C.J. Gehl1, B. Li1, K. Quade1, Y. Xing1, T.J. Shaik1, K. Nimmer1, X. Yang1, S.L. Kerns1, A.N. Kothari1 1Medical College of Wisconsin, Milwaukee, WI, USA
Introduction: Wearable devices are of growing interest in surgery due to their ability to objectively and longitudinally track patient activity and physiological parameters such as sleep and heart rate (HR). This offers an opportunity to use data collected by these devices to help perioperative providers measure postoperative recovery in real-time. The objective of this study was to measure activity and physiologic parameters collected by Fitbit devices in the postoperative period using the All of Us (AOU) Research Program.
Methods: Data was obtained using the AOU Research Workbench controlled tier dataset. Included were participants who underwent a surgical procedure, defined by NSQIP inclusion CPT codes, and who had pre- and postoperative Fitbit data. Excluded were patients who did not have Fitbit data within 90 days following their procedure and patients who underwent procedures that would substantially limit mobility (i.e., lower extremity fracture repair). HR variability was measured by calculating the standard deviation of the 5-minute NN intervals (SDANN), activity was measured by step counts, and sleep was measured by minutes asleep. Postoperative Fitbit metrics were represented as a percent of baseline. This was calculated by taking participants’ preoperative data and calculating baseline average per day; for example, an individual’s baseline daily step count was the average daily steps taken across any days prior to their procedure. Recovery to baseline was defined as achieving 95% of baseline for 3 consecutive days.
Results: Of 808 patients who met the inclusion criteria, 586 (72.5%) were women, 688 (85.1%) were White, and the mean age was 60.4 (SD 14.4). The average baseline daily step count was 7712.2 (SD 4892.8), the average baseline minutes asleep was 372.2 minutes (SD 76.3), and the average baseline SDANN was 115.5 (SD 30.9). There was a drop on postoperative day (POD) 1 in each measured domain, with participants averaging 42.8% (SD 46.1%) of their baseline step count, 87.3% (SD 44.7%) of their baseline minutes asleep, and 82.2% (SD 24.6%) of their baseline HR variability (Figure). The return to above 95% of baseline for three consecutive days varied for each domain, with recovery occurring on POD61 for step count, POD4 for sleep, and POD17 for HR variability.
Conclusions: We found a decline in activity, sleep, and HR variability from individual baseline immediately following elective surgery with a gradual increase back towards baseline with time; this return to baseline varied based on measured domain with the longest being for physical activity. These measures can inform preoperative decision-making and preparedness prior to elective surgery and eventually as quality measures for surgical recovery.