Abstract
Introduction
Recent advances in algorithms developed with passively collected sensor data from smart phones and watches demonstrate new, objective, metrics with the capacity to show qualitative gait characteristics. The purpose of this feasibility study was to assess the recovery of gait quality following primary total hip and knee arthroplasty collected using a smartphone-based care platform.
Methods
A secondary data analysis of an IRB approved multicenter prospective trial evaluating the use of a smartphone-based care platform for primary total knee arthroplasty (TKA, n=88), unicondylar knee arthroplasty (UKA, n=28), and total hip arthroplasty (THA, n=82). Subjects were followed from 6 weeks preoperative to 24 weeks postoperative. The group was comprised of 117 females and 81 males with a mean age of 61.4 and BMI of 30.7. Signals were collected from the participants' smartphones. These signals were used to estimate gait quality according to walking speed, step length, and timing asymmetry. Post-operative measures were compared to preoperative baseline levels using a Signed-Rank test (p<0.05).
Results
Mean walking speeds were lowest at postoperative week 2 for all three procedures (p<.001). The TKA population stabilized to preoperative speeds by week 17. For UKA cases, mean walking speeds rebounded to preoperative speed consistently by week 9 (p>.05). THA cases returned to preoperative walking speeds with a stable rebound starting at week 6 (p>.05), and improvement was seen at week 14 (p=.025).
The average weekly step length was lowest in postoperative week 2 for both TKA and UKA (p<.001), and at week 3 for THA (p<.001). The TKA population rebounded to preoperative step lengths at week 9 (p=0.109), UKA cases at week 7 (p=.123), and THA cases by week 6 (p=.946).
For TKA subjects, the change in average weekly gait asymmetry peaked at week 2 postoperatively (p <0.001), returning to baseline symmetry by week 13 (p=.161). For UKA cases, mean gait asymmetry also reached its maximum at week 2 (p =.006), returning to baseline beginning at week 7 (p=0.057). For THA cases mean asymmetry reached its maximum in week 2 (p <0.001) and was returned to baseline values at week 6 (p=.150).
Discussion and Conclusion
Monitoring gait quality in real-world patient care following hip and knee arthroplasty using smart phone technology demonstrated recovery curves similar to previously reported curves captured by traditional gait analysis methods and patient reported outcome scores. Capturing such real-world gait quality metrics passively through the phone may also provide the advantage of removing the Hawthorne effect related to typical gait assessments and in-clinic observations, leading to a more accurate picture of patient function.
