Reliability of continuous vital sign monitoring in post-operative patients employing consumer-grade fitness trackers: A randomised pilot trial

Abstract

Introduction Fitness trackers can provide continuous monitoring of vital signs and thus have the potential to become a complementary, mobile and effective tool for early detection of patient deterioration and post-operative complications. Methods To evaluate potential implementations in acute care setting, we included 36 patients after moderate to major surgery in a recent randomised pilot trial to compare the performance of vital sign monitoring by three different fitness trackers (Apple Watch 7, Garmin Fenix 6pro and Withings ScanWatch) with established standard clinical monitors in post-anaesthesia care units and monitoring wards. Results During a cumulative period of 56 days, a total of 53,197 heart rate (HR) measurements, as well as 12,219 measurements of the peripheral blood oxygen saturation (SpO 2 ) and 28,954 respiratory rate (RR) measurements were collected by fitness trackers. Under real-world conditions, HR monitoring was accurate and reliable across all benchmarked devices (r = [0.95;0.98], p < 0.001; Bias = [−0.74 bpm;−0.01 bpm]; MAPE∼2%). However, the performance of SpO 2 (r = [0.21;0.68]; p < 0.001; Bias = [−0.46%;−2.29%]; root-mean-square error = [2.82%;4.1%]) monitoring was substantially inferior. RR measurements could not be obtained for two of the devices, therefore exclusively the accuracy of the Garmin tracker could be evaluated (r = 0.28, p < 0.001; Bias = −1.46/min). Moreover, the time resolution of the vital sign measurements highly depends on the tracking device, ranging from 0.7 to 117.94 data points per hour. Conclusion According to the results of the present study, tracker devices are generally reliable and accurate for HR monitoring, whereas SpO 2 and RR measurements should be interpreted carefully, considering the clinical context of the respective patients.