Reliability and sensitivity of the 20-meter walk test among patients with knee osteoarthritis

Abstract

Purpose: The 20-meter walk test is a physical function measure commonly used in clinical research studies and rehabilitation clinics to measure gait speed and monitor changes in patients’ physical function over time. Unfortunately, the reliability and sensitivity of this walk test are not well defined and therefore limit our ability to evaluate changes in gait speed. The aim of this study was to assess the reliability and sensitivity of the 20-meter walk test, at a self-selected pace, among patients with knee osteoarthritis (OA). Methods: This was a measurement reliability study that included fifteen consecutive participants enrolled in a randomized-controlled trial of intra-articular corticosteroid injections for knee OA. The study sample included consecutive participants attending screening (Day 1) and baseline (Day 2) visits between July 2011 and December 2011. All participants met the American College of Rheumatology criteria for OA; had radiographic knee OA, as defined by Kellgren-Lawrence grade 2 or 3; and knee synovitis, defined by a synovial pouch depth greater than 2.0 mm measured by ultrasound. All participants were also required to have knee pain symptoms, defined as a WOMAC Osteoarthritis Index (version 3.1, 5-point Likert) pain subscore > 2 at the beginning of their first visit (Day 1). All participants completed 4 trials on 2 separate days, 7 to 21 days apart (8 trials total). Each day was divided into 2 sessions, which each involved 2 walking trials. Day 1 contained sessions 1 (trial 1 and 2) and session 2 (trial 3 and 4) and Day 2 contained session 3 (trial 5 and 6) and session 4 (trial 7 and 8). All trials were administered by the same investigator following a standardized script and protocol. We compared walk times between trials with Wilcoxon signed-rank tests. We also calculated Spearman correlation coefficients to assess the relationship between sessions. Finally, smallest detectable differences (SDD) were calculated to estimate the sensitivity of the 20-meter walk test. Results: Participants were 53% female, 67% (n 1⁄4 10) Caucasian, on average 61.0 7.8 years of age and had a mean body mass index of 28.9 5.4 kg/m2. Twelve participants (80%) had Kellegren-Lawrence Grade 1⁄4 3. WOMAC pain scores were 5.3 1.3 on Day 1 and 4.9 2.0 on Day 2. The figure depicts walk times across trials. We found that walking times in the first session were slower than the second session (median difference 1⁄4 0.53 seconds, -0.04 m/s; Table). We also found that the correlation between session 1 and sessions 3 and 4 were lower than the correlations between session 2 and sessions 3 and 4 (Table). Therefore, we considered the first session of each day a practice session and calculated the SDD between the second session of each day (session 2 and 4). SDD were -2.59 seconds (walking slower) and 1.65 seconds (walking faster). There was a potential systematic bias of participants walking slower during the second session on Day 2 compared to the second session of Day 1 (10 [67%] participants walked slower in session 4 compared to session 2; based on differences below zero). Conclusions:We found that the average of the first two 20-meter walk times did not agree with subsequent average walk times among participants with knee OA. Therefore, practice trials may be advised prior to a measuring a participants walk time. Despite concerns about thewalk times of session 1, the following sessions were reliable and had good agreement. Finally, changes in walk time between -2.59 seconds (walking slower) and 1.65 seconds (walking faster) should be considered within the range of normal variability of 20-meter walking speed.