Perturbation Training Post-Stroke: Examining Adaptation and Generalization for Fall-Risk Reduction

Falls are a common complication after stroke and an important public health issue (Batchelor et al. 2012b; Jorgensen et al. 2002). Falls may result in minor injuries such as bruises, grazes to serious injuries such as hip fracture, head injury, leading to severe debilitating physical and psychosocial consequences. These consequences have serious implications thereby deteriorating functional outcomes, limiting independence and worsening quality of life post-stroke (Batchelor et al. 2012b). In addition to being the leading cause of injury related to deaths, falls are also associated with significant economic burden totaling to a direct medical costs of $179 million for fatal falls and $19 billion for nonfatal fall injuries (Novitzke 2008; Stevens et al. 2006). Stroke induced sensorimotor deficits, cognitive impairment and postural dysfunction are the primary contributors towards high frequency of falls (Bansil et al. 2012; Sun et al. 2014; Yanohara et al. 2014). Studies report that about 14% to 65% of people with stroke fall at least once during acute hospital care (Davenport et al. 1996; Nyberg and Gustafson 1995; Teasell et al. 2002). Additionally, during inpatient rehabilitation about 10.5 to 47% experience a fall, with 5 to 27% falling equal to or more than twice. After discharge from the hospital, people with stroke continue to experiencing falls during the next six-months accounting to approximately 37% to 73% (Mackintosh et al. 2005; Mackintosh et al. 2006). Even during the chronic phase of recovery, about 40% to 70% of people with chronic stroke (PwCS) experience falls indicating that falls are a major concern during all phases of stroke recovery (Andersson et al. 2006; Hyndman et al. 2002; Lamb et al. 2003; Weerdesteyn et al. 2008a). Thus, despite achieving independent mobility, PwCS continue to demonstrate balance deficits due to the incomplete restoration of sensori-motor, balance and/or cognitive impairments resulting in high incidence of falls (Bansil et al. 2012; Sun et al. 2014; Yanohara et al. 2014). Moreover, regaining community ambulation further predisposes PwCS to falls from unpredicted environmental slip or trip perturbations accounting to up to 34% of overall falls (Schmid et al. 2013). Thus, with an aim to reduce fall-risk, it is crucial to develop comprehensive and effective fall-risk prevention interventions targeting on improving well-established fall-risk predictors among PwCS. Several clinical balance measures such as Berg Balance scale (Maeda et al. 2015), Timed Up and Go test (Bower et al. 2019) and Activity-specific balance confidence scale (Simpson 2011) along with instrumented spatio-temporal gait parameters (Bower et al. 2019) and Posturography have been established as sensitive fall-risk predictors among PwCS. Recently, perturbation based behavioral studies involving exposure to external slip and trip perturbations and inducing a loss of balance during stance and walking, similar to those in real-life have been used to accurately identify fall-risk factors (Inness et al. 2014; Joshi et al. 2018; Mansfield et al. 2012; Salot et al. 2016). These studies in PwCS have identified reduced stability (indicating the simultaneous control of center of mass (CoM) position and velocity relative to the perturbed base of support (BoS)), poor vertical limb support (reactive vertical force generated from the stance limb in order to resist downward fall) and impaired compensatory stepping responses (to re-establish the BoS and position the CoM within the destabilized BoS) as crucial biomechanical fall-risk predictors similar to those in healthy older adults (Joshi et al. 2018; Pai and Bhatt 2007; Salot et al. 2016). Given the multifactorial nature of falls, comprehensive conventional fall prevention interventions have been previously designed with a wide-ranging approach to improve the risk factors for falling among people with stroke (Batchelor et al. 2012b; Schmid et al. 2013). However, majority of these interventions have shown to be targeting voluntary movement and anticipatory postural control, without specifically training balance recovery mechanisms which are necessary to prevent an actual fall (Lubetzky-Vilnai and Kartin 2010; van Duijnhoven et al. 2016). Thus, these interventions showed limited success in fall reduction due to lack of task-specificity (Batchelor et al. 2012a; Grabiner et al. 2014; Verheyden et al. 2013), thereby demanding development of an effective balance training interventions for fall prevention. Reactive balance control in the form of a rapid and effective compensatory stepping response play a crucial role in regaining postural stability and reducing fall-risk following exposure to an unexpected perturbation (Maki et al. 2006). An effective compensatory stepping response helps to reposition the displaced center of mass and re-establish the base of support, in order to avert a fall (Maki et al. 2006; McIlroy and Maki 1996). Perturbation-based studies have indicated that PwCS have impaired reactive balance control resulting in delayed compensatory stepping, multiple stepping responses, shorter step length, delayed and abnormal muscle activation patterns along with a need for external assistance (Lakhani et al. 2011; Mansfield et al. 2012; Salot et al. 2016). Recently, perturbation training paradigm has emerged as a task-specific intervention that aims to improve reactive balance control via compensatory stepping mechanism, by simulating loss of balance in a safe laboratory environment similar to real-life falls (Gerards et al. 2017; McCrum et al. 2017; Pai and Bhatt 2007). Several studies performed in healthy older adults have demonstrated the potential of perturbation training to induce adaptive changes in fall-resisting skills via improving postural stability, stepping responses and reducing falls (Bhatt and Pai 2009a; Bhatt et al. 2006b; Bhatt et al. 2012; Mansfield et al. 2010; Okubo et al. 2019; Pai et al. 2014a). On similar lines, preliminary studies performed in people with stroke using small magnitude repeated pull-push perturbations have demonstrated improved reactive balance control (Mansfield et al. 2018) and stepping responses (Mansfield et al. 2011) along with reduced recovery steps (Schinkel-Ivy et al. 2019) indicating adaptive changes in the reactive balance control. Further, exposure to two, consecutive, mid-sized over ground walking slips have shown improved postural stability and reduced fall-risk among PwCS (Kajrolkar et al. 2014). Moreover, Bhatt (2019) demonstrated significant improvement in postural stability and laboratory falls while Nevisipour (2019) showed improved trunk control among PwCS following large magnitude, treadmill-based stance slip and trip perturbation training, respectively. Thus, there is considerable literature supportive of preserved ability of PwCS to undergo short- term reactive adaptation in fall-resisting skills and reducing fall-risk with a single session of training. Additionally, preliminary study by Bhatt (2019) also demonstrated the ability of PwCS to retain the acquired adaptive changes following a single session of stance perturbation training for up to 3 weeks post-training. Recently, multisession perturbation training has also been performed in individuals with stroke to understand the effect of greater training dosage on fall-resisting skills and improvement in retention of fall-resisting skills (Handelzalts et al. 2019; Pigman et al. 2019a; Punt et al. 2019; van Duijnhoven et al. 2018). In addition to adaptation, healthy adults have also shown the ability of the CNS to effectively transfer acquired motion state adaptations to untrained environmental and task constraints, and even to untrained limb (Morton and Bastian 2004; Morton et al. 2001; Sainburg and Wang 2002; Seidler 2004). Studies of generalization of fall-resisting skills from trained to untrained tasks depend on if the task requires motor skill learning or adaptive changes in sensory/perceptual systems (Reynolds and Bronstein 2004). In the context of fall resisting skills, healthy adults have demonstrated transfer from slip adaptation to novel trip task (Bhatt et al. 2013), from sit-to-stand slip to walking slip (Wang et al. 2011); stance trip adaptation to walking trip (Grabiner et al. 2012), from waist pull perturbations to treadmill slips (Martelli et al. 2018) and stepping in-place on the rotating disk to hopping on both feet (Earhart et al. 2002). Similarly, generalization of acquired skills fall-resisting to different effectors contribute towards inter-limb transfer (Bhatt and Pai 2008; Marcori et al. 2020; McCrum et al. 2020; Van Hedel et al. 2002). Likewise, motor learning studies in individuals with stroke have demonstrated preliminary evidence on presence of generalization of learned motor skills. Locomotor studies performed in PwCS have shown significant generalization of acquired adaptation in step symmetry from trained split-belt treadmill walking to untrained task of over-ground walking (Reisman et al. 2009; Savin et al. 2014). With regards to inter-limb transfer, few studies have been performed in upper limb training (Ausenda and Carnovali 2011; Iosa et al. 2013; Yoo et al. 2013) indicative of intact ability of post-stroke impaired nervous system to show some degree of generalization. However there are no studies focusing on inter-limb transfer effect of fall-resisting skills from trained to untrained limb in PwCS. Although the ability to acquire reactive adaptation is well-preserved among PwCS, most studies have provided low to mid-sized magnitude of perturbations, with most perturbations during standing. There is limited evidence on adaptive changes following large magnitude perturbations during stance and walking.