Reliability of Gait Parameters Using a Treadmill System
Reliability of Gait Parameters Using a Treadmill System
Average gait speed during over–ground walking (1.35 ± 0.14 m.s) was significantly faster than during treadmill walking (F = 15.0, P < .001). Mean spatial and temporal parameters for treadmill walking during the three gait sessions are presented in Table 1 . With the exception of foot rotation angle and step width, there were statistically significant main effects for test session for all temporal and spatial gait parameters. In session 3, participants walked, on average, 0.13 m.s faster, increased their cadence by 5 steps.min, and adopted stride lengths and step times that were around 6% longer and 5% shorter, respectively, than in sessions 1 and 2. Session three was also accompanied by a concomitant decrease in the duration of double support. There were no statistically significant differences, however, between the first and second gait trials for any spatiotemporal parameter.
Table 2 illustrates the topic maxima in the vertical GRF for the three test sessions. There was a significant main effect for the magnitude and timing of the vertical GRF braking peak and the final propulsive peak across the sessions. While the vertical propulsive peak was significantly lower in the first session compared to subsequent sessions, the braking peak was significantly higher in the final session when compared to the previous walking trials.
When expressed as a percentage of the mean, the SEM was less than 10% for the majority of gait variables, including step length (Figure 3), for both within– and between–day comparisons ( Table 3 ). The exceptions were step width (SEM 10% of between–day mean), left foot rotation angle (SEM 14% of between– and within–day mean) and right foot rotation angle (SEM 11% of between– and within–day mean). Foot rotation angle and self–selected walking speed had the greatest variability of all gait variables ( Table 3 ). Temporal gait parameters were typically more consistent than spatial measures and variability for all parameters was generally smaller within– rather than between–days. While the SEM of vertical ground reaction force peaks were less than 5% of the mean values for within– and between–day comparisons, the 95% limits of agreement ranged between ± 53 and ± 63 N ( Table 4 ).
(Enlarge Image)
Figure 3.
Bland and Altman plot for step length of the left leg. Bias (solid line) and RC95% (dashed lines) for step length between sessions 2 and 3 (within–day) (a) and between sessions 1 and 2 (between–day) (b) and between sessions 1 and 3 (between–day) (c). Note that the RC95% is mathematically identical to the Minimum Detectable Change, which represents the minimum change in score (at an individual level) that reflects true change (with 95% confidence), rather than measurement error alone. Negative values reflect an increase in step length.
Results
Average gait speed during over–ground walking (1.35 ± 0.14 m.s) was significantly faster than during treadmill walking (F = 15.0, P < .001). Mean spatial and temporal parameters for treadmill walking during the three gait sessions are presented in Table 1 . With the exception of foot rotation angle and step width, there were statistically significant main effects for test session for all temporal and spatial gait parameters. In session 3, participants walked, on average, 0.13 m.s faster, increased their cadence by 5 steps.min, and adopted stride lengths and step times that were around 6% longer and 5% shorter, respectively, than in sessions 1 and 2. Session three was also accompanied by a concomitant decrease in the duration of double support. There were no statistically significant differences, however, between the first and second gait trials for any spatiotemporal parameter.
Table 2 illustrates the topic maxima in the vertical GRF for the three test sessions. There was a significant main effect for the magnitude and timing of the vertical GRF braking peak and the final propulsive peak across the sessions. While the vertical propulsive peak was significantly lower in the first session compared to subsequent sessions, the braking peak was significantly higher in the final session when compared to the previous walking trials.
When expressed as a percentage of the mean, the SEM was less than 10% for the majority of gait variables, including step length (Figure 3), for both within– and between–day comparisons ( Table 3 ). The exceptions were step width (SEM 10% of between–day mean), left foot rotation angle (SEM 14% of between– and within–day mean) and right foot rotation angle (SEM 11% of between– and within–day mean). Foot rotation angle and self–selected walking speed had the greatest variability of all gait variables ( Table 3 ). Temporal gait parameters were typically more consistent than spatial measures and variability for all parameters was generally smaller within– rather than between–days. While the SEM of vertical ground reaction force peaks were less than 5% of the mean values for within– and between–day comparisons, the 95% limits of agreement ranged between ± 53 and ± 63 N ( Table 4 ).
(Enlarge Image)
Figure 3.
Bland and Altman plot for step length of the left leg. Bias (solid line) and RC95% (dashed lines) for step length between sessions 2 and 3 (within–day) (a) and between sessions 1 and 2 (between–day) (b) and between sessions 1 and 3 (between–day) (c). Note that the RC95% is mathematically identical to the Minimum Detectable Change, which represents the minimum change in score (at an individual level) that reflects true change (with 95% confidence), rather than measurement error alone. Negative values reflect an increase in step length.
