2017 ISAKOS Biennial Congress ePoster #1211

 

Using Mobile Technology to Track Patient Recovery after Knee Arthroplasty

Stephen L. Lyman, PhD, New York, NY UNITED STATES
Catherine MacLean, MD, PhD, New York, NY UNITED STATES
Kara Fields, MPH, New York, NY UNITED STATES
Prem N Ramkumar, MD, MBA, Cleveland, Ohio UNITED STATES
David Mayman, MD, New York, NY UNITED STATES

Hospital for Special Surgery, New York, NY, UNITED STATES

FDA Status Not Applicable

Summary

While patient reported outcomes measures are the gold standard for measuring functional recovery after total knee arthroplasty, mobile technology tracking patient movement after surgery holds promise as a real-time monitoring system.

Abstract

Introduction

Patient Reported Outcome Measures (PROMs) are important tools in assessing patient functional recovery, but require active patient engagement and are usually collected at discrete time points. With advances in mobile technology, smart phones are now equipped with sensors that can directly, continuously, and passively measure patient mobility. The objective of this study was to assess the validity and usability of a mobile fitness application (app) to track patient recovery after total knee arthroplasty (TKA) compared to traditional PROMs.

Methods

We prospectively enrolled 100 consecutive eligible TKA patients 4 weeks prior to surgery and followed them for 6 months. The Moves app, which uses the phone’s native accelerometer and GPS to continuously monitor steps taken and distance traveled, was installed on the personal phones of all participants. Patients also completed weekly PROMs, including pain numeric rating scale (NRS), KOOS, JR. (an abbreviated form of KOOS optimized for TKA patients), and KOOS Quality of Life via a web link. For each TKA patient we calculated the mean preoperative daily step count and the percentage of this mean step count that was achieved on each postoperative day. We then plotted the recovery trajectory of each patient as a moving average of the percentage of average preoperative steps achieved vs. time post surgery. An unsupervised machine learning algorithm, K-means clustering, was used to identify subgroups of patients with similar recovery trajectories. Spearman’s correlation was used to compare app-based mobility measurements and PROMs.

Results

Three clusters of TKA patients were identified based on their recovery trajectories. The least active group of patients recovered, on average, approximately 50% of their mean preoperative steps by 2 months and approximately 75% by 6 months. The moderately active group recovered, on average, approximately 100% of their mean preoperative steps by 2 months and remained stable through 6 months. The most active group achieved, on average, approximately 175% of their mean preoperative steps by 2 months and gradually increasing to 200% by 4 months. Pain and KOOS, JR. scores were similar in the three groups preoperatively. The correlation between percentage of mean preoperative steps achieved vs. KOOS, JR. was 0.38 (95% CI 0.35, 0.42) and for steps vs. Pain (NRS score) was -0.35 (95% CI -0.39, -0.31).

Conclusion

App-measured mobility identified 3 patterns of functional recovery, and these divergent patterns may suggest an opportunity to optimize rehabilitation programs in the different groups. The modest correlation with PROMs support the continued use of these patient reported metrics even as passive measures are incorporated. If technological challenges can be managed, using mobile technology to directly, passively monitor post-operative mobility may prove a cost-effective tool for clinicians and researchers.