiO 2 group compared with 62% in PRICO (mean within-participant difference (PRICO-CLiO 2 ) −14 percentage points, 95% CI −24 to −4; paired t-test p=0.01). A mixed-effects model accounting for period gave an 18 percentage points difference (82% vs 64%; treatment effect −0.92 on the logit scale, 95% CI −1.51 to −0.33, p=0.004). Time-weighted proportion spent above target range was lower with CLiO 2 (15% vs 30%). Time spent below target range was comparable between groups. No significant period or carryover effect was observed. Conclusions In mechanically ventilated neonates, adaptive CLiO 2 algorithm did maintain the oxygen saturation within target range for a greater proportion of time and reduced hyperoxic exposure compared with the rule-based PRICO algorithm. Trial registration number CTRI/2025/05/087301.