PULSE OXIMETRY IN VENTILATED PRETERM INFA NTS - THE DILEMMA OF OVERLAPPING ALARM LIMITS

Objective: To evaluate the reliability of pulse oximetry in detecting both hyper- and hypoxaemic states. To create clinically feasible alarm limits. Patients and methods: Subjects were 96 artificially ventilate d preterm babies (52 boys) with indwelling umbilical artery catheters. Statistical analysis: Predictive value analysis of 471 pulse oximeter readings related to synchronous arterial oxygen tension was carried o ut, and the receiver operating characteristic (ROC) curves were genera ted separately for the hyperoxaemic and the hypoxaemic range to illust rate the discriminative power for both states. Main results: ROC-curve s revealed nearly identical capacity of the pulse oximeter in identify ing both hypoxaemia and hyperoxaemia. However a clinically suitable an d safe range of alarm limits to maintain an arterial oxygen tension of 50 to 90 mmHg could only be established at a sensitivity level of 0.7 5. At this level the alarm range of transcutaneous oxygen saturation w as 93-96%. Conclusion: Based on our findings, we cannot recommend puls e oximetry as a sole means of oxygen monitoring for ventilated preterm infants. Combination of pulse oximetry with arterial blood gas sampli ng or other methods of blood gas monitoring seems mandatory.