Evaluation of pulse oximetry as a continuous monitoring technique in the neonatal intensive care unit.

We tested a number of pulse oximeters in a neonatal ICU to assess their performance as continuous monitors. Using a computer-based data acquisition system and a movement sensor, we determined the proportion of time that the oximeter was unreliable as the duration of monitored time that the oximeter heart rate (HR) differed from a "true" HR by greater than 10 beat/min using each of the oximeters in both their short and long averaging mode. The tested devices were found to be unreliable for 11.9% to 25% in the short averaging mode, and from 13.8% to 29% in the longer averaging mode. Most of the HR discrepancy was secondary to movement of the extremity. A newer model of one of the devices, which uses ECG R-wave detection to trigger optical pulse scanning, was associated with markedly reduced motion artifact (2.1% to 4.1%) when it accurately interpreted HR. The response of the devices to an acute loss of pulsation revealed differences in design and performance with some oximeters indicating zero saturation and others continuing to display a value while indicating a "low quality signal." Pulse oximeters are a valuable adjunct to clinical oxygen monitoring and, when properly applied and reliably indicating the infant's HR, will accurately reflect arterial saturation (r = .8, p less than .0001). ECG synchronization appears to reduce motion artifact when the ECG R-wave is detected.