Validating Wearable Sensors Using Self‐Reported Instability among Patients with Knee Osteoarthritis

Background Self‐perceived instability among patients with knee osteoarthritis (OA) is defined as giving way, buckling, or shifting of the knee during activities, especially walking. Although instability is a leading cause of mobility decline with knee OA, methods for quantifying the symptom, determining the mechanisms, and establishing effective interventions remain unclear. Recently, data outputs (ie, linear acceleration and its time‐derivative, jerk) from wearable sensors are showing strong associations with self‐perceived instability among patients with other knee pathologies and may offer insight into OA‐related instability. Objective To examine discriminant and convergent validity of using data outputs from wearable sensors to quantify self‐reported instability among patients with knee OA. Design Secondary analysis of a cross‐sectional study. Setting Primary recruitment from an institutional outpatient physical therapy clinic and collection completed in an institutional research laboratory. Patients Thirty‐nine total participants. The OA group included 26 participants with radiographic evidence of moderate to severe knee OA in the medial compartment; knee pain >3 out of 10, and a walking speed of ≥1.0 m/s. The control group included 13 participants with no history of knee OA. Participants with current or history of low back, hip, or foot/ankle injury; knee replacement; skeletal realignment surgery; or comorbidities that limit walking, pregnancy, and inability to walk without an assistive device were excluded. Interventions N/A Main Outcome Measures Data output from wearable sensors at the tibia. Results Midstance acceleration (P = .01) and jerk (P = .04) were significantly greater for those with than without knee OA. Acceleration was significantly associated with self‐reported instability (Spearman's rho = −0.63, P