Abstract Knee osteoarthritis is a significant age-related condition that impairs mobility. Lab-based gait measurement has identified specific gait mechanics that are associated with knee osteoarthritis severity and progression. However, when interventions that successfully correct aberrant gait mechanics in the lab are implemented in clinical trials, they do not meaningfully improve knee osteoarthritis symptoms or slow progression. This gap between in-lab success and real-world efficacy may be because individuals walk differently in daily life than in lab settings, with in-lab gait likely only being representative of a small portion of an individual’s real-world gait. Longitudinal measurement of real-world gait could identify new factors that relate to joint health or intervention effectiveness, enabling better prediction of knee osteoarthritis progression and improved intervention design. Before we can design or execute a longitudinal study of real-world gait, we need a better understanding of real- world gait data. Recent studies indicate that individuals walk slower and with shorter stride lengths on average when out of the lab compared to in lab settings. These findings suggest that gait measures that are specifically tied to joint health may also differ during daily life compared to the lab setting. Additionally, symptoms such as pain and fatigue that are known to affect gait mechanics during in-lab collection vary substantially within and between days in real-world settings (and likely to a greater extent than in lab settings). Despite advances in wearable sensors such as inertial measurement units (IMUs) and the popularity of using IMUs for gait analysis in the lab, the use of IMUs in real-world gait analysis has been limited because of challenges with analyzing unobserved data and interpreting outcomes from novel data. Our team has successfully implemented reliable methods for detecting and categorizing walking activity (e.g., level walking vs. stairs, straight walking vs. turns), orienting data to recognizable reference frames (e.g., gravitational or functional), and calculating interpretable outcomes that correspond to traditional gait measures and are relevant to joint function (e.g., knee joint range of motion, propulsive ankle angular velocity, inter-segment coordination). In this study, we will calculate our established gait measures during real-world gait collected over 3 full, consecutive days in 3 groups of participants: older adults with knee osteoarthritis, older asymptomatic adults, and young adults. We will use electronic messaging to collect ecologically valid measures of pain and fatigue 5 times each day. We will use these data to compare the magnitude and variance of knee range of motion, propulsive ankle joint velocity, and lower extremity coordination between groups, days, and between real-world and lab settings (Aim 1). We will model the relationships between our gait measures and participant self-reported pain and fatigue (Aim 2). Completion of these aims will provide preliminary data with which we can design a larger study (R01) to evaluate the role of real-world gait in knee osteoarthritis progression. In the long term, this knowledge will allow for earlier detection of mobility decline and improved intervention design and implementation.

Narrative Many gait interventions for knee osteoarthritis fail to improve clinical outcomes despite improving gait mechanics, possibly because these interventions are based off of in-lab gait patterns, which likely differ from the gait patterns that patients use during daily, real-world walking. In this study, we will establish foundational knowledge of real-world gait measures and the impact of common knee osteoarthritis symptoms (pain, fatigue) on daily gait measures. This foundational knowledge will allow us to design a larger study to evaluate the role of real-world gait and symptoms in knee osteoarthritis progression and, long term, to improve the design and implementation of interventions to slow the progression of knee osteoarthritis.