The Use of Activity Trackers in Postoperative Recovery: A Systematic Review and Meta-Analysis (Preprint)

BACKGROUND Postoperative mobilization is an important aspect of recovery from surgery, and the increasing prevalence of wearable activity trackers is a promising avenue for objective monitoring of this key outcome. This systematic review and meta-analysis provides the first comprehensive evaluation of the uses and validation of activity trackers in postoperative patients. OBJECTIVE We aimed to assess the extent and characteristics of wearable activity tracker usage in the postoperative patient population, as well as the consistency of results amongst various surgical disciplines and activity tracker models. This was undertaken with the ultimate objective of defining best practices for future research in this burgeoning field. METHODS A comprehensive literature search was conducted for studies which used a wearable activity tracker to measure step counts in adult patients within one year of surgery. Each selected study was reviewed and assessed for risk of bias by multiple investigators, with documentation of the methods used to validate tracker accuracy and clean tracker data before analysis. To calculate the overall mean step count at the preoperative and various postoperative timepoints across all studies, a fixed-effects inverse-variance weighted mean method was used for all timepoints with sufficient data available. RESULTS Of 5,694 studies identified, 109 were selected for inclusion and 73 were used in the meta-analysis. Orthopedics was most common surgical discipline represented (41%). Research-grade trackers such as the Actigraph GT3X+ were used in 50% of studies, with the remainder utilizing commercial trackers such as various Fitbit models (17% of all studies), the activPAL (14%), and the StepWatch Activity Monitor (12%). Patients moved significantly less in the first postoperative week than their preoperative activity levels (p<.001). Although the second and third postoperative weeks also showed less activity than the preoperative period, these differences were not statistically significant. Postoperative step counts four weeks or more after surgery were greater than preoperative step counts, although these differences did not reach statistical significance. Research activity trackers recorded higher mean step counts on aggregate at nearly all timepoints relative to commercial trackers, but these differences were only statistically significant at the one-week postoperative time point. Nearly one-quarter of studies simply used the raw step count data from trackers without any data cleaning or standardization, and only 12% of studies used another objective measure of step counts to verify the activity tracker values. CONCLUSIONS While postoperative patients move significantly less in the first week than their preoperative activity levels, later postoperative time points do not statistically differ from preoperative activity, potentially indicating that patients recover or exceed their prior level of function within a few weeks of many major surgeries. The wide variety of trackers used in addition to a lack of standardized best practices in tracker data cleaning and accuracy verification contribute to a wide variation in step counts before surgery and at all stages in the first postoperative year. Future researchers would benefit from the establishment of best practices such as reporting the intended duration of tracker use and patient compliance, defining non-wear time, and specifying reasons for excluding tracker data points from analyses.