BackgroundDetrended fluctuation analysis (DFA) and sample entropy (SE) measure the long-term correlations and regularity of gait patterns, respectively, having previously been used to identify participants at risk of falling, previous history of injury, or patients with motor diseases. Since these measures are more sensitive to gait impairment than linear measures (e.g., the standard deviation [SD] of stride time), they can be potentially used in military medicine to identify soldiers at risk of injury. However, clinometric properties are yet to be established. Research QuestionWhat is the repeatability of DFA, SE, and traditional linear measures of stride time variability (SD and coefficient of variation [CV]) under various load and speed constraints? MethodsFourteen Australian Army trainee soldiers (age: 25.6 ± 5.9 y, height: 1.74 ± 0.08 m, body mass: 77.2 ± 15.1 kg, service: 1.5 ± 1.8 y) attended three sessions over two weeks, completing four 12-minute walking trials on an instrumented treadmill in each session. Participants walked with a combination of 0 kg or 23 kg loads at a self-selected or 5.5 km/h speed. Heel contacts from the right foot were identified using treadmill-embedded force plates. From 512 stride time intervals, linear (SD and CV), and non-linear (DFA and SE) measures were obtained. To assess the between-session repeatability, intraclass correlations (ICC 2,1) were employed. Results and SignificanceThere was poor-to-moderate repeatability for the SD (ICC: 0.357–0.545) and CV (ICC: 0.371–0.529). DFA showed poor-to-moderate repeatability (ICC: 0.013–0.504), while SE had poor repeatability (ICC: 0.133–0.226). Previous studies have shown that differences of > 0.19 in DFA and > 0.66 in SE can differentiate between healthy and pathological gait. These values are greater than this study's reported standard error of measurement, indicating that clinically meaningful changes may still be detectable despite low repeatability.
Read full abstract