Unexpected perturbations training improves balance control and voluntary stepping times in older adults - a double blind randomized control trial

Falls are a major problem among elderly population; they are the leading cause of injury above the age of 65 [1]. Of those who fall in the U.S., 20 to 30 % suffer moderate to severe injuries that reduce mobility and independence, and increase the risk of death [2]. The financial impact in 2000, adjusted for inflation, was $30 billion and is expected to reach $67.7 billion by 2020 [1].

Walking is the major activity in which large proportion of falls in older adults occurs [3]. Sixty percent of outdoor falls among older adults resulted from slips or trips [4]. Even among older adults capable of independent walking, there could be a substantial decline in balance performance, which does not become evident until a slip or a trip happens [5]. In fact, the inability to step rapidly in response to unexpected loss of balance ultimately determines whether a fall occurs [6, 7]. Thus, a better way to improve balance, improve stepping and reduce risk of falls may be to direct preventive efforts towards older adults who have not yet fallen. Until recently these balance recovery responses were considered hardwired postural reflexes that could not be influenced by training. However, [8‐10] it was showed that older adults were able to adapt in a reactive manner after participation in a perturbation exercises that challenged the mechanisms responsible for dynamic stability (i.e., increase in base of support and counter-rotating segments around the center of mass).

A number of studies have begun to examine the effect of perturbation training on balance of older adults. Shimada et al. [11] found improvement in mobility and a trend to fall reduction after split treadmill training. This training method is a very unnatural condition, given that most people walk with the same velocity in each leg. Other studies have other issues. Pai et al. [12] showed a rapid decrease in loss of balance in response to multiple presentations of a slip perturbation after rising from sit to stand. Mansfield et al [13] found that older adults with a history of falls or instability reduced the frequency of multi-step reactions and foot collisions after perturbation training while standing or walking in place. The training methods perturb the balance of their participants from sit to stand or during standing or walking in place, which may not be as relevant in a natural setting as might be a perturbation while walking. Melzer and Oddsson [14] found improvement in voluntary stepping, and balance control, in an exercises program that incorporate mild external balance perturbation exercises applied by the instructors; and Halvarsson et al. [15] found that old fallers that suffered from fear of falling, decreased their fear of falling, and voluntary stepping times during dual-task performance and increased velocity of walking post perturbation training.. However, in these program the perturbations of posture were expected, and not random. Recently it was showed that unidirectional translational treadmill training (i.e., a laboratory-induced trip) reduced falls [16, 17]. Bhatt et al. [18] found that inducing unannounced right-leg slips, participants significantly reduced fall and balance loss incidence. Pai, et al. [19] found that a single session of repeated-slip exposure reduced older adults’ annual risk of falls from 34 to 15 % (p < 0.05) especially among those who had history of falls. The above protocol provided an anterior perturbation, causing a backwards “slip” initiated always by on the right foot. Participants might have learned and expected the right-leg slips perturbations. In a recent meta-analysis [20] that include 8 perturbation-based balance training studies (n = 404) participants reported fewer falls than those in the control groups.

Motivated by the above perturbations training studies and trying to accommodate for some of the issues mentioned above (i.e., perturbation training while standing or walking in place; highly predictable repeated-right leg slip exposure; a very unnatural walking on split treadmill), we aimed to explore whether unexpected multidirectional perturbation training while walking on a treadmill [21] can reduce risks of falls in independent older adults. A perturbation exercise while walking provides a more realistic balance training that is sufficiently task-specific so that responses on this training regime will be more likely to be transferred to other measures of balance control and voluntary stepping measures.

Our hypotheses were that following exposure to a gait training program that includes unexpected perturbations exercises during walking will significantly improve voluntary stepping times as well as balance control in older adults, two factors that are associated with falls and injuries related with falls [22‐27]. We believe that perturbation exercises that will challenge both balance control during walking as well as trigger a quick stepping responses to avoid fall during walking. These postural response following an external perturbation receives a higher priority than a voluntary action thus can be incorporated into centrally programmed voluntary movements [28]. This concept should be of importance for balance training and it further supports the notion that postural perturbations should be incorporated into balance training programs.

Other than age and height there were no significant differences between the groups at baseline (Table 2). During the training period we had 6 drop outs in the intervention group and 7 in the control group (Fig. 1).

Table 3 show that the perturbation training resulted in a significant group-by-time decrease in AP-sway (p = 0.012, [ES] = 0.59). There were also a trend towards a group-by-time decrease with a large effect size in sway velocity, sway area and ML-sway post-training (p = 0. 115, [ES] =0.72; p = 0.119, [ES] =0.74; and p = 0.142, [ES] = 0.58, respectively); although these differences were not statistically significant, the effect size is considered to be moderate. The Stabilogram-Diffusion analysis in eyes closed condition showed a significant group-by-time decrease in Dxs, (p = 0.012, [ES] =0.78) and a trend towards significance in Cdy and Dys (p = 0.028, [ES] = 0.92; and p = 0.095, [ES] =0.65, respectively) (Table 4).

Table 5 shows that the perturbation training resulted a significant group-by-time interaction for foot contact time of the voluntary step execution in both the single and dual task conditions (p = 0.002, effect size [ES] = 0.75 and p = 0.003 effect size [ES] = 0.89 respectively). In addition, a significant group-by-time interaction for the voluntary step reaction time in dual task condition (p = 0.010), and a trend towards a significant group-by-time interaction for single task condition (p = 0.057).

We found no significant group-by-time interaction effect for all components of the LLFDI as well as for FES and POMA.

With respect to side effects and adverse events, during the exercise training program, muscle soreness was experienced by some subjects, especially in the early stage of the training. Those effects were managed by adjusting the training intensity and the symptoms disappeared during training.

The results support in part our main hypotheses, unexpected perturbations training while walking can improve the ability to voluntarily step rapidly and standing balance control in older adults. These parameters have been shown in the past to predict injury from falls [19, 24]. This shows that the benefits of unexpected perturbation during walking were generalized to other aspects of balance. Our findings are supported by Pai and Bahtt [40] that suggested that the central nervous system makes adaptive improvements in proactive and reactive control of stability as a result of trial and error perturbation practice (i.e., a forward slip). They suggested that the central nervous system probably decreases the reliance on feedback corrective mechanisms for successful recovery and builds an internal representations to improve its feedforward control while walking. Pai and Bahtt training [40] used forward slip practice, the training program, while in the present study the direction of the perturbation was highly unpredictable (forward, backward, right and left perturbations), the intervention group subjects were unable to predict the direction of perturbation during the training. Thus it is unlikely that the improvement was due to feedforward control. Our results support the notion that the central nervous system probably increased the reliance on feedback corrective mechanisms for successful recovery. This is supported by the results of the SDA.

The significant improvement in the intervention group and the large effect size in the SDF parameters (ES = 0.65–0.92) in eyes closed conditions are promising. Age-related decrease in SDF parameters are well documented [30, 33] and indicates that a COP tends to drift away from the equilibrium point is a predictor to falls in elderly persons [26, 27, 29, 30, 41‐43]. The improvement in the Dxs and Cdy as well as tendency towards improvement in Dys of SDA in eyes closed condition showed in this study indicates that the deterioration of balance control could be reversed by perturbation balance training. AP balance control in eyes close condition (Dys and Cdy) were found to be an important risk indicator of falls and injurious falls [26, 27]. Laughton et al. [41] found a greater Dys in elderly fallers compared with the young’s, and greater AP and ML sway in older adults who demonstrated lower scores in the POMA. Kurz et al. [27] found that a deterioration of AP postural control present higher risk of serious injury. Most Studies usually used traditional balance measures, which provide descriptive information on the postural sway, however it is very difficult to understand the mechanism of postural control using traditional COP statistics, this we performed SDA. We found training effects on open- and close-loop control mechanism when vision was occluded as indicted by a significant decrease in Critical Displacement in AP direction (Cdy). . Improvements in and Short-term Effective diffusion coefficients in ML direction (Dxs) as well as a trend towards a decreased in Short-term Effective diffusion coefficients in AP direction (Dys) further support this notion. These parameters reflect the degree of sway in the short term region. A decreased tendency to continue sway in an ongoing direction (Table 4) indicates a behavior that reflects a more stable balance control system. This indicates that the intervention group was able to better detect COP movement under their feet and initiate a more effectively close loop balance control. These suggest that the improved closed-loop balance control when vision was occluded would likely be from proprioceptive and/or maybe vestibular sources.

The improvement in step execution in the intervention group was seen during the step execution times i.e., foot contact times, in both ST and DT (72 ms in ST and 131 ms in DT, see Table 5). A shorter step execution indicates improvement in the ability to prevent a fall if balance is lost, consequently, the risk of fall and injury maybe reduced [22‐24]. The improvement in the step execution in the intervention group was accompanied with improvement in step reaction times, in particular under dual task conditions. The step reaction time duration is mainly dependent on sensory detection thresholds, nerve conduction velocities and central neural processing times. A shortened step reaction time in dual task condition suggests that the central neural processing time was improved as a result of perturbation training. This indicate that the executive functioning was improved, subject were able to step quickly while their attention was allocated elsewhere. The executive functioning is related to the ability to rapidly shift attention from a cognitive task to the stepping task. This could be interpreted also in terms of automaticity of the stepping behavior as an essential characteristic of the central reorganization process. Therefore, we could assume that the interference effect during dual task stepping was reduced as a result of the treadmill perturbation training.

Halvarsson et al. [15] also found significant improvements in stepping performance, in a group of elderly fallers that performed perturbation exercises. Melzer and Oddsson [14], Mansfield et al. [13] and Rogers et al. [44] also found that specific step training that includes perturbation of balance improves stepping abilities in older adults. Pai et al. [12] have shown reductions in falls and balance loss following a repeated-slip during walking exposure. The same group had further showed that those gains could be retained for 6 months [18] and cut older adults’ annual risk of falls by 50 % especially among elderlies who had history of falls [19]. Our results and the results above suggests that specificity-of-training principle is a major factor in achieving treatment goals and therapists need to tailor balance perturbation training programs to target functional aspects of balance control such as the ability to step rapidly [45].

It is still unknown how the improvement seen in balance and stepping carries over to real-life falls. It may be that carryover to fear of falls and physical performance as measure with POMA did exist but was not detected by these outcomes measures due to insufficient sample size or due to ceiling effects, subjects were close to score the highest score in both measures. We also did not find significant carryover effects on self-reported and performance based function. This suggests that function is not derived only by the ability to perform balance tasks, but is influenced by environmental and behavioral factors. Most perturbation training programs did not measure self-reported physical function [16‐19, 40, 44]. However other studies found improvement in self-reported function [14] and fear of falls [15]. This two studies trained balance in a group setting adding also behavioral factors. This may suggest that physical intervention per-se without behavioral intervention cannot change the levels of physical function. The fact that Performance based measure are weakly associated with self-reported mobility in healthy elderly persons (r = 0.21–0.29) [46] support this notion.

This study has several limitations. First, the data came from a fairly small sample of independent older adults thus the results cannot be generalized to frail or institutionalized elderly persons. Second, the training program was done on a treadmill and not on leveled ground. Treadmill walking is somewhat different then over ground walking and might pose different demands on the trainee. However, it has been showed that fall-resisting skills acquired from such training can be transferred to over-ground walking [47]. Third, to provide stronger evidence of the clinical efficacy of this training procedure, future studies should compare the current intervention with an alternative treatment such as Tai Chi, or strength training as this was found beneficial in a recent review [48]. Also there is a need to determine whether this type of training improves the ability to cope with falls that resulted from slips and trips only, or can be transferred to any kind of loss of balance. Fourth, further research is needed in order to determine whether randomly ordered perturbations exercises (Random practice) would render better results than repeating the same perturbation exercises (blocked practice) in terms of acquisition, adaptation and retention.

Participation in a balance training program that includes unexpected perturbation of balance during treadmill walking reduced the risk factors for falls as presented by biomechanical markers that have been shown in the past to identify fallers and predict falls. However levels of physical functioning were not improved, this suggests that preventing sedentary lifestyle for the elderly require additional behavioral intervention.