Walking Stride Variability On Different Walking Surfaces

Abstract

Measurements of the fractal scaling index of gait (α) provide an estimate of gait stability and risk of falling. Little attention has been given to the effects of walking surface on this measurement. PURPOSE: This investigation compared stride-to-stride variability for healthy subjects walking on two different surfaces. METHODS: Seven healthy subjects (ages 20–56) performed 20–25 minute, self-paced walks (1100–1300 strides) on two different surfaces (tennis hard court and an all-weather rubberized running track). An oval walking route was used to minimize the effects of turns on stride-to-stride variability. Surface roughness was estimated from the standard deviation of the vertical accelerations (SDa) made by a remote control vehicle as it traveled over the surfaces at the same average speed as the walkers. Each subject wore a shoe constructed from wool felt. Three force-sensing resistors were placed in the sole of the shoe worn by the dominant foot (usually the left): at the heel and at the head of the first and fifth metatarsals. A recorder carried in a fanny pack digitized the output of the sensors (256 Hz, 12-bit A/D) after it was passed through a signal-conditioning preamp and the stride times extracted. Stride-to-stride variability was estimated from the standard deviation (SDw) and the coefficient of variation (CV). Detrended fluctuation analysis was used to estimate α. RESULTS: The fractal scaling index was significantly higher (i.e. risk of falling smaller) for walking on the rougher surface (P < 0.0001; Wilcoxon rank sum test): running track (SDa = 0.28 m/sec2): α ∼0.96 (range .88-1.10); tennis court (SDa = 0.09 m/sec2): α ∼ 0.83 (.76-.89). The fractal scaling index for shuffled gait stride data was similar to that for uncorrelated white noise (α ∼ 0.5). No significant difference between the two surfaces was observed for the SDw and CV of the gait strides (P > 0.4). CONCLUSIONS: Changes in walking surface roughness over a range typically encountered during activities of daily living affect α, but not other measurements of gait variability. In particular walking surfaces that are slightly rougher than those typically encountered in residential and clinical settings (e.g. SDa ∼0.1 m/sec2) may offer advantages for minimizing risk of falling.