Intraoperative pulse oximetry: frequency and distribution of discrepant data

To determine the types of discrepant data during intraoperative pulse oximetry and their frequency and duration.

Prospective study.

University medical center.

46 consecutive ASA physical status I-III patients undergoing general anesthesia for elective surgical operations.

With an integrated computer algorithm on the pulse oximeter and another computer linked to it, data were screened and the frequency and distribution of the following oximeter signals recorded: absent; low quality or interrupted, as detected by the pulse oximeter algorithm; nonphysiologic, identified by the personal computer as a heart rate change greater than 10 beats per minute within 2 consecutive 2-second samples, with no similar abrupt change reported simultaneously on ECG. The number of episodes per hour of discrepant oximeter data and the duration of the episodes were recorded by phase of anesthesia: induction, maintenance, and emergence. Discrepant data occurred most frequently and lasted longest during emergence (p < 0.05); the majority of episodes of discrepant data during emergence lasted less than 12 seconds. Excluding discrepant data that lasted less than 12 seconds decreased the frequency of discrepant data by 63% and excluding those that lasted less than 30 seconds decreased the frequency of discrepant data by 93%.

Pulse oximeters frequently report discrepant data intraoperatively, most frequently during emergence from anesthesia. An alarm delay triggered by discrepant data and lasting 12 to 30 seconds would keep most discrepant data from becoming false alarms and, thus, may reduce distracting sound pollution in the operating room.