Treadmill Perturbation

DISCUSSION
Treadmill-perturbation training could effectively reduce falls upon an unannounced, novel slip resulting from the improvements in both proactive and reactive control of stability among community-dwelling older adults (lending support to Hypothesis One). Though merely walking on the treadmill could improve participants’ proactive control of stability, it could not improve their reactive control of stability nor effectively reduce their risk of falls (lending support to Hypothesis Two). Finally, pertaining to its ability to reduce rate of falls and to improve stability, treadmill-perturbation training was indeed significantly inferior to overground-perturbation training (lending support to Hypothesis Three).

Treadmill-perturbation
training could improve older adults’ resistance to falls as evident that these subjects subsequently carried such training effects over to overground walking. They learned to proactively forward shifting their COM position relative to BOS (Fig. 5b), which, in line with the results from the young adults in our previous study (13). In addition, the feature common to both age samples was that the improvement in reactive control was far greater than that in proactive control of stability (12). The improvement in reactive control of stability in seniors was by 44.1% (Tt versus Tc) in comparison to only by 20.8% in proactive control (Figs. 5a and 6a). These findings supported the notion that treadmill-perturbation training similarly affecting people across a wide range of ages (19-79 years old).

Rather unexpectedly, even without perturbation, treadmill walking could still promote older adults’ anterior shift in their COM relative to BOS and could generalize such changes in subsequent overground walking (cf. Groups Tc and Oc in Fig. 5b). The anterior shift of COM position would increase their stability against backward loss of balance, and hence accounted for the fact that 6.7% of the participants who walked on treadmill without perturbation could still fully recover from a novel overground slip without the need of taking a protective step, which could never happen in the overground-control group (cf. Groups Tc and Oc in Fig. 4). Yet, while the proactive control of stability represents a first line of defense against falls, it is the reactive stability control that clearly played decisive role here (26). Contrasting the results of perturbation-training with volitional exercises, such as walking on treadmill without perturbation as in the current study, it appears that only perturbation training can induce changes in reactive control of stability. This could be a key distinction between volitional exercises (28-30) and perturbation training in which disturbance occurs without a person’s knowledge and in which movement errors will be generated (31). Sensing such errors is essential for the CNS to recalibrate its internal representation of the stability limits (8, 32) and to update the motor commands that can partially or fully correct the errors when a similar disturbance appears (33). The rapid changes in the control of stability made after just a few trials of perturbation training could indeed be better accounted for by assuming CNS’ error-driven recalibration of the stability limits rather than by the prospects of improved muscle strength, joint flexibility or newly acquired motor skills as from Tai-Chi practice obtained – all of which require weeks or months of volitional exercises.

The treadmill-perturbation was significantly inferior to overground-perturbation training when resisting falls resulted from a slip during overground walking as expected (Figs.
4 and 6a). The improvement in the reactive control of stability obtained after treadmill-perturbation training (Tt) was only about 40% of those resulted from overground-perturbation training (Ot) when using the difference between the reactive control of stability of Ot and Oc as the reference (Fig. 6a). This difference between these two groups could account for 13.9% more falls in the former than the latter group (Tt versus Ot, Fig. 4). The similarity between the training (from the 1st to the 23rd slip) and the generalization test (the 24th slip) in overground-perturbation training group might explain why its results were better in comparison to those of the treadmill-perturbation training group. During overground training, all slips occurred consistently under the right side (limb). During treadmill training, the device could not precisely control the onset timing of the perturbation with respect to the side (limb) or to the gait cycle, and hence subjects received slips under both sides or at different gait cycle (not always ~17 ms after right heel strike). Approximately half of perturbations occurred under the left side (11). In addition, subjects were not afforded the learning opportunity to change the relative relationship between the leading and the trailing limbs on treadmill as in comparison to the overground perturbation. Its belt speed was dictated by the computer profiles (Fig. 2c), and could not be actively controlled by the participant. In contrast, an individual control of the ground reaction force beneath each foot during the overground training allowed this subject to actively slow down the slip (BOS) velocity as well as to reduce the slip distance. These factors could also account for the differences found in the two training
groups.

In summary, the treadmill-perturbation training was able to improve both proactive and reactive control of stability in older adults. Although the improvements were less than that in the overground-perturbation training, the treadmill-perturbation training remarkably reduced the likelihood of falls upon a novel slip during overground walking, which suggested a considerable generalization of training effect in older population.