A Comparison of Wireless Pulse Oximetry Technologies during SustainedShivering in Post-Operative Patients

Abstract

Background: Pulse oximetry measures oxygen saturation in arterial blood non-invasively and is routinely practiced in many clinical settings. Technological advancements have expanded the performance capabilities of these devices to operate wirelessly, making pulse oximetry measures more convenient to obtain. Factors such as patient motion however can potentially complicate the ability to retrieve these measures. We examined the performance of two wireless pulse oximeters in patients who shiver postoperatively. Methods: Two additional pulse oximeter sensors (Masimo and Nonin) were added to 30 patients who shivered post-operatively and met eligibility criteria. Dropout rates were calculated by dividing the number of times that each instrument displayed no SpO2 or pulse rate value by the total time both sensors were in camera view for that subject. We used Wilcoxon matched pairs tests to compare the average time until the first reading displayed on the sensor for each device as well as dropout times. Dropout events were also compared using McNemar's test. Results: Drops in signal occurred in thirteen subjects. The Masimo sensor had more dropouts and a longer average drop duration, which was found to be statistically significant (p=0.02). Masimo's sensor required more s on average than Nonin's (8.86>8.78) to display the first reading, but this was not found to be statistically significant (p=0.90). Masimo also had a higher percentage of drops however this was not statistically significant (p=0.07). Conclusion: Overall incidence of shivering was low among patients in this setting (0.1%). More than 1/3 of subjects experienced drops during identical time intervals indicating that motion artifact may impact results regardless of technology. Additionally, Masimo's sensor is not meant for clinical use thereby raising questions regarding our statistically significant result in this particular setting. Combined with results from part I of this study, we cannot confidently detect a statistically significant difference between manufacturers.