|UFDC Home||myUFDC Home | Help|
This item has the following downloads:
1 MOTOR CONTROL SOLUTIONS TO DYNAMIC STABILITY DEFICITS DURING WALKING AFTER SPINAL CORD INJURY By KRISTIN ALAYNE VAMVAS DAY A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY UNIVERSITY OF FLORIDA 2010
2 2010 Kristin Alayne Vamvas Day
3 T o my husband, Scott and our future family
4 ACKNOWLEDGMENTS As I complete this major milestone in my academic career, I am amazed at what I have experienced and accomplished over the last four and a half years. At the same time I also am fully aware that I did not walk alone on this journey. W ords cannot fully express the gratitude I have to so many for standing by my side during this marathon process: mentors, family and friends. I would like to express my sincerest gratitude to my primary mentor, Dr. Andrea Behrman. The undying passion she holds for her work both as a researcher and clinician inspires me every single day. She has shown me what it means to be a true clinical researcher; I now know it is possible to have a career based on what your heart guides you to do. In addition to Dr. Behrman, I have been extremely grateful to have one of the strongest graduate committees possible: Drs. Steve Kautz, Dena Howland, and Craig Velozo. Every graduate student should be so fortunate to have a committee who cares enough to find the perfect balance between suppor t and challenge. It i s that balance which helped me find my potential as a budding clinical researcher. I would like to extend a sincere t hank you to the T32 Neuromuscular Plasticity Training program for providing several years of financial support and endless opportunities for growth. I am grateful to the Florida Brain and Spinal Cord Injury Program for the h onor of being an Early Career Rehabilitation Research Award re cipient My appreciation also goes to the Department of Physical Therapy and Ms. Glor ia Miller for developing a creative teaching assistantship position that continued to support me through the last semester of my doctoral training.
5 This journey would not have been as enjoyable as it was had it not been for the many friend s I have made along the way. Those friends are the many who welcomed me to the university back in 2006 and others who arrived along the way. I am especially grateful to my coffee chat friends for the great times when we just needed time to talk and people to listen. M y family also has been an incredible support team This includes not only my immediate family, but also my extended family and inlaws. Even with most of them several states away, I always knew they were proud of me, n o matter what I did or did not accompl ish In the end, they just wanted me to be happy and healthy Finally, I believe more than a lifetime would be needed to express to my husband Scott, exactly what he ha s meant to me throughout this process Every dream of mine, he has encouraged, and earning a doctorate has been no exception. In times of doubt, he showed me my capabilities. In times of frustration, he gave me laughter. When separated by almost a thousand miles for two years, he showed me what unconditional love and support really are. For everything, I am eternally grateful.
6 TABLE OF CONTENTS page ACKNOWLEDGMENTS .............................................................................................................. 4 TABLE OF CONTENTS .............................................................................................................. 6 LIST OF TABLES ......................................................................................................................... 9 LIST OF FIGURES ..................................................................................................................... 10 ABSTRACT ................................................................................................................................. 12 CHAPTER 1 LITERATURE REVIEW ...................................................................................................... 14 Introduction .......................................................................................................................... 14 Neural Control of Locomotion ........................................................................................... 17 Subtasks of Walking .................................................................................................... 17 Spinal (Central) Pattern Generation ......................................................................... 18 Evidence from Animal Literature ............................................................................... 19 Evidence from Human Literature............................................................................... 24 Spinal Cord Injury Rehabilitation ...................................................................................... 27 Traditional Rehabilitation of Walking ........................................................................ 27 Activity Dependent Plasticity ...................................................................................... 28 Evidence from Animal Literature ............................................................................... 28 Evidence from Human Literature............................................................................... 32 Pa radigm Shift toward Recovery ............................................................................... 35 Determinants of Balance Control ...................................................................................... 38 Neural Systems Underlying Balance Control ................................................................. 41 Brain Stem .................................................................................................................... 41 Cerebellum .................................................................................................................... 42 Basal Ganglia ............................................................................................................... 43 Cerebral Cortex ............................................................................................................ 44 Spinal Cord and Descending Pathways ................................................................... 44 Balance Control in SCI ....................................................................................................... 45 Outcome Measures of Walking Balance ......................................................................... 50 Current Laboratory Based Measures of Balance Control ..................................... 50 Limitations of Laboratory Based Measures ............................................................. 55 Current Clinical Measures of Balance Control ........................................................ 56 Limitations of Clinical Measures ................................................................................ 5 9 2 CLINICAL RELEVANCE AND RATIONALE ................................................................... 67
7 3 METHODOLOGY: PROCEDURES AND CONSIDERATIONS ................................... 73 Participants .......................................................................................................................... 73 Clinical Assessments .......................................................................................................... 74 Biomechanical Data Collection Procedures .................................................................... 74 D ata Acquisition and Processing ...................................................................................... 76 Dynamic Stability Outcome Measures and Rationale ................................................... 77 Experiment #1 .............................................................................................................. 77 Experiment #2 .............................................................................................................. 79 Experiment #3 .............................................................................................................. 80 Locomotor Training Intervention ....................................................................................... 82 4 EXPERIMENT 1: INFLUENCE OF ASSISTIVE DEVICES ON HEAD STABILIZATION DURING WALKING AFTER SPINAL CORD INJURY ................... 89 Methods ................................................................................................................................ 92 Participants ................................................................................................................... 92 Experimental Procedures ........................................................................................... 93 Data Acquisition and Processing............................................................................... 94 Outcome Measures ..................................................................................................... 95 Data Analysis ................................................................................................................ 96 Results .................................................................................................................................. 97 With versus Without Assistive Devices in SCI ........................................................ 97 SCI Without Assistive Devices versus Controls at Matched Speeds .................. 98 Discussion ............................................................................................................................ 99 Assistive Devices Impact Head Motion Relative to the Pelvis .............................. 99 Spinal Cord Injury Affects Head Stability Multidirectional ly ................................ 101 Conclusions ........................................................................................................................ 102 5 EXPERIMENT 2: FOOT PLACEMENT VARIABILITY AS A WALKING BALANCE CONTROL MECHANISM POST SPINAL CORD INJURY .................... 110 Methods .............................................................................................................................. 114 Participants ................................................................................................................. 114 Experimental Procedures ......................................................................................... 114 Data Acquisition and Processing............................................................................. 116 Outcome Measures ................................................................................................... 116 Data Analysis .............................................................................................................. 117 Results ................................................................................................................................ 118 Variability of Outcomes: SCI versus Controls (Hypothesis 1) ............................ 118 Variability of Outcomes within Participant Group (Hypothesis 2) ...................... 119 Associations Among Outcome Magnitudes by Participant Group (Hypothesis 3) ......................................................................................................... 119 Associations Among Outcome Variability by Participant Group (Hypothesis 3) ......................................................................................................... 120 Discussion .......................................................................................................................... 121 Spinal Cord Injury Alters Movement Variability ..................................................... 121
8 Variability Differences within Groups ...................................................................... 123 Variability May Be a Better Clinical Correlate for Balance than Magnitude ..... 124 Suggestions for Clinical Translation ....................................................................... 125 Further Clinical Considerations within a Recov ery Based Framework ............. 126 Conclusions ........................................................................................................................ 127 6 EXPERIMENT 3: DIFFERENTIAL EFFECTS OF MANUAL ASSISTED VERSUS ROBOTIC ASSISTED LOCOMOTOR TRAINING ON DYNAMIC STABILITY POST SCI ..................................................................................................... 136 Methods .............................................................................................................................. 139 Participants ................................................................................................................. 139 Locomotor Training Intervention.............................................................................. 140 Experimental Procedures ......................................................................................... 141 Data Acquisition and Processing............................................................................. 142 Dynamic Stability Biomechanical Outcome Measures ........................................ 143 Primary outcome ................................................................................................. 143 Secondary outcomes .......................................................................................... 143 Data Analysis .............................................................................................................. 144 Results ................................................................................................................................ 145 Pre Locomotor Training Biomechanical Outcomes .............................................. 145 Post Locomotor Training Biomechanical Outcomes ............................................ 145 Discussion .......................................................................................................................... 146 MLT and RLT Both Trained Dynamic Stability Post SCI ..................................... 147 Adaptive Movement Strategies Developed to Maintain Dynamic Stability Post LT ..................................................................................................................... 148 AIS Categorization Does Not Reflect Dynamic Stability ...................................... 150 Conclusions ........................................................................................................................ 151 7 CONCLUSIONS ................................................................................................................ 159 Experimental Limitations .................................................................................................. 159 Overall Conclusions .......................................................................................................... 160 Summary and Future Work ............................................................................................. 162 REFERENC ES ......................................................................................................................... 164 BIOGRAPHICAL SKETCH ..................................................................................................... 182
9 LIST OF TABLES Table page 4 1 Participant demographics ........................................................................................... 104 5 1 Participant demographics ........................................................................................... 129 5 2 Associations between mean outcome magnitudes for participants with SCI at SS speeds. .................................................................................................................... 134 5 3 Associations between mean outcome magnitudes for healthy controls at 0.3 m/s. ................................................................................................................................. 134 5 4 Associations between mean outcome magnitudes for healthy controls at 0.6 m/s. ................................................................................................................................. 134 5 5 Associations between outcome variabilities for participants with SCI at SS speeds. ........................................................................................................................... 135 5 6 Associat ions between outcome variabilities for healthy controls at 0.3 m/s. ...... 135 5 7 Associations between outcome variabilities for healthy c ontrols at 0.6 m/s. ...... 135 6 1 Participant demographics separated by intervention group .................................. 153 6 2 SCI self selected treadmill speeds before and after intervention ......................... 154 6 3 Magnitude of standardized outcomes separated by intervention ......................... 155
10 LIST OF FIGURES Figure page 1 1 Modulation of central/spinal pattern generator (CPG) circuitry ............................... 61 1 2 Progression of functional walking recovery after SCI ............................................... 62 1 3 Determinants of balance control .................................................................................. 63 1 4 Interaction of the multiple systems contributing to balance control. ....................... 64 1 5 Potential factors influencing walking recovery post SCI .......................................... 65 1 6 Diagram of the inverted pendulum model .................................................................. 66 2 1 Theoretical continuum of walking ability in individuals post SCI. ........................... 70 2 2 Dynamic stability measurement framework. .............................................................. 71 2 3 Rationale and progression of three experiments. ..................................................... 72 3 1 Split belt instrumented treadmill, harness and instrumentati on. ............................. 85 3 2 Spatial foot parameters using feet center of mass (CoM) as body reference points ................................................................................................................................ 86 3 3 Foot placement relative to center of mass (CoM) distances ................................... 86 3 4 Example of center of pressure (CoP), center of mass (CoM), and extrapolated center of mass (XcoM) trajectories during walking with margin of stability (MoS) represented. .......................................................................................... 87 3 5 Step length (a) and step width (b) defined for a given gait cycle ............................ 87 3 6 (A) Manual assisted locomotor training, (B) Robotic assisted locomotor training ............................................................................................................................. 88 4 1 Differences in self selected treadmill speeds with and without customary assistive devices (AD) for participants with SCI. ..................................................... 105 4 2 Example of 3D raw acceleration profiles of head and pelvis trajectories ............ 105 4 3 Acceleration attenuation coefficients for persons with SCI walking with and without assistive devices (ADs) ................................................................................. 106 4 4 Relative mot ion of head and pelvis displacements. ................................................ 107
11 4 5 (A) Mediolateral, (B) anteroposterior, and (C) vertical standardized attenuation coeff icients presented for each participant with SCI. ......................... 108 4 6 Standardized SCI data for relative motion of head and pelvis displacements. .. 109 5 1 Spatial foot parameters using feet center of mass (CoM) as body reference points .............................................................................................................................. 130 5 2 Foot placement relative to center of mass (CoM) distances ................................. 130 5 3 Raw center of pressure (CoP), center of mass (CoM), and extrapolated center of mass (XcoM) trajectories ........................................................................... 131 5 4 Variability of standardized outcomes for individual participants with SCI ........... 132 5 5 Variability of outcome measures by participant group ........................................... 133 6 1 Mean outcome magnitudes standardized to control data pre vs. post locomotor training f or 10 SCI participants. ............................................................... 157 6 2 Variability of outcome measures standardized to control data pre vs. post locomotor training for 10 SCI participants. ............................................................... 158
12 Abstract of Dissertation Presented to the Graduate School of the University of Florida in Partial Fulfillment of the Req uirements for the Degree of Doctor of Philosophy MOTOR CONTROL SOLUTIONS TO DYNAMIC STABILITY DEFICITS DURING WALKING AFTER SPINAL CORD INJURY By Kristin Alayne Vamvas Day May 2010 Chair: Andrea L. Behrman Major: Rehabilitation Science Dynamic stability, defined as the ability to control ones center of mass (CoM) within a moving base of support, is a co requisite of functional walking. Following a spinal cord injury (SCI) dynamic stability is impaired. O ne clinical approach for dynamic stabil ity deficits utilizes assistive devices (ADs) to compensate for functional loss es post SCI However, SCI rehabilitation has recently begun a transition from compensatory therapies toward activity based therapies which target walking recovery To remain co nsist ent with this therapeutic shift measurements should be conducted in this same recovery context Therefore, the purpose of this dissertation was to examine movements that individuals with SCI employ to maintain dynamic stability using a framework that parallels the rehabilitation paradigm shift. M easurements utilized were based on scientific evidence of nervous system priorities to maintain stability. The first experiment investigated head stability during walking with and without devices to understand the effect of ADs, as a conventional evaluation approach, on head stability. Additionally, this study aimed to determine how head stability differed between individuals after injury walking without ADs and healthy persons. This testing condition without d evices remained through subsequent studies. The second
13 experiment examined dynamic stability via foot placement analysis relative to the 1) opposite foot, 2) CoM and 3) CoM plus its velocity. Finally, the third experiment assessed the differential effects of manual assisted and robotic assisted locomotor training on dynamic stability. CoM trajectory length per stride was used as a primary indicator of stability control with trunk motion and spatial foot parameters examined as secondary outcomes. Collective findings across studies indicated that 1) ADs have a role in head stability and when ADs a re removed, individuals post injury exhibit less stability than controls ; 2) persons with SCI demonstrate greater variability in all measures of foot placement comp ared to controls and appear to maintain stability through a continuous pattern of corrective foot placements; and 3) both manual assisted and robotic assistive locomotor interventions have training benefits for dynamic stability, but the trunk and feet str ategies used to maintain CoM control vary between interventions This work is a first step in dynamic stability analysis post SCI and lays the groundwork for further investigations in this field
14 CHAPTER 1 LITERATURE REVIEW Introduction Spinal cord injury (SCI) is among the most devastating and incapacitating acquired medical conditions today. Every year, 12,000 new cases occur in the United States alone resulting in an estimated 255,700 persons chronically disabled by this condition currently ( N ational S pinal C ord I njury S tatistic a l C enter 2008) Although young adult males remain the most frequently injured population, a recent trend shows an increasing number of injured adults over 60 years of age. Furthermore, motor vehicle crashes are t he most common cause of injury, with falls rising as the second most frequent cause, above sports and violence. Despite the substantial amount of physical trauma often sustained by such incidents, the probability of sustaining a complete SCI with total par alysis and loss of sensory function has decreased in recent years secondary to improvements in emergency medical management (Bernhard et al. 2005) As a result, the percentage of persons discharged from the inpatient hospital with incomplete injuries, who demonstrate an emergence of sensory and/or motor function has increased. This clinical transition from complete to incomplete injuries has led to a higher number of individuals with a prognosis to recover functional abilities. Both the extent and location of a spinal cord lesion influence the possible combination of physic al and psychosocial impairments observed after an injury. Depending on the severity of these impairments, individuals post SCI commonly experience reduced functional independence and quality of life as they are l imited in the ability to fully return to their prior lifestyles Physical consequences of SCI may include, but are not limited to, paralysis or paresis, diminished sensation and proprioception,
15 incoordination, and spasticity (Scivoletto et al. 2008) Functionally, these consequences may manifest as difficulties in standing and walking. Reportedly, recovery of walking abilit y is a highly coveted goal relative to recovery of other functions fo r persons who have sustained a SCI regardless of injury level or chronicity (Ditunno et al. 2008) Additional research indicates that regaining walking ability is a first or second priority for approximately 38 percent of individuals with paraplegia and is of higher importance for persons with tetraplegi a who have been injured less than three years (Anderson 2004) Thu s, understanding how walking is generated and controlled as well as developing effective recovery based walking interventions and outcome measures are critical for individuals to achieve their personal goals and to maintain health while avoiding adverse co nsequences of prolonged sitting versus standing or walking (e.g. compromised skin integrity, postural limitations, cardiovascular and respiratory co morbidities, pain, osteoporosis) (Jacobs and Nash 2004) Literature in both humans and animal models post SCI highlights control of dynamic stability, or balance, as an essential component for successful walking recovery (Ladouceur et al. 1997; Deliagina et al. 2008; Scivoletto et al. 2008; Karayannidou et al. 2009) That is, an individual must acquire the capacity to walk upright, load through his/her lower extremities, and c ounteract the internally and externally generated forces created by the dynamics of the a moving body. However, because of the complexities of the task of walking, which involve integration of neural, muscular and biomechanical properties, in combination with the heterogeneity of each SCI, little attention has focused on the movement strategies or mechanisms necessary to achieve balance. Rather, walking balance control studies have investigated healthy
16 individuals (England and Granata 2007) elderly persons (Menz et al. 2003b; Kavanagh et al. 2005a) persons with musculoskeletal impairments (Hof et al. 2007) or individuals with other neurol ogical disorders such as Parkinsons disease, who generally comprise a more homogenous population, relative to persons with SCI, with characteristic movement patterns (Rochester et al. 2004; Oates et al. 2008) Only a few recent studies illustrate balance control in a sample of individuals with SCI evaluating either vestibulospinal integrity (Wydenkeller et al. 2006; Liechti et al. 2008) or automatic postural responses during standing perturbation tasks (Thigpen et al. 2009) Other studies in SCI that address walking evaluate stepping performance in the body weight support and treadmill environment overground with assistive devices (Dobkin et al. 2006; Behrman et al. 2008; Bowden et al. 2008) and/or when adapting to environmental constraints (e.g. inclines) (Leroux et al. 1999; Pepin et al. 2003b) Literature is needed, which directly investigates the mechanisms or movement strategies that persons with SCI employ t o maintain walking balance when required to load through their lower extremities and support themselves without physical assistance. Such investigations would provide insight into the intrinsic capability of the nervous system to perform the specific task of walking and assist in reaching a consensus for how walking balance should be measured. Without an initiation of these investigations, advancement of walking interventions as well as measurement of patient progress and intervention effectiveness will be limited to compensatory strategies, which confine walking performance to external assistive devices. Therefore, the purpose of this dissertation is to elucidate the motor control strategies that persons with incomplete SCI, who possess a heterogeneous ense mble
17 of impairments, implement to overcome deficits in dynamic stability during walking. Various biomechanical measurement techniques will be utilized to understand how individuals who exhibit any ability to self generate at least a few steps post injury i ndividually solve the ubiquitous dilemma of balance control. The literature review will progressively outline research in the fields of locomotion, SCI rehabilitation, and balance control and its measurement to provide a foundation for the t hree studies to follow. More specifically, major sections of the literature review include 1) the neural control of locomotion with emphasis on the subtasks of walking and neurophysiological evidence of spinal pattern generation, 2) traditional compensatory approaches of walking rehabilitation post SCI, evidence of activity dependent plasticity, and the concurrent shift in walking rehabilitation towards recovery based interventions, 3) balance control during standing and walking, its prerequisites, and contributory neural substrates, 4) the effect of SCI on balance control, and 5) current laboratory and clinical measures of balance. Subsequently, an explicit rationale and framework for the three complementary studies will be discussed, followed by methodological procedures and considerations in ins tituting each experiment. Neural Control of Locomotion Subtasks of W alking Human walking is a complex task that requires the neural, muscular, and biomechanical integration of three main subcomponents: 1) reciprocal stepping, including forward propulsion and hip extension at the stance to swing transition, 2) balance control, and 3) adaptability. Forssberg and colleagues first described this combination of fundamental tasks in the animal literature with locomotor studies in spi nalized cats (Forssberg et al. 1980a; Forssberg et al. 1980b) Since that time, both
18 basic and clinical scientists have continued to demonstrate and express th e necessity of these tasks for successful ambulation in animals and humans (Barbeau 2003) The basic reciprocal stepping pattern in walking involves the alternating flexion and extension of the lower limbs. Additionally, this component of walking requires that the trailing limb propel the body forward for a con tinual translation overground. Moreover, hip extension as the lower extremities progresses from the stance to swing phases is a critical sensory cue to elicit this alternating interlimb pattern (Dietz et al. 2002) However, while the lower limbs are constantly moving, the upper body in contrast, requires a high amount of stability to keep the head, which houses essential visual and vestibular balance con trol apparatuses, relatively undisturbed. Therefore, balance control also is necessary to maintain an upright body posture and equilibrium in spite of the challenges that stepping presents. Finally, individuals rarely walk in isolation from the demands of the outside world, and their walking behavior typically is goal directed. As a result, a person needs the ability to adapt successfully to the environment to achieve his/her personal goals (Shumway Cook et al. 2002) In a healthy, uninjured nervous system, these subtasks integrate together to form a seemingly simple and automatic walking pattern. However, in reality, the production of walking requires a high degree of ne ural control to create such a smooth, rhythmic output. Spinal (Central) Pattern Generation The central nervous system (CNS) has an innate capacity to generate a rhythmical, motor output, such as the stepping behavior seen in walking, without afferent or su praspinal input. Interneuronal networks known as central pattern generators are responsible for producing that motor pattern (Grillner and Wallen 1985) Although these neural circuitries have been found throughout the CNS for the
19 production of vital functions such as breathing and mastication, central pattern generators specific to locomotion are located in t he spinal cord and known as spinal pattern generators (SPGs) (Jordan et al. 1992; Marder and Calabrese 1996) Though SPGs can self sustain a stereotypical m otor pattern, the locomotor output also can be modulated and refined through incoming information from the periphery and supraspinal commands to produce a relatively automatic behavior even in the presence of internal and external demands (Dietz 2003; Edgerton et al. 2004) Following injury to the spinal cord, reliance on afferent input to SPGs may greatly increase secondary to disruption of descending neural flow ( Edgerton et al. 2004) (Figure 1 1). Findings from animal preparations of SCI described in the literature have been able to confirm the presence of SPGs (Edgley et al. 1988; Juvin et al. 2007) However, in humans, while many believe only indirect behavioral evidence exists secondary to the invasive procedures required to isolate the SPG circuitry (Illis 1995; Duysens an d Van de Crommert 1998) others contend that direct evidence does exist (Dimitrijevic et al. 2005) Evidence from Animal L iterature Early in the twentieth century, C.S. Sherrington and T.G. Brown published seminal work on the locomotor capabilities of cats, which would later change the way the scientific community viewed the intrinsic potential of the spinal cord and rehabilitation post SCI. Sherringtons initial experiments demonstrated that cats decerebrated at the brain stem level who received direct stimulation of the cross section of the spinal axis exhibited a rhythmic hindlimb flexion extension pattern that was of central origin (Sherrington 1910) Subsequently, Brown altered this animal preparation, showing that decerebrate, deafferented, T12 spinalized cats also were able to produce bilateral, reciprocati ng flexion and extension contractions of the tibialis anterior and
20 gastrocnemius muscles. Based on these findings, Brown proposed a model of a central mechanism consisting of antagonistic centres in the lumbar spinal cord. This model theorized that each center was paired such that when flexor activity was generated in one center, extensor activity was inhibited. The halves allowed for oscillation between hindlimb movements, thus allowing the forward progression in locomotion. Furthermore, Brown regarded p roprioceptive feedback to this central mechanism as highly important for adaptability in the environment, but that its presence was purely regulative, not causative. Despite the lack of intricate detail in this model relative to knowledge possessed today the general concept triggered the production of vast amounts of animal locomotor research to understand SPG circuitry. According to Burke et al., the features that best identify the existence of SPGs are recognizable and reproducible patterns of rhythmi c output in the absence of instructive external drive from other parts of the CNS or from peripheral sensory feedback (Burke et al. 2001) Therefore, some of the strongest evidence of SPG networks in animals use experimental paradigms that induce fictive locomotion. Fictive locomotion involves blocking neuromuscular activity via pharmacological means or motor nerve transection, thus preventing movement related sensory feedback. Following decerebration, spinalization, and/or spinal cord isolation as is noted in cat, rat and lamprey work (Edgley et al. 1988; Juvin et al. 2007; Mentel et al. 2008) sites within the CNS such as the mesencephalic locomotor region or spinal cord are electrically stimulated or perfused with neurotransmitters like dopamine or nialamide (Nielsen et al. 2005; Mentel et al. 2008) Subsequently, efferent neuronal recordings are obtained. The pattern of efferent activity seen has been consistent with the alternating rhythm of extensor or
21 flexor activity visible in intact vertebrate locomotion. Moreover, this activity has been observed in both the forelimbs and hindlimbs in spinal cats and rats to demonstrate the coordinated cou pling of limb movements in quadrupedal locomotion (Yamaguchi 1992; Nielsen et al. 2005; Juvin et al. 2007) One of the simplest vertebrate models of spinal pattern generation that also provides the greatest amount of mechanistic detail is the lamprey model The lamprey has elicited proof of SPGs from the most basic, cellular level. Using the isolated spinal cord in vitro, N methyl D aspartate ( NMDA ) receptor agonist baths have produced a cyclic pattern with a 1% phase lag of ventral root electrical bursts on either side of the spinal cord. This pattern has demonstrated the capability of rapidly activating 100 segments along the lampreys body to induce a highly coordinated swimm ing motion (Wallen and Williams 1984; Grillner et al. 1995) Moreover, mechanoceptors, known as edge cells, have been found to line the lamprey spinal cord and because of their sensitivity to small amounts of stretch can entrain the rhythmic motion and al ter its frequency and timing (McClellan and Sigvardt 1988) The edge cells consist of two types of neurons: excitatory and inhibitory. When the notochord/spinal cord is bent, the cells inhibit contralateral side movement, while exciting ipsilateral movement, and vice versa. Th is activity is thought to propagate movement via several smaller oscillating units along the spinal cord (Grillner et al. 1995) When divided into pieces, the lamprey spinal cord has been shown to produce rhythmic activity in each piece separately, suggestive of pattern generating networks throughout the length of the spinal cord for continual generation of movement (Grillner et al. 1995)
22 In addition to the lamprey, both ki ttens and adult cats have been used as quadrupedal models of locomotion to demonstrate the existence of SPGs. In a two study series, Forssberg et al. sought to determine how well the cat spinal cord could produce stereotypical, rhythmic stepping behavior (Forssberg et al. 1980a) and to what extent the hindlimbs could coordinate with one another (Forssberg et al. 1980b) Kittens spinalized between T10 and T12 initially exhibited alternating hindlimb patterns when the trunk was rotated to one side while lying on the floor. For some kittens, this pattern developed into the ability to reciprocally step while maintaining weight support on the treadmill. Despite deficits in equilibrium, muscle tone and step asymmetries (e.g. limping), distinct intralimb flexor (F) and extensor (E1, E2, and E3) phases were identified in the step cycles with EMG activity resembling that of intact cats (Forssberg et al. 1980a) Furthermore, increases in treadmill speed were shown to pr oduce concomitant increases in hindlimb speeds through shortening of the extensor phases and to a lesser degree the flexor phases Eventually, faster speeds altered the interlimb walking patterns from out of phase to in phase manifested as galloping movements (Forssberg et al. 1980b) In split belt treadmill situations where one belt was driven at a slow speed while the other was driven at a fast speed, the spinalized kittens also could immediately change the duration of the support and swing phases in each limb of the step cycle to stabilize the rhythm and maintain interlimb coordination. If the stepping was purely a product of belt speed, each limb would have continued to step at frequencies consistent with the belt speed. Based on these results of interlimb coordination, Forssberg and colleagues asserted that two separate, but interconnected spinal generators modified by peripheral feedback were accountable for the stepping
23 patterns observed (Fo rssberg et al. 1980b) This theory had previously been outlined based on cat fictive locomotion by Grillner and Zangger and later was reiterated by Howland et al. in T12 spinalized, neonatal kittens that developed a hindlimb stepping pattern on the trea dmill (Grillner and Zangger 1979; Howland et al. 1995) In contrast to both fictive and treadmill locomotion, another mode of detaili ng the presence of SPGs is through airstepping in spinalized animals. Airstepping eliminates limb loading or drive from treadmill motion and avoids the impact of chemicals sometimes used to induce stepping movements. Although initially requiring stimuli su ch as tailpinching to produce basic reflexes and airstepping, cats with complete low thoracic spinal lesions have shown the onset of a couple airstepping cycles immediately upon vertical lifting three weeks post transection. At five weeks, this pattern dev eloped into sustained, spontaneous airstepping with strong, appropriately timed flexor and extensor EMG bursts consistent with both treadmill and fictive locomotion (Giuliani and Smith 1985) The airstepping model has been suggested to demonstrate better the natural beha viors of pattern generating circuitry. In addition to the above animal models, motor behaviors in bipedal chicks have provided evidence for SPGs as well. Bekoff and colleagues (1987, 1989) have examined hatching and walking in post hatchling chicks before and after lumbosacral deafferentation alone, cervical spinal transection alone, and both procedures together. In normal chicks, walking possessed cyclical, alt ernating interlimb coordination; whereas hatching showed an episodic, synchronous coordination that was distinctly different from walking. However, after deafferentation, these motor patterns demonstrated a convergence suggesting that sensory input is the major modifier of the
24 same, or at least components of the same, SPG circuitry responsible for t hese individual behaviors (Bekoff and Sabichi 1987) Moreover, after C3 spinal transection alone, chicks a gain showed separate hatching and walking characteristics that resembled normal patterns (Bekoff et al. 1989). Furthermore, once the high cervical transection was combined with deafferentation, walking exhibited similar alternating patterns to those seen i n only deafferented chicks. These findings suggested that the spinal circuitry did not require supraspinal or afferent inputs to produce the cyclical, rhythmic output. However, authors could not dismiss the possibility that lower cervical or thoracic propr ioceptors may modulate activity of the SPGs in the absence of other ascending or descending inputs. Evidence from Human L iterature As mentioned earlier, fictive locomotion provides potentially the strongest evidence for SPGs in animals. However, because no equivalent of this methodology exists for humans (i.e. anatomically complete lesions with deafferentation), questions still emerge as to whether the spinal circuitry that we detect in humans is the same as that in animals (Illis 1995; Duysens and Van de Crommert 1998; Dietz and Harkema 2004) Regardless, evidence from both training and spinal cord stimulation studies in humans is mounting to s uggest strongly that SPGs are present in humans (Dietz and Harkema 2004; Dimitrijevic et al. 2005) A hallmark study claiming to be the first well defined example of a central rhythm generator for stepping in humans despite incompleteness of SCI was conducted by Calancie et al. (Cal ancie et al. 1994) This case report of an adult male, 17 years post traumatic, incomplete, cervical SCI described the onset of involuntary, rhythmic, alternating lower extremity movements while in a supine position with hips and knees
25 extended. This pa ttern began with the initiation of an intense ambulation training program. Of note were the sensory inputs which increased the frequency of the pattern (e.g. hip/knee/cervical extension, toe dorsiflexion) and those which decreased or ceased the movement (e .g. hip/knee/cervical flexion, toe plantar flexion, involuntary bladder emptying, complete vertical unloading in harness). Additionally, anaesthesia to the right hip which upon x ray was subluxed and sclerotic, attenuated the response. Furthermore, because of the predictability of electromyographic ( EMG ) activity and responses to sensory inputs indicating similarities with fictive locomotion studies in cats, the authors suggested that comparable SPGs were responsible for both. Since the Calancie et al. report, several other studies in humans have concurred that sensory input such as body loading and hip position influence locomotor pattern output via modulation of SPGs (Harkema et al. 1997; Dietz et al. 2002) Perhaps the most convincing, and by some researchers described as direct (Dimitrijevic et al. 2005) evidence for human SPGs involves the epidural electrical stimulation of the dorsal spinal cord in individuals with complete SCI (Dimitrijevic et al. 1998; Minassian et al. 2004; Minassian et al. 2007) Completeness of injury for such studies has been defined based on the Brain Motor Control Assessment as absence of suprasegmental motor unit activation below the level of lesion (Dimitrijevic et al. 1998) Because of the approximation of this definition to an anatomically complete SCI relative to a cli nically complete or discomplete SCI (Sherwood et al. 1992) humans with this type of injury are the best available models to compare to isolated spinal cord animal works. Tonic electrical stimulation of 25 to 60 Hz and 5 9 V via electrodes to the L2 epidural space produced a locomotor like pattern with synchronized EMG activity in
26 these individuals (Dimitrijevic et al. 1998; Minassian et al. 2007) This optimal range of stimulation was based on experimentation of different stimulation parameters. Levels outside this f requency and amplitude range induced tonic extensor activity in the lower limbs. These optimal parameters demonstrated that the interneuronal network in the lumbar spinal cord has the capability of utilizing tonic trains of electrical stimuli to initiate o scillatory output. In turn, the rhythmic motor output generates patterned afferent feedback for the network to continue producing stepping. Support for SPGs in humans and additionally that each limb has independent generators which cross coordinate to prod uce optimal walking patterns under different conditions has been shown in split belt treadmill paradigms, similar to the scenario discussed earlier in spinal kittens (Forssberg et al. 1980b) However, because of the inherent challenges presented by doing such a paradigm in persons with complete SCI, a healthy infant model has been used secondary to the limited cortical influence over stepping (Yang et al. 2005) When split treadmills are driven at different belt speeds or in opposite directions, infants have exhibited the ability to adopt a coordinated walking pattern. Each limb is thought to have an autonomous SPG because of the ability to modify stance phase durations and move legs in opposite directions concurrently, but SPGs on either side are believed to communicate in order to ens ure only one limb is in swing at any given moment. These findings and conclusions are in agreement with literature in spinal cats and fictive locomotion in cats suggestive of SPG presence (Grillner and Zangger 1979; Forssberg et al. 1980b)
27 Spinal Cord Injury Rehabilitation Traditional Rehabilitation of W alking Until recently with the emergence of evidence supporting properties of SPGs in humans, rehabilitation of walking for individuals post SCI reflected the hierarchical model of motor control (Shumway Cook and Woollacott 2001) This theoretical model purported that the CNS was hardwired, irreparable, and nonmalleable. Moreover, the spinal cord simply was nothing more than a conduit for information from the brain to the rest of the body. If an injury occurred to the spinal cord, the relay of information was suspended. Rehabilitation strategies for SCI emphasized compensation for the physical losses presumed to never return function, consistent with assumptions of the hierarchical model. In many cases, particularly those in which patients exhibited little to no voluntary muscle activity, relearning standing or walking was deemed impossible. Additionally, the common sympt oms of spasticity and clonus were considered negative consequences of SCI, further limiting the probability of walking (Beres Jones et al. 2003) Thus, therapists avoided any treatment regarding standing or walking and solely taught wheelchair propulsion, transfers, and other functional mobility skills. If standing or walking ever were attempted, therapists instructed patients in the use of assistive devices, suppor tive bracing, and alternative biomechanical strategies such as utilizing momentum to initiate movement or creating a wide base of support with the legs to increase stability. These strategies would assist the individuals in resuming some level of mobility using remaining physical strengths and external aids in light of the presumed irreparable spinal cord and inability to convey information from the brain to the muscles (Atrice 2005; Behrman and Harkema 2007)
28 Activity D ependent P lasticity Over recent years, evidence of activity dependent plasticity in animals and humans has suggested a strong potential f or the resumption of walking ability after SCI if the training conditions are optimized, thus providing the basis for rehabilitation strategies alternative to the traditional approach. Activity dependent plasticity can be defined as persistent changes with in the CNS that result from prior experiences and influence future motor behaviors (Wolpaw and Tennissen 2001) In essence, the CNS is exhibiting motor learning as a function of experience. Although activity dependent plasticity traditionally is deemed to occur supraspinally (Kleim and Jones 2008) a great deal of animal and human literature also supports plasticity in the spinal cord due to peripheral inputs and/or descending influe nces from the brain (Barbeau and Rossignol 1987; Behrman et al. 2008) Repetition of specific types of sensory input as well as variability in activity intensity and challenging function through increasing postural and locomotor demands are essential to modify spinal cord neuronal structure and synapses and to induce the desired functional recover y after SCI (Barbeau 2003) Evidence from Animal L iterature Research in adult cats an d kittens with SCI have illustrated some of the most compelling evidence of locomotor recoveries consistent with activity dependent plasticity. Through the use of locomotor training in the treadmill environment with appropriate sensory inputs provided by s peed of the treadmill, bodyweight loading, and manual cues, spinalized cats have shown improved hindlimb stepping abilities over varying times periods post transection (Lovely et al. 1986; Barbeau and Rossignol 1987; Hodgson et al. 1994) This is in contrast to work that has demonstrated poor,
29 uncoordinated stepping performance and difficulty maintaining weight support after two months training in midthoracic spinalized cats (Eidelberg et al. 1980) Although plasticity is considered to be greater in younger, developing nervou s systems as in the kitten (Smith et al. 1982) Barbeau and Rossignol demonstrated that appropriate training strategies can produce stepping recovery in adult cats as well (Barbeau and Rossignol 1987) Cats spinalized between T10 and T12 were trained in the treadmill environment two to three times per week. With use of a thoracic jacket and/or manual assistance at the tail, bodyweight support was gradually increased and treadmill speeds were varied according to the cats progress. Over periods of several weeks up to one year of training, deficits in balance and volitional movement remained. However, cats remarkably exhibited the ability to maintain hindquarter weight support and to generate rhythmic, coordinated stepping with overall similar joint excursions and EMG activity compared to both spinal kittens and intact cats. As shown in Barbeau and Ross ignols work (1987), providing correct afferent input can result in improved stepping performance. Additionally, one criterion associated with sensory input for activity dependent plasticity to occur is task specificity (Barbeau and Fung 2001) That is, learning of a particular task such as walking is observed when that ta sk is specifically trained. However, walking training may not translate to the performance of other tasks such as standing, presumably because the sensory experience is different between the two tasks. Kang and Dingwell (2006) presented evidence consistent with this notion, identifying that the mechanisms controlling walking and standing balance are inherently different based on local dynamic stability analyses (Kang and Dingwell 2006) Therefore, rather than training standing tasks as a method
30 for i mproving walking, the repetitive afferent input from walking itself is thought to entrain that particular task into the CNS via neuroplastic changes. For example, the abilities of T12 13 spinalized cats randomized to either a non trained, treadmill trained, or stand trained group one month post transection have been explored (Hodgson et al. 1994) Following two to three months of training, all standtrained cats could maintain standing for an extended time period with very little stimulation at the tail; however, they were unable to step with the exception of a few cats taking uncoordinated steps at extremely slow speeds under 0.2 m/s. In contrast, the steptrained cats required maximal stimulation at t heir tails to maintain standing but could step at an average of 0.62 m/s. Followup experiments initiated the stand or step training just one week after transection, continued for six to eight months, then crossed cats to the other training regimen. After the cross over, the cats learned to perform the newest task in which they were trained, but they deteriorated in the ability to perform the or iginal task; that is, if they originally learned to step before switching to stand training, then later they had difficulty stepping, but could stand well (Hodgson et al. 1994) Furthermore, activity dependent plasticity has been shown through comparisons of spontaneous recovery and recovery with locomotor training (Lovely et al. 1986; de Leon et al. 1998) In a sample of spinalized cats like those described by Hodgson et al. (1994), Lovely et al. randomly assigned cats to either a non trained group or a treadmill trained group. Five to seven months later, cats in the trained group exhibited a steeper rate of improvement in speed and achieved a significantly greater maximal treadmill speed with full weight bearing steps compared to the non trained group (Lo vely et al. 1986) Subsequently, de Leon et al. found comparable results in a similar study, but
31 cats started training after one week and continued for three months. Kinematic and EMG characteristics were quantified and compared between groups and to ac tivity before spinalization (de Leon et al. 1998) Several changes occurred consistent with greater stepping and speedrelated improvements in the trained cats. For example, increased terminal hip extension in stance, higher vertical ankle displacement during swing, greater forward placem ent of the paw at initial touchdown, and increased tibialis anterior and iliopsoas EMG amplitudes all were noted. In contrast to studies in the cat with complete SCI, evidence to support spinal cord plasticity with repetitive training has been demonstrated in the incomplete SCI chick model also (Muir and Steeves 1995) Since walking and swimming produce similar rhythmic, alternating movements, but differ in the amount of phasic sensory feed back available, recovery of these behaviors were evaluated in hatchling chicks following thoracic hemisections. After two weeks of both locomotor training on a runway and swim training, walking recovered to normal, but swimming motion remained poor relativ e to normal values. However, when phasic cutaneous input from a buoy was provided to the chicks as needed during limb extension (in the absence of a loading stimulus as in walking), swimmi ng improved after five days; yet removal of the stimulus caused dete rioration of the movement. By two weeks, however, even without the cutaneous input, swimming returned to normal (Muir and Steeves 1995) Therefore, task specificity, by way of repeated sen sory inputs, which are timed appropriately within the task, is thought to promote activity dependent plasticity and locomotor recovery after SCI
32 Finally, activity has illustrated the capacity to induce changes at the molecular level in animals as well (Vaynman and Gomez Pinilla 2005) Exercise in rodents with hemisectioned spinal cords has revealed the upregulation of certain neurotrophins as activity increases. Specifically, brainderived neurotrophic factor (BDNF) has shown elevated levels on the same side as the lesioned lumbar spinal cord. Additionally, activity has promoted expression of BDNF products in the CNS, such as syn apsin I, involved in synaptic trans mission and neurotransmitter release, as well as cyclic adenosine monophosphate ( AMP ) response element binding protein (CREB), involved in gene transcription. These findings have suggested that the spinal cord has the potential to learn via rehabilitation post SCI since these neural factors are essential components of synaptic plasticity for memory and learning (Vaynman and Gomez Pinilla 2005) Evidence from Human Liter ature Based on knowledge gained from animal literature, human studies in SCI evolved using the training fundamentals that revealed positive functional changes in animals. Specifically, the concept of activity dependent plasticity stimulated the implementat ion of activity based therapies in humans (Behrman and Harkema 2007) Activity based therapies utilize the principle of task specificity to enhance the neurophysiological changes that promote funct ional gains. The goal is to retrain the nervous system below the level of the lesion and take advantage of any spared neural pathways that may interact with specialized circuitries above and below the lesion. Evidence from the literature in humans has inc luded both individuals with complete and incomplete SCI, cervical, thoracic and lumbar levels, adults and children. Participants have engaged in various locomotor training strategies from manual -
33 assisted to robotic assisted training, strengthening exercise s to functional electrical stimulation (Behrman and Harkema 2000; FieldFote et al. 2005; Wirz et al. 2005; Gregor y et al. 2007) However, to remain consistent with the definition of activity dependent plasticity, only those interventions task specific to walking are described here. In persons with acute and chronic incomplete SCI of numerous etiologies and ability levels, four to twenty weeks of manual locomotor training were conducted. This training resulted in the majority of initially wheelchair bound persons becoming able to walk without physical assistance. Furthermore, persons who were able to walk before training improved gait speed and endurance. Overall, these changes were maintained at followup, six months to six a nd half years later (Wernig et al. 1999) Such persistent changes in function indirectly corroborate activity dependent plasticity in the CNS given the retention of training effects. Similar findings have been described after manual locomotor training with trans lation to overground walking in persons with incomplete SCI (Behrman and Harkema 2000; Behrman et al. 2005) An adult who was non ambulatory and initially had minimal voluntary activation in th e lower extremities progressed to ambulating full time with a straight cane and greatly improved volitional leg strength. Other individuals who were ambulatory with assistive devices prior to training advanced to walking with less restrictive devices and a t faster gait speeds. Although these persons were all within a year of injury onset and spontaneous recovery may have influenced the outcomes, certain features of the training responses suggested that neural plasticity based on the task specific rehabilitation strategy was at least partly responsible. For example, the
34 individual who had little voluntary leg movement prior to training demonstrated the ability to reciprocally step on the treadmill before any improvements in voluntary muscle control were obser ved. For other individuals who were able to walk before enrolling in training, improvements were seen in step symmetry and kinematics on the treadmill that did not immediately translate to the overground situation (Behrman and Harkema 2000; Behrman et al. 2005) Similar to the adult with incomplete SCI who gained both strength and walking abilities despite initially having little leg activity, a non ambulatory child with a chronic, severe, incomplete SCI also recovered walking function. However, after developing the ability to step and walk full time with a walker following 76 manual locomotor training sessions, this child demonstrated no changes in voluntary muscle activity during that time (lower extremity motor score remained 4/50) (Behrman et al. 2008) Moreover, although stepping was present, equilibrium reactions were absent alluding to spinal pattern generated activity with minimal descending control for balance strategies. Additionally, the rhythmical, sustained stepping pattern was absent in a supine position, presumably because of the removal of afferent input due to loading while upright (similar rationale to the previously mentioned chick model of Mu ir & Steeves, 1995). This last finding, in addition to the evolution of walking with minimal, volitional lower extremity movement, is highly suggestive of activity dependent plasticity in this childs CNS as well as supraspinal, descending activation. As discussed in spinalized cats, persons with clinically complete SCI also have exhibited the capacity step in the treadmill environment, albeit without transfer to the overground (Dietz et al. 1994; Harkema et al. 1997; Behrman and Harkema 2000)
35 Reciprocal stepping has been achieved using the same sensory inputs as those employed for persons with incomplete injuries (e. g. progressively increasing load, accentuating hip extension and upright trunk, triggering tibialis anterior activation for swing, altering speeds). The reduction in amount of assistance required, from full manual contact throughout the gait cycle to assis tance for foot placement only or even a few independent steps, has been suggestive of plasticity in the neuromuscular system below the level of the spinal cord lesion. Moreover, EMG patterns over the course of a manual locomotor training program in patient s with complete paraplegia demonstrate appropriate timing of the lower extremity flexors and extensors with increasing gastrocnemius and decreasing tibialis anterior amplitudes (Dietz et al. 1994; Dietz et al. 1995) The rate of change in gastrocnemius activity directly related to changes in loading during the program as well. Dietz et al. contend that these changes are not only indicative of a lumbosacral spinal circuitry, but also that the circuitry is capable of learning in response to sensory input such as loading (Dietz et al. 1997) In contrast to stepping ability, preliminary data in both individuals with clinically complete and incomplete SCI report an improved ability to stand after stand and step training (Harkema 2001) Despite apparent lack of supraspinal input, person s with complete injuries gradually assumed standing with less bodyweight over time. One person was reported to have stood with only 10% bodyweight support ( BWS ) for 45 seconds. In comparison, those individuals with incomplete injuries have demonstrated ind ependent standing (i.e. without BWS) for several minutes post training. Paradigm Shift toward R ecovery Although the conventional approach to SCI rehabilitation remains widely used presently, both evidence of activity dependent plasticity and SPGs are changing the
36 way scientists and rehabilitation professionals view the capabilities of the CNS and rehabilitation post SCI. A paradigm shift is emerging that transitions our rehabilitation mindset from one of compensation toward one of recovery (Behrman et al. 2006) Moreover, the principles of motor learning, which emphasize task specificity and repetition of activity consistent with the task to be relearned, encompass this shift toward recovery of function. Specific sensory inputs influence the plasticity of both the brain and spinal cord (Kleim and Jones 2008) Thus, if recovery of walking is desired, the task of walking must be performed with appropriate sensory cues to the CNS. These cues limit the use of compensatory tools like assistive devices or bracing. Rather, recovery of walking promotes loading through the lower extremities, minimizing load through the upper extremities, optimizing kinematics and kinetics (e.g. trunk position, foot trajectory, hip position), and walking at a speed consistent with normal walking (Behrman and Harkema 2007) Barbeau et al. (1999) presented a model of functional walking recovery consistent with this paradigm shift in rehabilitation, which could aid in directing clinicians towards appropriate evaluation and treatment strategies for patients after SCI. Figure 1 2 depicts a progression of walking recovery as a function of cha nges in control and capacity Control indicates a persons potential to alter four variables during walking : 1) generation/absorption of energy at specific points of the gait cycle, 2) trajectory of the foot during swing phase, 3) support of ones own body weight, and 4) balance of the upper body. Capacity refers to the ability to maximize those variables as needed in order to meet environmental demands or personal goals. Image 1 of Figure 1 2 presents an individual in a permissive environment of BWS and a treadmill, which
37 allows that person to accomplish and optimize the necessary subtasks of walking. This in dividual possesses both low control and capacity for functional walking as he/she is unable to perform any component of walking overground, even with assistance of a walker, cane or crutches. Images 2 and 3 demonstrate varying degrees of increases in capac ity and control as the individuals transition to an overground environment; however, they are only able to perform walking within the constraints of assistive devices. In Image 2, the person with a walker may be able to take a longer or higher step, for ex ample, representing greater capacity of that variable, but at the expense of upper body control and lower extremity loading. That is, this individual could be compensating and increasing weight bearing through the upper extremities to give the lower extrem ity the biomechanical advantage of creating a maximal change in foot trajectory. In contrast, the individual with a cane in I mage 3 demonstrates greater control of the head, arms and trunk as well as weight bearing through the legs, although it may be at t he expense of achieving a maximal step length or height to st ep over an obstacle. Finally, I mage 4 shows an overall increase in both control and capacity as a person demonstrates the ability to perform all the requisite subtasks of walking and modulate them appropriately for adaptability in the real world (e.g. no device, uneven terrain, stairs, etc.). During rehabilitation, the BWS and treadmill environment may allow for thorough evaluation and optimal training of the control and capacity of variables esse ntial to walking recovery. Since the emphasis of such an environment is on functional recovery rather than compensation, an individuals true ability can be discerned in the absence of assistive devices and thus maximized. Furthermore, the shift in rehabil itation focus
38 towards recovery of walking behavior may enable individuals to experience and learn more appropriate motor control strategies for balance and adaptability without external devices that provide upper body support and minimize body loading. Det erminants of Balance C ontrol Human balance control is innately challenging, particularly in comparison to quadrupedal balance control requirements. Q uadrupeds are naturally stabl e during standing and walking with a horizontal trunk position, low center of mass, and a broad base of support (BoS) created by three to four limbs in contact with the ground at all times In contrast, the human body is an intrinsically unsteady ensemble. This instability results from a large mass consisting of the head, arms and t runk being placed in an upright position approximately twothirds of body height above the ground. Furthermore, the only support structures are one or two feet on the ground forming a small BoS (Winter 1995) When humans perform tasks more demanding than s tatic standing, such as reaching, turning, walking, or running these functional skills continually create forces that challenge the structural organization of the body. Therefore, this multisegmental system requires balance control mechanisms that are capable of responding to the demands imposed by movement dynamics (Figure 13) Maintaining balance concerns the integration of two essential control components: postural control and equilibrium control (Massion & Woollacott 1996) Posture refers to the orientation of body segments relative to the direction of gr avity. Thus postural control requires the body to combat Earths gravitational pull to prevent collapsing and to remain upright. Complementary to postural control, equilibrium control is the stab ilization needed to counteract the linear and angular accelerations attempting to unbalance various segments of the body. S elf initiated movements such as walking or
39 external perturbations such as tripping on an obstacle create these destabilizing accelerations. To avoid falling, the magnitude of equilibrium control will depend upon movement speed and the inertial mass of the moving segment That is, the faster the movement or the heavier the moving segment, the large r accelerations will be. Consequently, g reater equilibrium control will be required to stabilize the interconnected segments against such forces ( Massion & Woollacott 1996) Ultimately, both postural and equilibrium control are achieved by two primary means: proactive and reactive mechanisms (Patla and Prentice 1995) Proactive balance mechanisms utilize both visual cues and prediction to prevent falls during voluntary movements. Initially, vision allows a person to scan the environment from a distance and evaluate impending challenges. Then, predictive mechanisms incorporate knowledge of prior experiences and calculations of potential forces that are likely to act on the body in order to select, plan and execute anticipatory postural adjustments (Huxham et al. 2001; MacLellan and Patla 2006; Misiaszek 2006) These feedforward responses set postural muscles in preparation for the expected destabili zation induced by a voluntary task (Cordo and Nashner 1982) However, when proactive mechanisms under or overestimate postural preparations or alternatively when perturbations are unexpected, the CNS responds to sensory feedback and employs reactive balance mechanisms. Unlike proactive mechanisms that have time to utilize higher executive processing, initial reactive mechanisms must occur in such a rapid manner that cortical control is not possible (Tang et al. 1998; MacLellan and Patla 2006; Misiaszek 2006) In those cases, the primary feedback comes from somatosensory and ves tibular sources
40 since the response times of those systems are much greater than vision (Woollacott and Tang 1997) In addition to the balance control mechanisms intrinsic to the individual, the task performed and the environment in which the task occurs both affect the strategies utilized in balance control (Figure 1 3) (Huxham et al. 2001) Specifically, the task and environment may differentially dictate the biomechanic al configurations of the body as well as the degree of information processing required to accomplish balance control successfully. For example, as a movement becomes more intricate (e.g. quiet standing to running), changes occur in the relationship between body segments and the accelerations that need to be controlled. Similarly, walking on a level linoleum floor will adjust the bodys kinematics and kinetics differently and to a lesser extent than negotiating gravel terrain or avoiding moving obstacles. Fu rthermore, the complexity and familiarity of the task and environment both influence the amount of information processing (e.g. attention, prediction) that is necessary. Shumway Cook and Woollacott (2001) have presented a systems approach to balance contr ol that illustrates this interaction of the individual, task and environment via a convergence of musculoskeletal and neural compone nts (Figure 14 ). More specifically, musculoskeletal components include the relationships among the various body segments in cluding range of motion, flexibility and strength. In contrast, neural components encompass sensory and motor strategies, sensory systems, internal representations of the body (body schema) and higher executive influences on motor responses through antic ipatory or adaptive strategies.
41 Neural Systems Underlying Balance C ontrol Due to the intricate interrelationships between neural and muscular systems to coordinate motor output for functional balance, identifying the underlying neural substrates responsibl e for responses of balance control is difficult. In both animals and humans, suggestions of these substrates have been deduced from behavioral outcome studies primarily. Literature using animal models has analyzed postural and equilibrium control in lampre ys, rabbits, and cats, to name a few (Deliagina et al. 2008) Moreover, research in humans traditionally has examined patient populations who have disorders related to various neural structures in order to understand how the postural responses of individuals with an abnormally functioning structure differ from respon ses of healthy individuals (Morton and Bastian 2004; Jacobs and Horak 2007a) Based on this kind of work, the brain stem, cerebellum, basal ganglia, cerebral cortex, spinal cord, and several descending neural pathways all have been suggested as critical for balance control; however, a consensus on their definitive contributions remains to be determined. Therefore, the following provides an overview of the possible roles of these neural substrates on different aspects of static and dynamic balance control. Brain Stem Many of the implications that the brain stem plays a critical role in balance control have evolved from literature of animal models. S pecific brainstem sites in the cat model, such as the subthalamic and mesencephalic locomotor regions as well as the ventral and dorsal tegmental fields, show functions that integrate posture and walking when stimulated (Mori et al. 1992) Additionally, evidence has demonstrated that chronic, T6 spinalized cats lack the capacity to produce the necessary automatic postural responses to perturbations while standing, even though they exhibited hindlimb weight
42 support (Macpherson and Fung 1999) The maintained weight support was thought to result from muscular stiffness and tonic extensor activity elicited from spinal reflex mechanisms. However, the brainstem (and cerebellum) was implicated as a site required for balance control because it receives and integrates visual, vestibular and somatosensory input from the periphery (Horak and Macpherson 1996) Similarly, Torres Oviedo et al. extracted a set of muscular synergy patterns from both quiet stance and automat ic postural responses in cats exposed to platform perturbations (Torres Oviedo et al. 2006) The brainstem and cerebellum again were suggested as integration sites for task specific sensory information which ultimately simplified and organized the mo tor output into functional synergy patterns. Furthermore, the brainstem is thought to contain already established synergy patterns which are selected and modified based on incoming afferent input about the context of a situation (Jacobs and Horak 2007b) Cerebellum In addition to the above proposed roles of the cerebellum in conjunction with the brainstem, severa l other theories have been presented regarding its function. Early work by Nashner and Grimm found delayed and unorganized muscle responses to standing surface perturbations in patients with cerebellar disorders (1978). Based on these findings, the coordin ation of muscle synergies in postural responses was hypothesized as one major role of the cerebellum. However, more recent investigations have posited that the cerebellum is involved in predictive balance control primarily (Bastian 2006) ; that is, using prior experiences to shape motor output. Evidence from individuals with cerebellar damage supported this hypothesis, showing hypermetric postural responses to both unexpected and expected platform perturbations. Moreover, this patient group also demonstrated the inability to scale response magnitudes to
43 expected perturbations, regardless of having previous exposures (Horak and Diener 1994) Other studies also have noted deficits in scaling magnitude of automatic postural responses; however, they indicated more specifically that the cerebellum had greater involvement in gain control of response magnitudes to changes in conditions (Timmann and Horak 1997; Mummel et al. 1998) Basal Ganglia Research aiming to discover the basal ganglias role in balance control has largely utilized patients with Parkinsons disease (PD) as models. Chong et al. gained support for their hypothesis that the basal ganglia is essential for the ability to quickly adapt postural set to a sudden change in task context (Chong et al. 2000) Although reactive muscle response onsets and synergy pattern organization were normal in persons with PD multiple trials were required to switch postural set and select a different synergy pattern. In contrast, delayed muscle onsets and relative timing as well as decreased muscular amplitudes were shown in a voluntary task of rising on toes (Frank et al. 2000) Nevertheless, the scaling amplitude of pos tural responses to changes in velocity and excursion of movement was appropriately produced. These findings from the Frank et al. study and similar results from additional studies led to the conclusion that the basal ganglia contribute to the ability to in itiate and generate quick, sufficient force to control center of mass movement in both voluntary and externally cued perturbation tasks (Horak et al. 1996; Burleigh Jacobs et al. 1997) Furthermore, functional imaging studies in animals and humans also have supported the role of the basal ganglia in activating necessary postural muscles while inhibiting unnecessar y antagonists when preparing for and initiating responses to centrally driven and externally triggered movements (Brooks 2001) For example, when basal ganglia dysfunction is present, an
44 overall increase in background postural EMG (co contraction or inability to inhibit antagonistic muscles) has been observed during quiet standing, manifesting itself as muscular rigidity (Horak et al. 1996) Cerebral Cortex Involvement of the cerebral cortex in balance control remains a debatable issue, particularly for reactive mechanisms. Some researchers have cited that postural response latencies are too short to gain cortical input, thereby deeming subcortical mechanisms responsible (Diener et al. 1984) However, reactive postural responses may actually last longer than the initial, automatic response mediated by brainstem or spinal mechanisms. As a result, the later phases involving feedback loops to correct for errors, likely require cortical influence (Jacobs and Horak 2007a) Later phases include compensatory strategies, such as stepping to form a wider base of support or reaching for external support, beyond the potentially ineffective ankle or hip strategies produced by short or med ium latency responses. Moreover, cerebral cortex contributions have been observed via electroencephalography potentials in studies of anticipatory responses to expected perturbations and voluntary movement as well (Slobounov et al. 2005; Jacobs and Horak 2007a; Jacobs et al. 2008) Spinal Cord and Descending P athways In addition to the initial reactive myotatic stretch reflex induced by standing perturbations, evidence asserts the ability of descending pathways to modify se gmental spinal reflexes. Specifically, vestibulospinal and reticulospinal neuron recordings have shown increased responses in cats walking on an inclined treadmill angle (Matsuyama and Drew 2000) Reticulospinal drive also has revealed increased antigravity extensor tone during locomotor activities, likely due to the exc itation of the mesencephalic
45 locomotor region (Mori 1987) Furthermore, enhanced corticospinal pathway activity h as been detected and well correlated with cats balance corrections to standing platform tilts (Beloozerova et al. 2005) Addition ally, studies in both adult humans, infants, and spinal cats have implicated spinal networks and reflexive pathways for the quick, stumblingcorrective reaction that occurs in response to stimuli contacting specific extremity locations ( e.g. dorsum of paw/ foot) and during certain phases of the gait cycle (Forssberg et al. 1975; Forssberg 1979; Zehr et al. 1998; Lam et al. 2003) These pathways, as well as the interaction of pathways linking the above mentioned major neural structures, like the basal gangliabrainstem system, have been suggested as participants in the regulation of balance control (Drew et al. 2004; Takakusaki et al. 2004) Balance Control in SCI For individuals post SCI who regain any degree of walking ability, the challenge remains to identify which components in the triad of walking subtasks (stepping, balance, and/or adaptability) continue to limit their return to complete independence. Althoug h difficult to determine secondary to the multitude of factors comprising each subtask, Brotherton et al. noted that one of the top three perceived factors associated with the high falls incidence in this population was loss of balance (Brotherton et al. 2007a) Anderson (2004) further suggested that balance is a limiting factor for this population by indicating that recovery of trunk stability was an important quality of life factor for persons with paraplegia, particularly those less than three years since injury. As mentioned previously, human walking requires one to balance the large trunk mass above a continually moving base of support (Winter 1990). Therefore, trunk instability would create yet an additional challenge to accomplish the task of walking successfully.
46 Moreover, another study showed a significant correlation between balance and walking performance in SCI (Scivoletto et al. 2008) These authors recommended that balance become a highly emphasized component of walking reh abilitation in SCI, commenting that the SCI literature only discussed balance with respect to trunk activity in the wheelchair. A more recent search of the literature revealed that a few studies do exist which examine standing balance control in this popul ation (Wydenkeller et al. 2006; Liechti et al. 2008; Thigpen et al. 2009) Although walking balance control in SCI awaits investigation, studies provide evidence of sensorimotor complications that may contribute to balance dysfunction during walking and further inhibit recovery of this task. Th e systems approach to balance control by Shumway Cook and Woollacott ( 2001) presented earlier (Figure 14) supports the possible impact of these complications on walking balance control, inclusive of the proactive and reactive strategies necessary for part icipation in the any real world environment. Additionally, Figure 15 models a bi directional interaction of example sensory, motor, and postural deficits including balance following SCI, further illustrating the potential influence and interference of the se problems on walking recovery (Barbeau et al. 1999) As Barbeau et al. (1999) suggest, the various complications from SCI are not mutually exclusive and have the potential to impact the functioning of other biological systems. Norea u et al. (2000) presented the prevalence of secondary impairments resulting from SCI and highlighted the multiple body systems affected by these injuries that could disrupt function and quality of life. After surveying almost 500 individuals post SCI of va rying age, gender and injury severity, this investigation confirmed that in addition to nervous system involvement, the cardiovascular, respiratory,
47 musculoskeletal, intestinal, cutaneous/integumentary, and urinary/renal systems also were compromised (Noreau et al. 2000) Although some of these systems appear removed from the control of walking, many conceivably could disrupt this task and its subtask of balance, either directly or indirectly. On a musculoskeletal level, primary complaints follow ing SCI are atrophy and weakness in muscle groups that have diminished neural innervations secondary to the level of injury (Jayaraman et al. 2006; Shah et al. 2006) These deficits often include decr ements in trunk muscle activation (e.g. paraspinals and/or abdominals), which are key muscles in trunk stabilization during unsupported functional tasks (Borghuis et al 2008; Bjerkefors et al. 2009) Residual, isolated muscle strength using the AIS evaluation after acute injury reportedly strongly predicts motor recovery (Waters et al. 1994) although more recent evidence argues that changes in lower extremity motor scores on the American Spinal Injury Association Impairment Sc ale (AIS) do not always accompany changes in walking function, acutely or chronically (Wirz et al. 2005; Wirz et al. 2006; Behrman et al. 2008) Other skeletal muscle properties deemed necessary for skilled movement, such as the ability to alter timing and power of motor output, have been implicated as limiting factors in walking recovery. Specifically, persons with SCI demonstrat e drastically reduced rates of torque production as well as voluntary peak torque in affected muscles such as the plantar flexors and quadriceps (Jayaraman et al. 2006; G regory et al. 2007) In healthy individuals, energy generated from plantar flexor torque and power is transferred up the lower extremity to provide trunk support and forward progression during walking (Zajac et al. 2003) When disrupted, persons with
48 SCI may require alternative movement strategies to generate an upright walking pattern while fully loaded through their lower extremities. Thigpen et al. (2009) additionally showed that indi viduals with mild to moderate sensorimotor impairments (e.g. deficits in light touch, pin prick sensation, lower extremity strength, and proprioception) secondary to SCI are able to adapt automatic postural responses appropriately to expected and unexpected changes in standing perturbation conditions. However, muscle onset times are delayed and activation magnitudes are decreased compared to control subjects (Thigpen et al. 2009) Furthermore, persons with motor incomplete SCI display cocontraction of antagonistic lower extremity muscle groups (tibialis anterior/soleus, quadriceps/hamstrings) during treadmill walking with BWS (Gorassini et al. 2009) Some of these individuals also develop three to four times greater activation in proximal leg muscles compared to healthy indiv iduals; presumably, this abnormal increase may be an effort to maintain postural stability in the absence of adequate strength in other muscles. In the presence of lower extremity weakness, spasticity is speculated clinically to be a partial compensatory m echanism for loss of supraspinal drive by providing extensor support during stance post SCI (Dietz 2002) Yet, Scivoletto et al. (2008) detected a strong negative relationship between spasticity and walking performance, promoting spasticity as one of the best negative predictors of walking ability. Reportedly, spasticity has potential to fragment walking motion, thus diminishing its smoothness. Spasticity develops as a seque la of disinh ibition and hyperexcitation of monosynaptic stretch reflexes as well as loss of long latency reflexes resulting from lack of feedback control (Dietz 2002). Proprioceptive feedback from joint movement, muscle
49 stretch, skin pressure or noxious stimuli provides information about the location of the body in space and contributes to the regulation of reflex activity. Often after SCI, proprioception is diminished or absent. When modulation of these reflex loops is reduced secondary to proprioceptive deficits, wal king control is impaired. As a result, the nervous system relies more heavily on the vestibular system. However, evidence of compromised vestibulospinal tract integrity in samples of individuals post motor incomplete SCI has emerged recently (Wydenkeller et al. 2006; Liechti et al. 2008) While undergoing galvanic vestibular stimulation during different standing sensory conditions, perso ns with SCI demonstrate longer muscle and center of pressure movement onset latencies suggesting the vestibular system has difficulty substituting for sensory impairments and maintaining balance in the same manner a healthy individuals vestibular system d oes. Multiple impairments secondary to SCI exist which may be contributing to balance dysfunction and the aforementioned high falls incidence in this population. These impairments create functional deficits that necessitate reliance on assistive devices f or stability and mobility and further limit maneuverability in a variety of real world contexts (e.g. up and down stairs, inclines, uneven terrain) (Pepin et al. 2003a; Leroux et al. 2006; Musselman and Yang 2007) Inst ability post SCI also imposes altered movement strategies where the trunk and arms may have to compensate for impairments in the legs (Melis et al. 1999) Such biomechanical compensations may further limit the physiological activity tolerance required for functional adaptations like altering walking speeds or walking for long distances in the community (Waters and Mulroy 1999; Jacobs and Mahoney 2002) Additionally, individuals post SCI have reported several
50 secondary injuries due to loss of balance and falls, including fractures, joint disloca tion, muscle/ligament strain or sprain, and loss of consciousness (Brotherton et al. 2007a) which could further postpone and limit walking recovery. Krause (2004) asserted that those individuals with less severe SCIs, classified as AIS D are more likely to sustain subsequent injuries. P erhaps this is because they demonstrate some recovery of function and ability to ambulat e, but may have difficulty with the task (Krause 2004) .Thus, the study of balance control in the spinal cord injured population is essential to minimize risk of falls and maximize recovery potential for ones ultimate return to functional community walking and participation. Outcome Measures of Walking Balance Current Laboratory Based Measures of B alance C ontrol Balance control has been measured objectively in several ways utilizing different modes of technology in the laboratory setting. The reason for the dev elopment of various measurement strategies is that studies aim to investigate diverse components of a persons ability to balance. Because of the abstract and vast nature of balance as well as the infinite number of everyday tasks available to explore, r esearchers often consider individual variables that contribute to balance as a whole rather than objectively measuring the construct of balance. For example, a researcher may be seeking to understand possible neural mechanisms involved in the production of balance strategies while walking or rather to investigate the ability of muscles to coordinate their timing appropriately in order to achieve those strategies. Alternatively, a studys goal might be to determine which sensory systems are relied upon more or less for maintaining upright standing balance in specific patient populations. Therefore, measurement selections differ depending on whether one is investigating the neural,
51 biomechanical, or muscular mechan isms underlying balance control, and furthermo re which type of task (standing versus walking) is of interest. A review of common objective balance measurements are provided here with the disclaimer and understanding that researchers frequently alter or combine information from basic measurements to fit their particular scientific inquiries. As a result, a large number of measurement variations exist. One central, well established concept in balance control that has pervaded the literature is based on the inverted pendulum model (Figure 16) (Hof 2008) This model assumes the body is a single mass located at the center of mass (CoM), supported by a leg that is in contact with the ground at the center of pressure (CoP). The whole body CoM is often considered the bal ance point of the body and is the weighted average of all the body segments individual centers of mass. In contrast, CoP is the weighted average of all the contact areas of the feet with the ground. When the CoM falls to one side of the CoP, the CoP must shift to prevent falling. Therefore, based on this model, a persons CoM must remain within the CoP or base of support during standing or must transition along the trajectory of the swinging limb towards a new base of support in walking in order to remain stable. This relationship between the CoM and CoP has become the underlying theme in the development of many laboratory based balance measurement tools. Both the static and dynamic relationships of the whole body CoM to CoP have been used to examine balanc e in standing and walking tasks (MacKinnon and Winter 1993; Pai and Patton 1997; Hof et al. 2005) Biomechanical models from motion analysis data can be created to determine the CoM position and velocity. Subsequently,
52 CoM data may be compared to the CoP obtained from ground reaction forces on force plates. A static relationship which evaluates the horizontal distance between the vertical projection of the CoM location and the CoP remains a frequent measurement tool (Martin et al. 2002; Chou et al. 2004; Hughey and Fung 2005) As the distance between CoM and CoP becomes smaller, balance becomes compromised. However, given the dynamic nature of tasks such as walking where time and movement are important components, several researchers have recognized the contribution of CoM velocity in this relationship. M easur e s such as the margin of stability (aka. dynamic stability margin) have been proposed to account for velocitys influence (Pai and Patton 1997; Hof et al. 2005; Hof 2008) In addition to CoP and CoM relationships, l ocation and variability in CoP patterns alone (Goldie et al. 1989; Maki and McIlroy 1996; Doyle et al. 2007) are measures commonly used in quiet and perturbed standing assessments. Single or dual force platforms are capable of providing CoP information averaged under both feet or for each foot individually. Within testing protocols, individuals are often instructed to stand statically with eyes opened or cl osed, on a compliant surface, and/or in single limb or tandem stance (Jonsson et al. 2004) The primary goal of these experimental ma nipulations is to induce sensory conflict, stressing the visual, vestibular, or somatosensory systems separately or in combination. The direction and amplitude of CoP change (i.e. postural sway) that results from ankle strategies typically, but also hip, stepping, and suspensory strategies, is measured. In a similar manner, CoP patterns have been recognized as a means to assess postural sway secondary to unilateral galvanic vestibular stimulation techniques as well (Latt et al. 2003) This information
53 provides details regarding how the vestibular system interacts with the other sensory systems involved in postural control. As a result, insight into the neural mechanisms of balance can be ob tained. In contrast to CoM and CoP quantities, electromyography (EMG) of the limb and paraspinal muscles is utilized to detect whether or not appropriate muscles are activated during tasks and the timing of their activation. Specifically, muscular onset ti mes and response latencies have been used extensively to determine the initiation of reactive balance strategies after external perturbations are imposed to an individual while quietly standing or walking (Nashner 1980; Horak and Nashner 1986; Bakker et al. 2006) A great deal of the balance literature has adopted or adapted the classic perturbation paradigm developed by Nashner, which introduces unexpected platform rotations and translations (Nashner 1976) Muscle response times after the initiation of platform movement can provide information about the neuromuscular, coordinative mechanisms of a persons response; that is, determining if a delay in response is present or if additional muscle groups are compensating for deficits in normally activated muscles. Sim ilarly, muscle onset times can also offer insight into anticipatory postural strategies to internal perturbations in which muscles are activated in preparation for a task like initiating gait or reaching with the arms (Timmann and Horak 2001; Mochizuki et al. 2004) In such tasks, visual input as well as prior experience has instructed the central nervous system to set postu ral muscles in advance of a dynamic task to avoid an uncontrolled CoM displacement. Additionally, EMG has been used to develop another measure for balance control Applying non negative matrix factorization techniques to EMG data, the specific set of
54 syner gies responsible for a balance response to a perturbation task during standing can be identified. These synergies are interpreted as the way in which the nervous system simplifies the many degrees of freedom available in human movement and coordinates the motor output into stereotypical response patterns (Ting and Macpherson 2005) Alternatively, kinematic changes primarily in the head, trunk, pelvis, hip, knee, or ankle angles have been measured via motion analysis to indicate normal and abnormal balance patterns during standing and walking (Krebs et al. 1992; Leroux et al. 2002; Nadeau et al. 2003; Grabiner et al. 2008) Angles of trunk or CoM inclination and angular dispersions or mediolateral linear excursions of head and trunk segments are examples of objective measures utilized (Ha hn and Chou 2003; Nadeau et al. 2003; Lee and Chou 2006) Other temporal spatial parameters derived from kinematics including step width (interfoot distance) and step velocity have likewise been considered indicators of balance control during gait (Timmann and Horak 2001; Krebs et al. 2002) Unlike the above measures, nonlinear measurements using kinematics have been devised over recent years for dynamic tasks specifically (Dingwell and Cusumano 2000; England and Granata 2007; Segal et al. 2008) These measurements, such as the maximum finite time Lyapunov exponent, are calculated based on the rate of kinematic variability at various joints during a movement. These tools provide an indicator of local dynamic stability as a result of internal disturbances such as neuromuscular errors within the body. Final ly, acceleration patterns of the upper body from the head to the pelvis have been measured via accelerometers or double differentiation of motion analysis position data to illustrate the way the postural control system coordinates and attenuates the
55 oscill ations imposed by the dynamics of movement. Segmental coupling, peak amplitude, frequency, regularity and smoothness of the patterns are all outcomes that have been described (Menz et al. 2003a; Kavanagh et al. 2004; Kavanagh et al. 2005a; Kavanagh et al. 2006) Given the influence of major sensory systems on balance control as well as studies in healthy individuals suggesting that the head is well stabilized during various locomotor tasks (Pozzo et al. 1990; Nadeau et al. 2003) acceleration may provide information indirectly about the ability of the body to gain visual stabilization and accurate vestibular inputs at various gait speeds and over uneven terrain. Limitation s of Laboratory Based M easures Measurements of balance gained within the laboratory setting provide extensive knowledge and generate further inquiries into the complexities of the balance control system. Nevertheless, limitations exist in these measures. First, the amount of time required to collect, process, and analyze data is substantially greater than the time allotted to conduct a therapeutic examination in clinical settings. Second, the equipment needed to obtain these measures is costly. When a major goal of creatin g laboratory measures is to eventually translate findings to the clinic for use with patient populations, time and expense must be factors considered (Allum and Carpenter 2005) Additionally, the currently developed measures necessitate integration with one another to achieve insight into the cause and effect of balance maintenance and loss through a host of contributing variables. Optimally, only a few measures would be able to offer that depth of information.
56 Current C linical Measures of Balance C ontrol Unlike the highly technical measures acquired in laboratory settings, clinical measures of balance typical ly use little equipment and can be conducted and interpreted in a short period of time. With the exception of more specialized balance clinics, which utilize Computerized Dynamic Posturography systems such as the Balance Master or Equitest, the objective clinical measures most frequently used in the clinic are standardized assessments based on observational analysis. Numerous measures have been developed over the years with several redundant balance items across tests (e.g. inclusion of the Functional Reach in the Berg Balance Scale). As a result, only the most commonly used assessments in the clinic currently with balance as the principal construct are presented here: Berg Balance Scale, Dynamic Gait Index, Tinetti Performance Oriented Mobility Assessment and the Clinical Test of Sensory Integration and Balance (Huxham et al. 2001) Additionally, the most recently developed and still evolving balance tool, the Balance Evaluation Systems Test, which aims to be a comprehensive evaluation of balance will be described (Horak et al. 2009) The Berg Balance Scale (BBS) consists of 14 items, each scored on a scale of 0 to 4 (Berg et al. 1989) Scores are based on the amount of assistance required to complete the tasks as well as ability to maintain positions for specified time periods or ability to perform a task within a time constraint. All items on the scale prohibit the use of any assistive devices. Tasks range from static sitting and tra nsferring over level surfaces to a functional reach test, turning in a circle, and picking up an object from the floor. The BBS has been shown to have excellent interrater ( intraclass correlation coefficient ( ICC ) =.97) and intrarater (ICC=.98) reliability in individuals with acute stroke
57 and the elderly (Berg et al. 1995) as well as high test retest (ICC=.96) and interrater reliability (ICC=.96) in persons with multiple sclerosis (Cattaneo et al. 2007) In contrast to the BBS, the Dynamic Gait Index (DGI) assesses an indi viduals ability to perform eight gait related tasks with his/her usual assistive device(s) as needed (Shumway Cook and Woollacott 2001) Each item on the DGI is scored from 0 to 3, ranging from severe impairment to normal. Differences in scores are based on use of an assistive device, physical assistance needed, imbalance, gait deviations and/or time required to complete a task. Examples of tasks on this assessment include changing gait speeds, stepping over an obstacle, gait with vertical head turns, and stairs. In persons with chronic stroke, the DGI exhibits high test retest and interrater reliability (ICC=.96, .96 respectively) and concurrent construct validity with the BBS (r=.83) (Jonsdottir and Cattaneo 2007) High reliability has also been found for persons with vestibular dysfunction (kappa=.95) (Wrisley et al. 2003) and multiple sclerosis (ICC=.98) (McConvey and Bennett 2005) Next, the Tinetti Performance Oriented Mobility Assessment (POMA), originally designed to predict falls in an institutionalized population, has two subscales: a gait assessment (G POMA) a nd a balance assessment (B POMA) (Tinetti 1986; Tinetti et al. 1986) G POMA instructs an individual to walk at a preferred pace with an ass istive device as needed, while nine qualities of performance are evaluated as 1 (normal) or 2 (abnormal/compensatory). Performance qualities observed include initiation of gait, step length, path deviation, and trunk stability. In contrast, B POMA consists of 13 tasks that are scored as 1 (normal), 2 (adaptive response), and 3 (abnormal). Although tasks are similar to those on the BB S, a reactive balance task in which a nudge is applied to the
58 sternum while standing is also assessed. Both subscales have demonstrated good interrater (R=.80 to .93) and test retest reliability (R=.72 .86) in a sample of elderly participants (Faber et al. 2006) Although not measured along an ordinal scale as with the above measures, the Clinical Test of Sensory Integration and Balance (CTSIB) developed by Shumway Cook and Horak is frequently administered in the clinic (Shumway Cook and Horak 1986) The CTSIB tests a persons ability to maintain static standing for 30 seconds while compensating for altered sensory inputs in six different conditions. The conditions are 1) eyes open while standing on a firm surface, 2) eyes closed on a firm surface 3) eyes open with a visual conflict dome on a firm surface, 4) eyes open on a compliant surface, 5) eyes closed on a compliant surface, and 6) visual conflict on a compliant surface. Maintenance of normal balance, increased postural sway, or loss of balance in each condition indicates which sensory system (visual, somatosensory or vestibular) a person relies upon primarily or more so than other systems. Finally, the Balance Evaluation Systems Test (BESTest) was developed recently in an attempt to identify the underlying systems contributing to balance control deficits and emphasize the importance of the systems approach to motor control evaluation (Figure 1 4) (Horak et al. 2009) Unlike oth er tools, which were created for determining balance dysfunction in older adults in particular, this assessment aims to more broadly apply to a variety of patient populations in order to approach rehabilitation most effe ctively. The BESTest assesses six balance control systems (biomechanical constraints, stability limits/verticality, anticipatory postural adjustments, postural responses, sensory orientation, and stability in gait) across 36 test items. Some items
59 are newly created if unavailable on current tests and some items are borrowed from already published standardized tests (e.g. BBS, DGI, CTSIB) Although still in development, this tool thus far demonstrates excellent interrater reliability overall (ICC=.91). Limitations of Clinical M easures Therapists select balance measures with the intention of revealing a patients risk of falls and areas of balance deficits and of guiding a plan for treatment. Additionally, these measures are used to assess progression over time and recovery after injury, following surgery, or from a disease process. Recovery as described here is defined as Levin et al. (2009) proposed: restoring the ability to perform a movement in the same manner as it was performed before injury. However, the current measures of bal ance present limitations for therapists in accurat ely obtaining this information. Although most falls occur during walking tasks (Berg et al. 1997) a limited number of measures assess balance during walking. Moreover, the assessments that have b een developed allow an individual to perform tasks with the use of an assistive device such as walkers, canes or crutches (e.g. DGI, BESTest: Stability of Gait). Given the literature emphasizing critical differences between recovery of walking function ver sus functional compensation (e.g. with assistive devices) (Behrman and Harkema 2007) and research showing altered muscle activation or balance strategies with upper extremity support (Visintin and Barbeau 1994) or even light surface contact (Jeka and Lackner 1994) our current assessments for walking balance appear to place emphasis on compensatory balance only. Obtaining information regarding compensatory walking balance may be misguiding therapists as to th e true underlying balance deficits in their
60 patients, thus preventing implementation of appropriate treatment strategies for progression toward recovery of function. Furthermore, the balance measures that do emphasize recovery by eliminating use of assist i ve devices (e.g. BBS) neglect assessment of walking and primarily examine sitting, standing, and transitional movements like transfers. However, with the principle of task specificity for motor learning infused throughout the literature (Lovely et al. 1986; Barbeau 2003) if information regarding walking balance is needed, then walking should be the task assessed. Nevertheless, current measures like the BBS still are used to globally assess fall risk and balance for all tasks secondary to strong psychometric properties and lack of other appropriate measures (Patterson et al. 2007; van de Port et al. 2008) Finally, balance is comprised of many sub components of con trol via proactive and reactive mechanisms. Yet, current clinical bala nce measures primarily test proactive strategies and minimally assess reactive, which occur quite frequently in everyday life (Huxham et al. 2001) Although individuals often have to mo tor plan in advance how they are going to approach certain obstacles, they also need to respond suddenly to unexpected perturbations on a regular basis. Laboratory investigations are capable of examining factors such as these to provide a more in depth understanding of the construct of balance for eventual translation to the clinic.
61 Figure 11. Modulation of central/spinal pattern generator (CPG) circuitry in (A) healthy and (B) complete spinal cord injured nervous systems through afferent inputs, descending neural pathways, and/or therapeutic interventions. Adapted from Edgerton VR, Tillakaratne NJ, Bigbee AJ, de Leon RD, Roy RR (2004) Plasticity of the spinal neural circuitry after injury. Annu Rev Neurosci 27: 145167.
62 Figure 1 2. Progression of functional walking recovery after SCI Adapted from Barbeau H, Ladouceur M, Norman KE, Pepin A, Leroux A (1999) Walking after spinal cord injury: evaluation, treatment, and functional recovery. Arch Phys Med Rehabil 80: 225235.
63 F igure 13. Determinants of balance control. Adapted from Huxham FE, Goldie PA, Patla AE (2001) Theoretical considerations in balance assessment. Aust J Physiother 47: 89100.
64 Figure 14. Interaction of the multiple systems contributing to balance co ntrol. Adapted from Shumway Cook A, Woollacott M (2001) Motor Control: Theory and Practical Applications. Lippincott Williams and Wilkins, Philadelphia.
65 Figure 15. Potential factors influencing walking recovery post SCI Adapted from Barbeau H, Ladou ceur M, Norman KE, Pepin A, Leroux A (1999) Walking after spinal cord injury: evaluation, treatment, and functional recovery. Arch Phys Med Rehabil 80: 225235.
66 Figure 16. Diagram of the inverted pendulum model. mg=mass*gravity, l=length, CoM=center of mass, CoM(x)=vertical projection of center of mass onto ground, CoP(ux)=center of pressure, Fy=vertical ground reaction force at CoP Adapted from Hof AL (2008) The 'extrapolated center of mass' concept suggests a simple control of balance in walking. Hum Mov Sci 27: 112125.
67 CHAPTER 2 CLINICAL RELEVANCE AND RATIONALE Of the continuum of individuals who sustain a SCI, approximately 52 percent are diagnosed with incomplete injuries at the time of hospital discharge with a greater potential to r egain walking ability than those individuals with complete injuries ( National Spinal Cord Injury Statistical Center 2008) A smaller subset of persons with motor incomplete injuries regains some degree of walking function with or without assistance of an a ssistive device and/or brace. Of those who demonstrate any walking ability, an even smaller, unknown percentage is capable of walking without a device or bodyweight support (BWS) and avoiding falling simultaneously (Figure 2 1). This cohort exhibits a leve l of neural compensation and functional balance recovery (Levin et al. 2009) post injury that others are not able to achieve. Even more importantly, through their functional recovery, we may begin to understand the way they accomplish this task and mechanisms underlying such recovery. Ultimately, this knowledge will allow research ers and clinicians the opportunity to determine how to progress others with SCI towards functional balance and community ambulation. As mentioned in previous sections, current measures of balance possess limitations. Specifically, when clinicians wish to e valuate balance control for the task of walking post SCI, standardized tests that use assistive devices such as the Dynamic Gait Index, mask a persons actual capacity to maintain their posture and equilibrium secondary to compensation through the upper ex tremities. In addition to measures of balance, the primary clinical prognostic indicator for overall walking potential, which globally includes the ability to balance as a prerequisite, is the A merican Spinal Injury Association (ASIA) Impairment Scale (AI S) (Waters et al. 1994) Through an evaluation
68 of voluntary, isolated, limb strength and sensory integrity, persons who regain some to all isolated movement have injuries classified into three main categories: AIS C through E. However, given the anatomical, physiological, and functional heterogeneity p resent in persons with SCI, the ability to maintain balance during walking varies greatly within each category. Thus, clinicians still await adequate tools to detect the underlying balance control that reflects taskspecific functional recovery rather than compensation. In order to develop appropriate clinical tools, laboratory investigations are critical to understand the construct of balance and all its relevant neural, muscular and biomechanical components. This will enable the determination of clinic al correlates to laboratory based measures such that efficient and true assessments of balance can occur in the clinical environment. With the intention of measuring task specific functional recovery, the studies in the following chapters were developed wi thin the framework in Figure 22. This framework illustrates a linear progression through the paradigm shift occurring in neurological rehabilitation today: one that started as a compensatory approach and is now shifting toward a recovery based approac h. The first study begins with an examination of dynamic stability in the context of traditional rehabilitation (i.e. with assistive devices). In the second study, that layer of compensation with external devices is removed to evaluate stability during a f unctionally uncompensated walking experience. Finally, the third study remains within the paradigm shift to investigate the effects of two recovery based locomotor training interventions on dynamic stability. A dynamic stability measurement model also is presented in Figure 22 to demonstrate various avenues in which to evaluate the framework above. These
69 measurement strategies elucidate the interplay of body segment dynamics along the entire body axis from the head to the interaction of the feet with the ground. The rationale for this topdown, bottom up model of dynamic stability is to illustrate that the human body is series of interconnected elements that all possess a role in balance control. According to the systems approach to balance control, thes e elements unite on a multiple system level when considering the intersegmental biomechanics as well as the many sensory inputs, their integration, and motor outputs along the neural axis (Hora k et al. 2009) However, the studies presented here represent the first time a biomechanical investigation has evaluated several elements along the entire body in each person as well as conducted assessments during walking conditions without assistance of any type in individuals who regularly walk with assistive devices. The specific measurement tools selected capture what are purported to be critical components or priorities for maintaining balance while walking. Motor strategies observed in persons wit h SCI across these studies may be able to provide insights into potentially affected balance control systems for examination in future research as well as the impact of recovery based locomotor interventions on dynamic stability of individuals with particu lar motor strategies. One motor strategy (e.g. head stabilization, foot placement to capture the moving center of mass) may compensate for insufficiencies in other typically normal strategies or locomotor training may influence the development of certain motor strategies (e.g. control of center of mass, trunk movement, spatial parameters). Details regarding specific measurement tools mentioned in the developed model are described in Chapter 3, Methodology: P rocedures and Considerations.
70 Figure 21. Theoretical continuum of walking ability in individuals post SCI.
71 Figure 22 Dynamic stability measurement framework
72 Figure 23. Rationale and progression of three experiments.
73 CHAPTER 3 METHODOLOGY: PROCEDU RES AND CONSIDERATIONS Participants Ten i ndividuals with chronic, incomplete SCI and ten healthy persons comprising a control group participated in each study Those with SCI were involved in a larger randomized clinical trial (RCT) and healthy controls were included as part of a larger, ongoing cross sectional study By including the same individuals across studies findings from each could be assimilated to evaluate dynamic stability collectively in each person with SCI according to various measurement tools and within the overall framework presented. As part of the RCT, p articipants with SCI were enrolled based on the following inclusion criteria: 1) at least 18 years of age, 2) injury sustained at least 6 months prior to the study, 3) upper motor neuron, motor incomplete spinal cord lesion, 4) ability to ambulate at least 10m with or without an assistive device, and 5 ) injury of traumatic or non traumatic origin, excluding those of congenital etiology. From those enrolled in the RCT, a subsample was included in the three studies presented here, based on ones ability to generate at least 3 steps without an assistive device, bodyweight support or physical assistance. Healthy adult controls also were 18 years of age or older; however, they ambulated full time without assistive devices or physical assistance. All experimental procedures were conducted at the Brain Rehabilitation Research Center, Malcom G. Randall Veteran Affairs Medical Center in Gainesville, Florida. Each participant signed a written informed c onsent approved by both the VA Subcommittee for Clinical Investigation and the University of Florida Health Science Center Institutional Review Board.
74 Clinical Assessmen t s A licensed physical therapist assess ed bila teral upper and lower extremity motor and sensory function in persons with SCI based on the American Spinal Injury Association (ASIA) International Standards for Neurological and Functional Classification of Spinal Cord Injury (American Spinal Injury Association 2002) This assessment establish ed SCI severity and categorized injuries according to the ASIA Impairment Scale (AIS). A physical therapist also assessed performance on standardized balance assessment s, th e Berg Balance Scale (BBS) (Berg et al. 1992) and the Dynamic Gait Index (DGI) (Shumway Cook and Woollacott 2001) to provide further clinical descriptions of each individual s balance ability. Biomechanical Data Collection Procedures For each study, the following paragraphs detail the overall data collection process. This process generated all data required for the entire dissertation. Therefore, not every component described here is included in each study. Both participants with SCI and healthy controls had reflective marker balls and rigid body clusters positioned on specified body landmarks to acquire threedimensional ( 3D ) motion data. Marker positions we re based on the Vicon PlugInGait marker set (modif ied Helen Hayes set) All i ndividuals wore a safety harness attached to an ov erhead cable and track system that was suspended from the laboratory ceiling. Walking trials lasted a maximum of 30 seconds and were performed over a split belt instrumented t readmill (Tec machine, Inc.) (Figure 31) W alking practice was permitted to becom e accustomed to walking on the treadmill and to obtain the best possible steady state walking speed. When comfortable, data collections commence d Individuals with SCI performed two trials: the first with their customary assistive device,
75 the second without any device. Although participants wore a safety harness should they stumble or fall, neither BWS nor manual assistance was p rovided during any trial. All individuals were instructed to walk to the b est of their abilities at self selected (SS) treadmill speeds during both trials Trials conducted without assistive devices were intended to reveal the underlying capability of each individual to walk without external assistance and the way each persons nervous system solved the problem of balance contr ol. Either sitting o r standing rest periods were provided as needed between bouts of activity. H ealthy control participants performed four separate walking trials Three trials involved randomly introduced speeds at 0.3, 0.6, and 0.9 m/s. T hese slow speeds were selected a priori to represent a likely range within which persons with SCI would elect to walk. During t he fourth trial healthy individuals walked at their SS treadmill speeds Although community walking occurs overground rather than over a treadmi ll and debate continues about the similarities and differences in biomechanics and muscle activity between the two environments, particularly in neurologically impaired populations (Harris Love et al. 2001; Parvataneni et al. 2009) measurement of walking in the treadmill environment was necessary for several reasons. First, given the continuum of functional abilities across persons with SCI selected for these studies, some were able to generate only a limited number of steps without an assistive device. Walking overground without a device and without a safety harness would have created an unsafe environment in which to learn about balance strategies in this subset of individuals. Second, since Experiment 2 required 3D ground r eaction forces under each foot separately for outcome me asurement walking needed to occur over forceplates.
76 However, walkways that require subjects to place their feet on two or three individual forceplates, also require several passes across the walkway to capture enough footfalls and to ensure that subjects landed the whole foot on each forceplate. Many individuals with SCI would be unable to complete multiple passes, primarily due to fatigue and balance deficits. In contrast to a walkway an instrumente d split belt treadmill allows one to capture kinematic and kinetic data from a continuous number of steps in one walking trial. The treadmill can record every footfall without concern that the entire foot was captured. Additionally, if persons with SCI wer e limited to walking on a treadmill, then healthy controls were required to as well to maintain a consistent envi ronment and equal comparison between both groups. Finally, the treadmill also allowed for a controlled comparison of walking speeds, one of the few variables capable of being controlled in such a heterogeneous population. D ata Acquisition and Processing Twelve camera passive motion analysis (Vicon Motion Systems) and 3D grou nd reaction forces (GRFs) from four piezoelectric force transducers (Adv anced Medical Technology, Inc.) located beneath each half of the treadmill were acquired continuously during walking trials. The split be lt treadmill system allowed collection of GRFs for each stance phase over multiple steps of the gait cycle. Raw kinemat ic data w ere collected at 100 Hz, then low pass filt ered using a fourthorder, zerolag Butterworth filter with a 6 Hz cut off frequ ency. GRFs were acquired at a sampling rate of 2000 Hz, and low pass filtered using a fourth order, zero lag Butterworth fil ter wi th a 20 Hz cut off frequency. A 13segment musculoskeletal model was created using Visual 3D (V3D) processing that fit the model to marker trajectories. Anthropometric and inertial values defined within V3D w ere applied for segment modeling and segme ntal center of mass (CoM)
77 calculations. V3D models were used to conduct inverse dynamics analyses for calculation of intersegmental joint kinetics. Custom Matlab programming (Mathworks, Inc.) was developed to calculate the outcome measures E xcellent abso lute agreement has been reported between Vicon motion analysis systems and GaitRite instrumented walkways for measurement of spatiotemporal parameters (ICC= 0.92 0.99) (Webster et al. 2005) In addition center of mass es timati ons calculated via the Vicon system and a Crossbow accelerometry system demonstrated excellent concurrent validity (r = 0.87) (Lee et al. 2007) Furthermore, the split belt instrumented treadmill system utilized in these three studies has been validated as an instrument for gait analysis as well (Tesio and Rota 2008) Dynamic Stability Outcome Measures and Rationale Experiment #1 Head and pelvic motion. A literature base is mounting which evaluates dynamic stability based on the ability of the normally functioning human body to reduce perturbations to the head, which houses critical sensory apparatuses for balance. This function allows for enhanced gaze stabilization and integration of accurate vestibular inputs for balance control while moving. Recently, common measurement techniques have examined the degree of head acceleration attenuation relative to the pelvis (Menz et al. 2003a; Kavanagh et al. 2004; Mazza et al. 2008) Using this literature as a foundation, both displacement and acceleration patterns of the upper body were evaluated. Segmental CoM positions were used to calculate 3D displacements of head and pelvis CoMs were determined for each step in a trial. Subs equently, position data were double differentiated to acquire 3D linear accelerations for the same steps. Visual inspection of acceleration profiles located abnormally high and low spikes, which were
78 interpreted as non physiological noise. Therefore, maximal and minimal thresholds were manually selected on the profiles via customized Matlab coding, and steps with accelerations exceeding the establish ed thresholds were eliminated. Accelerations. Root mean squares (RMS) of frameby frame acceleration val ues for both the head and pelvis CoMs in each direction were calculated from valid steps. RMS values then were used to calculate attenuation coefficients (AC) for the entire trial as described by Mazza et al. ( 2008; 2009) AC= (1 (RMSH/RMSP))* 100, where H = head, P = pelvis Attenuation coefficients, expressed as percentages, rep resent the ability of the individual to reduce accelerations from the pelvis to the head. A positive, larger coefficient indicates greater head stability relative to the pelvis. In contrast, negative coefficients represent larger head accelerations (less s tability) relative to the pelvis. Displacements. The means and standard deviations of 3D head and pelvis CoM displacements were calculated. Holt et al. (1999) quantified vertical (VT) head displacements in the sagittal plane using a single head marker as an indicator of head stability. W e expanded our methodology to include the anteroposterior (AP) and mediolateral (ML) directions in order to paral lel our 3D acceleration outcomes and additionally selected the segmental head CoM as a more robust measurement of head motion than a single marker Using similar methodology to Hong and Earhart (2010) for the relative motion between two segments, the difference between pelvis and head mean displacements (mean difference, meanp elvismeanh ead) was an outcome indicating which segment produced greater displacements over the walking trial. The difference between pelvis and head displacement standard deviations (SD) (variability
79 difference, SDp elvisSDh ead) was an outcome indicating whether the head or pelvis produced more variable disp lacements over the course of a walking trial. Similar to the acceleration attenuation coefficients, positive values for difference calculations represent less head motion relative to the pelvis, while negative values represent greater head motion relative to the pelvis. Experiment #2 Spatial foot parameters and f oot placement relative to pelvis center of mass. Since walking creates a situation in which the CoM is outside the base of support for 80 percent of the gait cycle (Winter 1995), the CoM is continually falling forward with each step. Thus, awareness of where a foot lands at initial contact provides insight into how the CoM is restabilized with each step, and consequently, the way in which a person maintains balance over a series of steps. The place ment of one foot relative to another du ring double support (e.g. step width, step length) is commonly documented in the clinic using observational analyses (Krebs et al. 2002) Yet that method may or may not provide insight into the control of the CoM as the balance point of the body. Additional measures of foot placement indicating the location of the foot relative to the falling body might provide another level of information (Redfern and Schumann 1994; Pai and Patton 1997; Balasubramanian et al. 2010) Overall four measures were defined as the following distances at each initial contact: 1) leading foot CoM to trailing foot CoM in the ML direction (Step width) and in the AP direction (Step length) (Figure 3 2) and 2) leading foot CoM to pelvis CoM in the ML direction (ML foot placement) and in the AP direction (AP foot placement) ( Figure 3 3 ) Margin of stability (MoS). MoS, a measure developed by Hof et al. (2005), estimates dynamic stability using a method derived from the conventional comparison of
80 CoM position to center of pressure (CoP). In contrast to the conventional comparis on, Hof and colleagues accounted for the influence of velocity on the CoM position (2005) As such, MoS compares the shortest perpendicular, mediolate ral distance between the CoP and the extrapolated center of mass (XcoM) during single limb stance of gait (Figure 3 4 ). The XcoM is the vertical projection of the CoM in the direction of its velocity. 1. XcoM = | (x + ( = CoM 2. MoS = | umax XcoM |, where umax = boundary of CoP A larger MoS is consistent with increased dynamic stability as a greater margin exists between the XcoM and the CoP Furthermore, the size of the margin is proportional to the impulse that would be required to destabilize or unbalance a person. That is, by adding a critical amount of CoM velocity (i.e. impulse) in the direction of the nearest CoP, an unstable situation can be created. Experiment #3 Center of mass (CoM) trajectory. Critical biomechanical control elements are responsible for preventing collapse of the body as it produces a continuous, smooth forward progression during walking. Research of dynamic stability during walking suggests that the endproduct o f these control elements is motion of the bodys CoM. During walking, the upper and lower body must complement one another and shift in a coordinated effort to recapture the CoM within the CoP during each step since the CoM lies outside the CoP f or most of the gait cycle (Winter 1995 ). Therefore, trajectory of the CoM indirectly exemplifies the compensatory strategies of all other body segments to avoid falling.
81 For each stride in a walking trial, the length of the whole body CoM trajectory normal ized to strid e length was calculated in the transverse plane (Figure 3 4 ). Stride length was defined as the anteroposterior distance from the initial contact of one foot to the next initial contact of that same foot (heel marker to heel marker) (Figure 3 5 ) (Perry 1992) Length of the CoM trajectory may illustrate whether an individual progresses normally in a smooth sinusoidal nature or whether the trajectory deviates in alternate directions in an attempt to control the CoM during a given step. The latter situation would create a longer CoM trajectory A lternatively, a shorter than normal CoM trajectory may also be possible should other biomechanical characteris tics such as step width or perhaps neuromuscular properties such as hypertonicity or muscle tightness, limit CoM motion. Trunk excursions. Literatur e investigating falls in the elderly population has reported that a primary factor, which distinguishes older adults who fall from those who do not, is the ability to limit trunk movements after a perturbation during walking (Grabiner et al. 2008) Additionally, w hen the trunk is in a flexed position other body segments suc h as th e pelvis and lower extremities, compensate by alter ing their alignment to restore control of the CoM (Saha et al. 2008) Therefore, t runk excursions were used as an adjunctive measure of balance to complement the CoM trajectory length in Experiment 3 Thre e dimensional trunk angular displacement range s were determined for each step in a walking trial. Trunk displacements were defined as the angles of rotation (degrees) about each axis. The established reference axes were positioned orthogonal to one another with the x axis directed mediolaterally, the y axis anteroposteriorly, and the z axis vertically. Proximal and distal ends of the modeled
82 trunk segment were used to calculate movement about the x and y axes, while markers positioned on bilateral acromion processes were used to calculate rotation about the z a xis. Spatial foot parameters. As mentioned previously in Experiment 2, foot location relative to the contralateral foot is a common clinical measurement. The ease of describing spatial parameters (wide base of support, variable step lengths) (Krebs et al. 2002) or even collecting spatial measurements using a lowcost instrumented walkway makes these parameters particularly clinic friendly (Brach et al. 2001) Unlike the spatial foot parameters de scrib ed earlier in Experiment 2, which utilize foot CoM locations and require complex motion analysis equipment, step width and step length also were measured using body landmarks more applicable to the way in which these parameters could be gathered in the clinic environment. Step width was defined as the mediolateral distance between the heel marker of the leading foot at initial contact and the heel marker of the contralateral, trailing limb at that same point in time. Step l ength was defined as the anteroposterior distance between the heel marker of the leading foot at initial contact and the heel marker of the contralateral, trailing limb at that same point in time (Figure 3 5 ) (Perry 1992) Locomotor Training Intervention The locomotor training (LT) intervent ion for individuals with SCI in Experiment 3 occurred in either a manual or robotic environment, both activity based therapies (Figure 3 6 ) (Dromerick et al. 2006) While training in the manual environment, a harness suspended participants with varying amounts of BWS over a moving treadmill as human trainers physically facilitated walking biomechanics. Three to four trainers place d their hands at the pelvis/trunk for rotation, shifting, and upright posture, at
83 bilateral lower extremities for optimal stepping kinematics wit h appropriately timed gait events, and at bilateral upper extremities to promote arm swing, as needed (Behrman and Harkema 2000) Although training in the robotic environment (Lokomat, Hocoma) also occu r red with a harness, BWS, and a treadmill, th is training scenario immobilized the pelvis and legs in an exoskeleton and utilized a computerized system to drive stepping in lieu of human trainers. The robotic environment possess ed an additional feature called guidance for ce, which systematically reduced the percentage of stepping assist ance provided. This feature intended to mimic the progressive decrease of human trainer assistance in manual training as participants exhibit ed greater independence (Colombo et al. 2000; Reinkensmeyer et al. 2004; Galvez et al. 2007) Participants involved in Experiment 3 were randomized to one of these two environments as part of a larger randomiz ed clinical trial. Training occurred three to five days per week for 45 sessions total At least one licensed physical therapist was present always during training. The treatment goal was for participants to achieve 30 minutes of quality stepping per session; however, the number and length of individual bouts necessary to obtain this goal varied daily depending on each participants functional ability and/or trainer fatigue. Over the 45 sessions, participants were challenged continually as they demonst rated greater independence while exhibiting appropriate stepping patterns, including an upright trunk. Specifically, BWS decreased, treadmill speed increased and guidance force or trainer manual assistance decreased. BWS began at approximately 40% unloadin g with the ultimate goal of attaining 0%. Treadmill speed was initiated slower th an normal walking speed (normal 0 .8m/s) (Perry et al. 1995) then increased as soon as possible to normal and beyond according to
84 participant and/or trainer abilities. Guidance force on the Lokomat began at 100% and reduced toward 0% once treadmill speed was at least in a normal range and BWS was below 20%. Rest periods, typically standing with assistance for posture and BWS lowered to promote loading, were provided between stepping bouts. B lood pressure and heart rate were obtained before, at regular intervals during and after training sessions to assess participants physiological responses to exercise and to ensure episodes of autonomic dysreflexia or alternatively postural hypotension wer e avoided or addressed as necessary. Borg Ratings of Perceived Exertion also were recorded regularly according to the participants report.
85 Figure 31 Split belt instrumented treadmill, harness and instrumentation.
86 Figure 32. Spatial foot parameters using feet center of mass (CoM) as body reference points. White arrows represent (A) s tep width an d (B) s tep Length. This representative motion analysis graphic depicts an individual with SCI walking with a left forefoot initial con tact. Figure 33 Foot placement relative to center of mass (CoM) distances. Body reference points are the leading foot CoM and the pelvis CoM. White arrows represent (A) m e diolateral foot placement, (B) a nteroposterior foot placement. The same individual with SCI seen in Figure 32 is depicted here.
87 Figure 34 Example of center of pressure (CoP), center of mass (CoM), and extrapolated center of mass (XcoM) trajectories during walking with margin of stability (MoS) represented at the transition fr om double limb support to single limb support CoM trajectory length also demarcated between dotted lines for a single step (1/2 stride) in the transverse plane. Figure 3 5 Step length (a) and step width (b) defined for a given gait cycle. Gray circles represent reflective heel markers.
88 Figure 36 (A) Manual assisted locomotor training (B) R obotic assisted locomotor training
89 CHAPTER 4 EXPERIMENT 1 : INFLUENCE OF ASSISTI VE DEVICES ON HEAD S TABILIZATION DURING WALKING AFTER SPINAL CORD INJURY Following a spinal cord injury (SC I), individuals possess altered, and what could be described as new nervous systems (Edgerton et al. 2004) This injured nervous system is particularly plastic and inclined to learning based on the task specific sensorimotor experiences in which a person engages Therefore, as an individual attempts to relearn walking, the experiences introduced influence neural changes which manifest in 1) walking recover y in the way one walked prior to injury or alternatively 2) compensatory strategies that cope with losses and use only residual motor function s (Barbeau 2003) Traditionally in walking rehabilitation post SCI clinicians introduce assistive devices (ADs) to patients The intention is to provide support through the upper extremities and compensate for impairments contributing to balance dysfunction. H owever, these ADs provide externally controlled stability that may be preventing individuals from exploring normal balance strategies (Bateni and Maki 2005) Moreover, ADs may be teaching a task with sensory and biomechanical attributes different from normal walking (Jeka 1997; Dromerick et al. 2006; Ivanenko et al. 2009) Thus, a person likely is learning balance strategies in the context of having an AD present, rather than relearning how to develop balance strategies in the absence of external assistance. E valuating an indiv idual while walking without his/her AD therefore, may reveal the way the persons nervous system is coping with SCI mediated deficits and the solutions found in an attempt to recover balance. In contrast to a person with an injured central nervous system a person with a healthy nervous system has an array of possibilities for controlling walking balance An
90 individuals body segments coordinate together to create a n apparent effortless and smooth walking trajectory. Research suggests that this smoothness is necessary in order to stabilize the position of the head (Menz et al. 2003a; Kav anagh et al. 2004; Kavanagh et al. 2005b) Stabilization of the head is essential for acquisition of accurate sensory inputs to the visual and vestibular sy stems since these two systems have primary roles in maintain ing balance control (Shumway Cook and Woollacott 2001) In addition, when the feet initially contact the ground during walking, high linear accelerations at the legs transfer perturbations from body segment to body segment (e.g from shank to thigh to trunk) (Kavanagh et al. 2006) In order to optimize head stability and maintain balance control under these conditions linear accelerations decrease in a proximal to distal manner fr om the pelvis to the head (Winter et al. 1990; Menz et al. 2003a) The variable upper body movements often observed in persons post SCI have potential to inundate the nervous system with conflicting visual and vestibular information thus further propagating balance dysfunction The conventional i mplementation of ADs to provide leg an d trunk support may also reduce upper body motion and afford head stabilization. This result may further enhance compensated balance Additionally, visual field stabilization and balance control can improve through a self elected red uction in walking speeds, resulting in enhanced vestibulo ocular reflexes (Mamoto et al. 2002) Melis et al. (1999) observed s low walking speeds in persons with SCI when walking with ADs Speeds appeared to be at least partially attributed to type of device used with walker users exhibiting slower speeds than both crutch and cane users Consequently, given the clinical util ization of ADs to provide
91 postural stability during walking as well as the associated compromised walking speeds wh ile using devices, it is plausi ble to suggest that ADs may assist in achieving head stabilization, thus affording externally derived balance control for the SCI population. T he refore, t he main purpose s of this study were to investigate 1) the effect of ADs on head stability post SCI and 2) the effect of SCI on head stability during walking without ADs compared to healthy individuals. The first part of our investigation will provide insig hts into whether ADs have a role in stabilizing the head relative to the pelvis for balance control T he second part of the study will ass ist in understanding the abilities of individual s to maintain normal levels of head stability after injury when tested in a recovery based environment without ADs This particular assessment also could elucidate the possible longterm influences of not only the SCI but also chronic AD usage. Based on clinical intentions for ADs to stabilize and support the upper body, w e hypothesized that persons with SCI would demonstrate greater head stability while walking with ADs compared to withou t Specifically, head accelerations would attenuate to a greater degree relative to the pelvis and head disp lacement s would pos sess less m agnitude and variability relative to the pelvis when walking with ADs. Additionally, because the individuals with SCI will have experienced only walking conditions with ADs since injury, we anticipated they immediately would exhibit reduced head stability when devices were removed compared to healthy individuals. In p articular, persons with SCI would show increased head accelerations and larger, more variable head displacements relative to the pelvis compared to healthy persons. T he ability to preserve head stability with and without ADs may offer insight s into how a persons
92 nervous system prioritizes and solves walking balance dysfunction after injury when given the opportunity to explore two different sensory experiences M ethods P articipants A convenien ce sample of 10 i ndividuals with chronic, incomplete SCI (6 males; mean age=42.6 years, SD=14.2) (Table 41) and 10 healthy persons comprising a control group (3 males; mean age=56.1 years, SD=3.3) participated in this cross sectional study. Participants w ith SCI were involved in a larger randomized controlled trial with the following inclusion criteria: 1) at least 18 years of age, 2) injury sustained at least 6 months prior to the study, 3) upper motor neuron, motor incomplete sp inal cord lesion, 4) abili ty to ambulate at least 10m with or without an assistive device, and 5) injury of traumatic or nontraumatic origin, excluding those of congenital etiology. The subset of individuals selected for this study from the larger trial included those who could generate at least three steps without an assistive device, BWS, or physical assistance. The healthy adult controls were a convenience sample from a larger ongoing cross sectional study. All controls were18 years of age or older and full time ambulators witho ut assistive devices or physical assistance. All experimental procedures were conducted at the Brain Rehabilitation Research Center, Malcom G. Randall Veteran Affairs Medical Center in Gainesville, Florida. Each participant signed a written informed consent approved by both the VA Subcommittee for Clinical Investigation and the University of Florida Health Science Center Institutional Review B oard.
93 Experimental Procedures A licensed physical therapist assess ed bila teral upper and lower extremity motor and sensory function in persons with SCI based on the American Spinal Injury Association (ASIA) International Standards for Neurological and Functional Classification of Spinal Cord Injury ( American Spinal Injury Association 2002) This assessment establi sh ed SCI severity and categorized injuries according to the ASIA Impairment Scale (AIS). A physical therapist also assessed performance on standardized balance assessment s, the Berg Balance Scale (BBS) (Berg et al. 1992) and the Dynamic Gait Index (DGI) (Shumway Cook and Woollacott 2001) to provide further clinical descriptions of each individual s balance ability Both participants with SCI and healthy controls had reflective marker balls and rigid body clusters positioned on specified body landmarks to acquire 3D motion data. Marker positions we re based on the Vicon PlugInGait marker set (modif ied Helen Hayes set). All i ndividuals wore a safety harness attached to an overhead cable and track system tha t was suspended from the laboratory ceiling. Walking trials lasted a maximum of 30 seconds and were performed over a split belt instrumented t readmill (Tec machine, Inc.). Walking practice was permitted to become accustomed to walking on the treadmill and to obtain the best possible steady state walking speed. When comfortable, data collections commence d Those individuals with SCI perform ed two trials during which they were instructed to walk to the best of their abilities at a self selected (SS) t readmill speed (Figure 4 1) : the first trial with their customary assistive device, the second without any device. Although participants wore a safety harness should they stumble or fall, neither BWS nor manual assistance was p rovided during any trial. This testing condition was intended to capture true walking capacity post -
94 injury Either sitting o r standing rest periods were provided as needed between bouts of activity. Additionally, healthy controls performed three separate walking trials at 0.3 and 0.6 m/s as well as a SS treadmill speed. The first two speeds were selected to acquire normal comparisons at the slower speeds that persons with SCI elected to walk. Pepin et al. (2003b) suggest matching for speeds between SCI and control groups to eliminate speed as a potential confound and distinguish outcomes that are consequence s of the injury rather than speed. Walking at SS speeds also was conducted as much of the scientific literature assessing head stability in healthy adults utilizes SS speeds (Menz et al. 2003a; Kavanagh et al. 2004; Mazza et al. 2008) Data from SS speeds validated the similarity of our healthy adult results with other studies This step was necessary t o confirm our methodology and calculations prior to initiating comparisons at slower speeds Data Acquisition and Processing Twelve camera passive motion analysis (Vicon Motion Systems) and 3D grou nd reaction forces (GRFs) from four piezoelectric force tr ansducers (Advanced Medical Technology, Inc.) located beneath each half of the treadmill were acquired continuously during walking trials. The split be lt treadmill system allowed collection of GRFs for each stance phase over multiple steps of the gait cycl e. Raw kinematic data w ere collected at 100 Hz, then low pass filt ered using a fourthorder, zerolag Butterworth filter with a 6 Hz cut off frequ ency. GRFs were acquired at a sampling rate of 2000 Hz, and low pass filtered using a fourth order, zero lag B utterworth filter wi th a 20 Hz cut off frequency. A 13segment musculoskeletal model was created using Visual 3D (V3D) processing that fit the model to marker trajectories. Anthropometric and inertial values defined within V3D w ere applied for segment modeling and segmental center of mass (CoM)
95 calculations. V3D models were used to conduct inverse dynamics analyses for calculation of intersegmental joint kinetics. Custom Matlab programming (Mathworks, Inc.) was developed to calculat e the outcome measures Outcome Measures Using calculated segmental CoM positions, 3D displacements of head and pelvis CoMs were determined for each step in a trial. Subsequently, position data were doubledifferentiated to acquire 3D linear accelerations for the same steps. Visual inspection of acceleration profiles located abnormally high and low spikes, which were interpreted as non physiological noise. Therefore, m aximal and minimal thresholds were manually selected on the profil es via customized Matlab coding, and steps with accelerations exceeding the establish ed thresholds were eliminated. Accelerations. Root mean squares (RMS) of frameby frame acceleration values for both the head and pelvis CoMs in each direction were calculated from valid steps. R MS values then were used to calculate attenuation coefficients (AC) for the entire trial as described by Mazza et al. (2008; 2009). AC= (1 (RMSH/RMSP))* 100, where H = head, P = pelvis Attenuation coefficients, expressed as percentages, represent the abilit y of the individual to reduce accelerations from the pelvis to the head. A positive, larger coefficient indicates greater head stability relative to the pelvis. In contrast, negative coefficients represent larger head accelerations (less stability) relativ e to the pelvis. Displacements. T he mean s and standard deviations of 3D h ead and pelvis CoM displ acement s were calculated. Holt et al. (1999) quantified vertical head displacements as an indicator of head stability, and we expanded our methodology to include the anteroposterior and mediolateral directions in order to parallel our 3D acceleration
96 outcomes. Using similar methodology to Hong and Earhart (2010) for the relative motion between two segments t he difference between pelvis and head mean displacements ( mean difference, meanp elvismeanh ead) was an outcome indicating which segment produced greater displacements over the walking trial. The difference between pelvis and head displacement standard deviations ( variability difference, SDp elvisSDh ead) was an outcome indicating whether the head or pelvis produced more variable displacements over the course of a walking trial Similar to the acceleration attenuation coefficients, p ositive values for difference calculations represent less head motion relative to the pelvis, while negative values represent greater head motion relative to the pelvis. Data Analysis All statistical analyses were conducte d using SAS 9.13 software. A series of normality tests were conducted to ensure that SCI and control data for each outcome met all assumptions for parametric testing. P aired t tests were conducted to examine differences in the SCI group with and without ADs. For comparisons between the SCI and control group at matched speeds o utcomes for each person with SCI were standardiz ed using standard differences from the control group walking at a similar speed (SCI v alue minus control group mean divided by control group standard deviation) Thus, the control group mean equaled zero, variance equaled one and SCI values beyond 2 standardardized scores were considered outliers Speed matching was established as follow s: SCI 0.3 m/s = controls at 0.3 m/s; and SCI >0.3 m/s to 0.6 m/s = controls at 0.6 m/s. Visual analysis of SCI standardized data indicated heterogeneity in the sample with variability in both positive and negative directions relative to controls. Ther efore, i n order to avoid masking true differences in the SCI
97 sample and regressing SCI data toward a mean value, which might demonstrate no difference from controls, absolute values were calculated to determine a participants distance from a central point of zero for each outcome. The mean of all participants absolute values was used for group comparisons of SCI and healthy individuals using a permutation test. Alpha level for significance was set at .05. Results With versus Without Assistive Devices in S CI Accelerations. Figure 4 2 illustrates raw head and pelvis trajectories for one representative participant with SCI (SCI2) walking without a device. 3D a ttenuation coefficients (ACs) exhibited no statistical difference between persons with SCI walking wi th and without ADs ( p 0.05) (Figure 43) On average in both walking conditions participants demonstrated increased head accelerations relative to the pelvis in both the mediolateral (ML) and vertical (VT) directions. In the anteroposterior (AP) direction, head accelerations were al so higher when walking with ADs; but without ADs slightly lower head accelerations relative to the pelvis were exhibited (i.e. greater head stability). Figure 4 3 further illustrates the large standard deviations for all ACs. The individual participants presented with varying combinations of AC values (e.g. a positive value with AD and negative without AD or vice versa as well as positive or negative values in both walking conditions). Although averaged AC values showed no differences across conditions, RMS of the head and pelvis, which were used to calculate attenuation coefficients showed significantly greater accelerations at both the head and pelvis in the ML direction when walking without ADs ( p = 0.009 and p =0.006, respectively) RMS of the pelvis in the AP direction also was significantly greater without
98 ADs ( p =0.002) Remaining RMS values for the head and pelvis were not significantly different between walking conditions ( p 0.05). Displacements. Figures 44 A&B depict displacement mean differences and variability differences. In the ML direction, both the mean difference and variability difference demonstrated significantly greater head movement relative to the pelvis when walking without ADs (m ean, p =0.038; vari ability, p =0.039). Regardless of walking condition, ML head displacements were larger than the pelvis (negative values). All m ean and variability differences for the AP an d VT directions show ed no differences between walking conditions ( p 0.05). On average, the mean differences in these two directions with and without ADs revealed smaller head displacements relative to the pelvis. The opposite pattern was detected in variability as head displacements were overall more variable than the pelvis in the AP and VT directions; although without ADs, AP head and pelvis variability were similar (variability difference ~ 0 ). SCI Without Assistive Devices versus Controls at Matched Speeds Accelerations. Figure 4 5 Illustrates individual subject data for attenuation coefficients. When matched for walking speeds using standard differences from the controls, the SCI group walking without their customary ADs demonstrated significantly different AC values in the AP and VT directions compared to controls ( p 001). No differences were detected in the ML direction ( p =0.2823). Regardless of direction, ACs were reduced compared to the control mean, indicating a decreased ability to attenuate head accelerations. RMS values for the head and pelvis in all three direc tions were significantly different from controls ( p Displacements. Figure 4 6 A&B illustrates the heterogeneity of individual subject data for both mean differences and variability differences, respectively. As a group,
99 those with SCI showed signifi cant differences in displacement mean differences in the VT direction only compared to controls ( p Yet, t he variability in head displacements relat ive to pelvis displacements was significantly different from controls in all three directions ( p ). Discussion The framework and progression of comparisons in this study were developed to parallel the current paradigm shift occurring in SCI rehabil i tation. We began with an investigation to understand if our traditional rehabilitation approach usi ng ADs affected walking balance, specifically head stability, differently than when individuals we re challenged to walk without ADs. Our model then shifted to evaluate both persons with SCI and healthy individuals in the same walking context (i.e. without any external assistance besides a safety harness). Although a more common experimental design is for studies to first show comparisons of patient populations to healthy controls, we specifically wanted to highlight the transition from a conventional evalua tion approach to one that is task specific and emphasizes walking recovery Assistive Devices Impact Head Motion Relative to the Pelvis This is the first study to examine head stability, as an indicator of balance control, in this neurological population. In both acceleration and displacement outcomes, and regardless of whether devices were utilized, persons with SCI demonstrated great heterogeneity in their head stabilization as exemplified by large standard deviations around the mean values. Although no statistical differences were detected in ACs when wa lking with versus without device s, ACs revealed the same patterns of change as shown in the displacement mean and variability differences for their respective ML and AP directions Specifically, ADs appear ed to play a role in reducing the magnitude and
100 variability of ML head displacements relative to the pelvis While sign i fican t differences were not detected in the AP d irection, the opposite pattern of change emerged compared to the ML direction. Tha t is, the displacement mean and variability differences actually became more positive when ADs we re removed to the extent that the mean value across participants with SCI reflected less head motion overall compared to the pelvis (i.e. a positive value) Th is occurrence corroborates research in healthy individuals, which found greater AP head stability relative to the pelvis in comparison to the ML or VT directions (Kavanagh et al. 2004; Mazza et al. 2009) These findings suggest that preservation of stability in the AP direction may be a high priority for the nervous system during forward walking. Perhaps when devices are removed after injury that particular direction of stability remains a top priority, potentially to achieve accurate visual and vestibular sensory inp uts as one progresses forward. This prioritization may occur even at the expense of increasing ML head motion as seen here Interestingly, the VT direction demonstrated remarkably higher head accelerations relative to the pelvis with and without ADs. This would appear highly unusual in reference to the healthy literature, which suggests that VT accelerations at the head and pelvis should be the same because the spine mechanically links the two segments (Mazza et al. 2008) However, in clinical observations of walking persons with SCI often use a quick head thrust i n an upward and backw ard rotation to biomechanically achieve momentum and progress the lower body forward. Even though the head may not be vertically displaced much fa rther than the pelvis as seen in our study, which is likely due to the spines mechanical constraint, it is plausible that the quick motion could manifest
101 in higher head accelerations. Moreover, our findings in the vertical ACs s uggest that this biomechanical strategy persists regardless of AD use. Spinal Cord Injury Affects Head Stability Mul ti directionally Compared to normal in dividuals walking at comparably slow speeds, ACs in the AP and VT directions were significantly reduced in persons with SCI walking without devices (i.e. they presented with less head stability than is observed in healthy persons). For all participants, ML attenuation levels were similar to healthy controls Interestingly, both the ML and AP displacement mean differences were not different from healthy individuals ; y et, the variability in all directions was signifi cantly different. An increasing amount of literature proposes that the lack of consistency in ones movement, as opposed to its magnitude, is a more appropriate indicator of balance during walking (Brach et al. 2001; Granata and Lockhart 2008) In accordance with this view, head stability research also has expanded to include m easurement s of variability (H olt et al. 1999; Laudani et al. 2006) Thus the increased multidirectional variability of head motion relative to the pelvis post SCI is suggestive of reduced balance control. If one is unable to create head stability ( due to excessive trunk movements perhaps ) he/she may be compensating with other sensory inputs. For example, individuals post incomplete SCI may have spared neural pathways that retain adequate ankle, knee and/or hip proprioception or plantar somatosensation. Thus their alternative stra tegies to avoid falling may be to precisely place their feet and counteract the trunk and head motion with compensatory leg movements However, a nother important factor when evaluating someones head motion may be to consider whether the variability might be secondary to muscle weakness, in particular paraspinal muscle weakness. Although the passive mechanics of the spine
102 may have a stabilizing effect, the paraspinals also have been credited with stabilizing the up per body as they activate in a top down pat tern during walking (Prince et al. 1994). Presumably this recruitment order is a feedforward mechanism to stabilize the head in anticipation of perturbations. Our partic ipant sample with SCI included seven individuals with cervical injuries and three wit h upper thoracic injuries. Upper t runk and neck muscle weakness occurs with injuries along those levels (Larson et al. 2010) therefore although not specifically tested, it is plausible to suggest that those with SCI had difficult ly producing anticipatory muscle responses during walking. As a result, head stability was compromised. Although Thigpen et al. (2009) reported relatively normal anticipatory muscle responses in the lower extremities during a standing perturbation task in individuals post SCI, it cannot be assumed that anticipatory responses of the trunk muscles would respond the same way during walking since standing and walking have been shown to possess different stability requirements (Kang and Dingwell 2006) Conclusions The ability to maintain head stability after SC I varied greatly both with and without assistive devices and relative to healthy individuals. However, patterns of stability emerged which suggest that ADs may influence mediolateral head stability relative to the pelvis and that in the absence of ADs, individuals prioritize the ability to stabilize anteroposterior head motions. Variability of head displacements relative to the pelvis appeared particularly deficient compared to controls in all three dimensions as did the ability to attenuate accelerations in the AP and VT dir ections. Future work on head stabilization in this population should evaluate other segmental relationships located between the head and pelvis (i.e. pelvis to trunk, trunk to neck, neck to head) to provide a more in depth view of how head stability is or is not achieved. Continued efforts
103 should be taken to understand the impact of ADs on dynamic stability their influence on the plasticity of the nervous system, and their utility versus other evaluations strategies (e.g. treadmill based or overground with a harness only) in assessments of dynamic stability
104 Table 4 1. Participant demographics Participant Age (yrs) Sex Injury site Time post SCI (mos) LEMS (max:50) AIS BBS (max:56) DGI (max:24) Assistive device SCI1 45 M C5 6 10 43 D 46 17 RW SCI2 55 M C4 45 45 D 31 14 SPC SCI3 48 F C5 25.5 46 D 51 12 SPC SCI4 26 M T3 4 11 40 D 21 15 RW SCI5 66 M C7 78 49 D 48 17 SPC SCI6 47 F C4 6.5 43 D 19 12 RW SCI7 40 F T2 3 11 38 D 10 9 RW SCI8 21 M C6 7 45 D 17 8 RW SCI9 27 M T6 12 40 D 12 11 RW SCI10 51 F C4 5 7.5 45 D 42 14 SPC LEMS: Lower Extremity Motor Score, AIS: American Spinal Injury Association Impairment Scale, BBS: Berg Balance Scale, DGI: Dynamic Gait Index, RW: Rolling Walker, SPC: Single Point Cane
105 Figure 41. Differences in self selected treadmill speeds with and without customary assistive devices (AD) for participants with SCI. Figure 42. Example of 3D raw acceleration profiles of head and pelvis trajectories (SCI2 without device ). Dashed oval s outline regions of reduced head accelerations relative to the pelvis for a particular percent of the gait cycle. ML: mediolateral, AP: anteroposterior, VT: vertical, acc: acceleration.
106 F igure 43 A cceleration attenuation coefficients for persons with SCI walking with and without assistive devices (ADs) ML: mediolateral, AP: anteroposterior, VT: vertical. Error bars indicate standard deviations. No significant difference s were detected between walking conditions
107 Figure 44 Relative motion of head and pelvis displacements. (A) M ean differences and (B) variability differences in persons with SCI walking with and without assistive devices (ADs). ML: mediolateral, AP: anteroposterior, VT: vertical. Error bars indicate standard deviations. (*) denotes significant difference between walking conditions.
108 Figure 45 (A) Mediolateral, (B) anteroposterior, and (C) vertical standardized attenuation coefficients presented for each participant with SCI compared to a control mean of zero (variance=1). (*) denotes SCI group significant ly different from controls.
109 Figure 46 Standardized SCI data for relative motion of head and pelvis displacements. (A) M ean differences and (B) variability differences in the mediolateral, anteroposterior, and vertical directions compared to a control mean of zero (variance=1). (*) denotes SCI group significant ly different from controls.
110 CHAPTER 5 EXPERIMENT 2 : FOOT PLACEM ENT VARIABILITY AS A WALKING BALANCE CONTROL MECHANISM POST SPINAL CORD INJURY Balance during walking requires both upright posture and equilibrium control to avoid falling (Winter et al. 1990) In conjunction with the ability to reciprocally step and to adapt to the environment, balance is critical in achieving functional ambulation (Forss berg et al. 1980a; Barbeau et al. 2006) P ersons with spinal cord injury (SCI) re port balance dysfunction as a contributor to this populations high falls incidence and secondary injuries (Brotherton et al 2007a) L aboratory gait analyse s have reported that reliance on assistive devices that restrict excessive movements, the inability to bear weight fully through the lower extremities, and trunk flexion adaptatio ns all indicate poorer balance control post SCI (Karcnik and Kralj 1999; FieldFote et al. 2001; Behrman et al. 2005; Leroux et al. 2006; Scivol etto et al. 2008) Spinal cord lesions affect the sensorimotor systems involved in balance control thus creating precarious walking cond itions prone to falls, particularly when compensatory assistive devices are removed (Shumway Cook and Woollacott 2001; van Hedel et al. 2005; Horak et al. 2009) In order to rehabilitate this loss of balance control, m easurement tools that can quantify distinctive balance strategies of the SCI population are necessary to understand the effectiveness of interventions as w ell as an individuals progress over time Clinicians commonly document the quality of spatial parameters at the feet such as wide base of support when observing an increased step width to describe the stability in a persons walking pattern. However, these clinical observations do not account for the dynamic control of the bodys center of mass (CoM), which is essential to avoid falling. A s walking progresses forward the CoM accelerates past the single
111 stance limb and causes the CoM to fall with each step. Consequently precise placement of the contralateral swing foot is critical to recapture the CoM within the base of support (Patla et al. 1999) If the foot contacts the ground in a sub optimal position, subsequent steps must adjust to compensate for this error thus inducing small perturbations. Studies of soleus H reflexes during healthy walking confirm this nervous system priority to control placement of the foot and ensure stability when balance is perturbed (Krauss and Misiaszek 2007) Following SCI, each step dur ing level overground walking c ould be viewed as its own perturbation to the neuromuscular system necessitating corrective foot placements Increased or decreased variability in a patients spatial walking pattern relative to the amount of variability observed in healthy individuals oft en is considered indicative of instability (Stolze et al. 2000; Brach et al 2001) However the ability to balance may be more related to where one places his/her foot relative to the CoM rather than relative to the opposite foot Balasubramanian et al. ( 2010) reported a significant positive relationship between step length asymmetry and asymmetry in the anterior placement of the foot relative to the pelvic CoM in persons post stroke; however, no relationship existed between step width and the lateral placement of the foot relative to the CoM This disparity between measures suggests that a clinical examination of foot to foot distances as an indicator of balance may not always reflect a persons true ability to control CoM motion through foot placements. An additional measure has been developed in the laboratory to characterize balance control beyond spatial parameters and foot placement relative to the CoM. Gi ven the dynamic nature of walking, an expansion of th e se static measurement
112 approaches has been suggested to account for the impact of velocity on the CoM position during this task. Hof et al. proposed a measure called the margin of stability (MoS) (Hof et al. 2005; Hof 2008) This measure compares the shortest mediolateral distance between the center of pressure ( CoP ) and the extrapolated cent er of mass (XcoM) during double limb support as weight acceptance begins (from initial contact of the leading limb to toe off of the trailing limb) The XcoM is the vertical projection of the CoM in the direction of its velocity If the CoM reaches the boundary of the CoP and cont inues to have an outwardly directed velocity vector a person must take a step to increase the base of support or a fall result s. From the clinic to the laboratory, three methods for quantifying walking balance have been described in a rather hierarchical manner from basic locations of the feet to inclusion of the more body centric CoM and its velocity Yet, their collective potential to examine balance in the SCI population has not been explored. Thus, t he primary purpose of this study w as to investigate walking balance through spatial foot parameters, f oot placement relative to the CoM and MoS in persons post SCI. We aimed to understand how the variability of these measures differed between persons with SCI and healthy controls as well as how the variabi lity differed within those groups. Additionally, we sought to bridge the laboratory and clinical interpretations of balance to identify the utility of spatial foot parameters (both magnitude and variability) as potential clinical correlates for more complex balance tools that incorporate CoM and XcoM Three main hypotheses were established. First, we hypothesized that persons with SCI would demonstrate significantly different v ariability in all measures compared to healthy individuals Second, within the healthy group, the variability among spatial foot
113 parameters would be similar to MoS variability as well as to the variability of foot placement s relative to the pelvis Co M in t heir respective direction. H owever, those with SCI would exhibit significant within group differences among the same measures Third, we anticipated that both the magnitude and variability of spatial foot parameters would show a weak positive association w ith foot placements relative to the CoM in their respective directions as well as to MoS in persons with SCI; yet, a strong positive association would be detected in healthy individuals. To evaluate these hypotheses, w e specifically chose to study those individuals who could walk at least a few steps without devices or assistance even if they usually walk with a walker or cane for the following reasons For those individuals with SCI who are able to walk in any form and prevent themselves from falling, outcomes could provide a window into biomechanically how they are able to accomplish the task. Variability differences between and within SCI and healthy groups were anticipated due to disruptions of central nervous system circuitry post SCI. Although the spinal cir cuitry for stepping propagation would be intact in the individuals examined for this study the cortical and subcortical modulatory control responsible for posture and equilibrium would be interrupted (Jacobs and Horak 2007a) W ith the balance deficits and lack of distal limb control observed in the SCI population step length and width may exhibit great variability whereas the foot location relative to a moving CoM may be more tightly regulated from one step to the next. This regulation could be the underlying reason why falls are av oided for some individuals
114 Methods Participants A convenience sample of 10 i ndividuals with chronic, incomplete SCI (6 males; m ean age=42.6 years, SD= 1 4.2) (Table 51) and 10 healthy persons comprising a control group (3 males; mean age=56.1 years, SD=3.3) participated in this cross sectional study. Participants with SCI were involved in a larger randomized controlled trial with the following inclusion criteria: 1) at least 18 years of age, 2) injury sustained at least 6 months prior to the study, 3) upper motor neuron, motor incomplete sp inal cord lesion, 4) ability to ambulate at least 10m with or without an assistive device, and 5) injury of traumatic or nontraumatic origin, excluding those of congenital etiology. Th e subset of individuals selected for this study from the larger trial included those who could generate at least three steps without an assistive device, BWS, or physical assistance. The h ealthy adult controls were a convenience sample from a larger ongoin g cross sectional study All controls were 18 years of age or older and full time ambulator s without assistive devices or physical assistance. All experimental procedures were conducted at the Brain Rehabilitation Research Center, Malcom G. Randall Veteran Affairs Medical Center in Gainesville, Florida. Each participant signed a written informed consent approved by both the VA Subcommittee for Clinical Investigation and the University of Florida Health Science Cent er Institutional Review Board. Experimental Procedures A licensed physical therapist assess ed bila teral upper and lower extremity motor and sensory function in persons with SCI based on the American Spinal Injury Association (ASIA) International Standards for Neurological and Functional
115 Classificati on of Spinal Cord Injury (American Spinal Injury Association 2002). This assessment establish ed SCI severity and categorized injuries according to the ASIA Impairment Scale (AIS). A physical therapist also assessed performance on standardized balance assessments, the Berg Balance Scale (BBS ) (Berg et al. 1992) and the Dynamic Gait Index (DGI) (Shumway Cook and Woollacott 2001) to provide further clinical descrip tions of each individual. Both participants with SCI and healthy controls had reflective marker balls and rigid body clusters positioned on specified body landmarks to acquire 3D motion data. Marker positions we re based on the Vicon PlugInGait marker set ( modif ied Helen Hayes set). All individuals wore a safety harness attached to an overhead cable and track system that was suspended from the laboratory ceiling. Walking trials lasted a maximum of 30 seconds and were performed over a split belt instrumented treadmill ( Tec machine Inc.). Walking practice was permitted to become accustomed to walking on the treadmill and to obtain the best possible steady state walking speed. When comfortable, data collections commence d Those individuals with SCI perform ed one walking trial during which they were instructed to walk to the best of their abilities at their self selected (SS) t readmill speed. All walked without assistive devices or braces. In addition, although participants with SCI wore a safety harness for each trial should they stumble or fall, neither bodyweight support ( BWS ) nor manual assistance was provided during any trial. This testing condition was intended to capture true walking capacity post injury Either sitting o r standing rest periods were provided as needed between bouts of activity. Additionally, healthy controls performed two separate walking trials at
116 0.3 and 0.6 m/s for normal comparisons to the speeds which persons with SCI elected to walk. Data Acquisition and Proce ssing Twelve camera passive motion analysis (Vicon Motion Systems) and 3D grou nd reaction forces (GRFs) from four piezoelectric force transducers (Advanced Medical Technology, Inc.) located beneath each half of the treadmill were acquired continuously duri ng walking trials. The split be lt treadmill system allowed collection of GRFs for each stance phase over multiple steps of the gait cycle. Raw kinematic data w ere collected at 100 Hz, then low pass filt ered using a fourthorder, zerolag Butterworth filter with a 6 Hz cut off frequ ency. GRFs were acquired at a sampling rate of 2000 Hz, and low pass filtered using a fourth order, zero lag Butterworth filter wi th a 20 Hz cut off frequency. A 13segment musculoskeletal model was created using Visual 3D (V3D) p rocessing that fit the model to marker trajectories. Anthropometric and inertial values defined within V3D w ere applied for segment modeling and segmental center of mass (CoM) calculations. V3D models were used to conduct inverse dynamics analyses for calc ulation of intersegmental joint kinetics. Custom Matlab programming (Mathworks, Inc.) was developed to calculate the outcome measures described below. Outcome Measures M eans were calculated using individual step data in a walking trial to indicate the average magnitude of each outcome for each person. Values for b oth right and left legs were entered into calculations for controls as well as those with SCI. Variability across each persons walking trial was calculated using standard deviations Spatial a nd f oot placement parameters. Two spatial foot parameters and two foot placement parameters were calculated. The two spatial foot parameters were us ed
117 to indicate the mediolateral ( ML ) and anteroposterior (AP) distances of one foot relative to the contralateral foot. The two foot placement parameters w ere us ed to indicate the ML and AP distances of the foot relative to the falling body (Balasubramanian et al. 2010) Spec ifically, these four measures we re defined as the following distances at each initial contact : 1) leadin g foot CoM to trailing foot CoM in the ML direction (Step width) and in the AP direction (Step length) and 2) leading foot CoM to pelvis CoM in the ML direction ( ML foot placement ) and in the AP direction ( AP foot placement ) (Figure 5 1) Margin of stability (MoS). Based on Hof and colleagues definition (2005), MoS was calculated as the shortest perpendicular ML distance between the CoP and the extrapolated center of mass ( XcoM ) d uring double limb support (Figure 53 ) 1. XcoM = | (x + ( 2. MoS = | umax XcoM |, where umax = boundary of CoP A larger MoS is consistent with increased dynam ic stability as a greater boundary exists between the maximum XcoM an d the CoP Furthermore, the size of the margin is proportional to the impulse that would be required to destabilize or unbalance a person. That is, by adding a critical amount of CoM velocity (i.e. impulse) in the direction of the nearest CoP, an uns table situation can be created. Data Analysis All statistical analyses were conducte d using SAS 9.13 software. Means and standard deviations were calculated for all biomechanical measures. Standard deviations (i.e. variability) of data for each person with SCI were standardiz ed using standard differences from the control group walking at a matched speed. Thus, the
118 control group mean equaled zero, variance equaled one, and SCI values beyond 2 standardardized scores were considered outliers Speed matching was established as follows: SCI 0.3 m/s matched to controls at 0.3 m/s; and SCI >0.3 m/s to matched to controls at 0.6 m/s. Visual analysis of SCI standardized data indicated heterogeneity in the sample with variability in both positive and negative directions r elative to controls. Therefore, i n order to avoid masking true differences in the SCI sample and regressing SCI data toward a mean value, which might demonstrate no difference from controls, absolute values were calculated to determine a participants distance from a central point of zero for each outcome. The mean of all participants absolute values was used for group comparisons of variability outcomes between the SCI group and healthy individuals using a permutation test. Using nonsta ndardized data, a paired t test examined differences between measures of variability within the SCI group and within the control group at 0.3 and again at 0.6 m/s. Also using nonstandardized data, P earsons correlations investigated the relati onships betw een mean magnitudes of each measure as well as between their variability in the SCI group and in controls at both treadmill speeds Alpha level for significance was set at .05. Results Variability of Outcomes: SCI versus Controls (Hypothesis 1) Of the five outcome measures, participants with SCI displayed significantly different variability from controls as hypothesized ( p 0.007). Figures 54 a c show standardized data for individual participants which illustrate direction and magnitude of differences from a control mean of zero, in addition to the various combinations and degrees of variability across participants. T he majority of participants exceeded + 2 standardized scores for step width, step length, and ML and AP foot placement
119 variability ; in compariso n, fewer participants with SCI exhibited MoS variability outside of th is value Values beyond two standardized scores demonstrate greater movement variability compared to the variability exhibited by most healthy individuals Ho wever, it should be noted th at there were individuals with SCI in our sample (e.g. SCI3 in all five measures) who demonstrated negative deviations from the control mean. Negative values indicate decreased variability, but no participants presented with such limited variability that v alues occurred below 2 standardized scores. Variability of Outcomes within Participant Group (Hypothesis 2) Figure s 5 5 a c illustrate within group differences in outcome variability. Both the SCI group and controls walking at 0.3 and 0.6 m/s revealed significantly greater variability in step length compared to in AP foot placement ( p 0.001). Each group also exhibited significantly greater variability in step width compared to MoS ( p 0. 05) Variability of step width was significantly greater than the ML foot placement for controls walking at 0.6 m/s ( p = 0.0006) and was only marginally significant at 0.3 m/s ( p = 0.073) However, no significant difference was detected in this comparison for the SCI group ( p = 0.126) Step length was significantly more variable than MoS for both those with SCI ( p = 0.004) and controls at 0.3 m/s ( p = 0.027) but not for controls walking at the faster speed ( p = 0.276) These findings are partially c ontrary to our hypothesis as controls demonstrated significant differences between some measures of variability and the SCI group did not reach significance between some measures. Associations Among Outcome M agnitudes by Participant Group (Hypothesis 3) Tables 5 2 to 54 highlight the associations between mean outcome magnitudes within participant groups. The SCI group only exhibited two significant positive relationships: step width with ML foot placement (r=0.901, p =0.004) and step length with
120 AP foot placement (r=0.641 p =0.042) Control s walking at 0.3 m/s showed the same two relationships as the SCI group, but additionally revealed a significant inverse association between step length and MoS (r= 0.792, p =0.011) which are distances orthogonal to one another. Interestingly, no relationship was observed between MoS and s tep length when controls walked at 0.6 m/s but step width and MoS were significantly positively related ( r=0. 751, p =0.0 20) However, as seen at 0.3 m/s and also i n those with SCI, the step lengt h and AP foot placements were significantly related in the positive direction (r=0.663, p =0.051) ; step width and ML foot placement were marginally related (r=0.614, p =0.059) Control s at 0.6 m/s also showed a significant positive association between the st ep length and the ML foot placement (r=0.695, p =0.038) As with hypothesis two, hypothesis three was only partially supported. Significant positive associations were detected within both groups rather than solel y within the control group. Moreover, although we anticipated all positive relationships across measures, regardless of group, one inverse relationship resulted. Associations Among Outcome V ariability by Participant Group (Hypothesis 3) Tables 5 5 to 57 present associations among m ean outcome variability for the SCI group and for the control group at the two different treadmill speeds. Unlike the outcome magnitudes in which only two associations were noted the SCI group displayed significant positive relationships between spatial p arameter variabilities and all other measures This was true regar dless of direction and included the association between step width and step length themselves (r 0. 802, p 0.005) T he relationship between MoS and step length variability demonstrated a tren d toward significance (r=0.597, p =0.068) Fewer associations were detected within the control group. At 0.3 m/s, only the step length and AP foot placement variability were significantly correlated
121 (r=0.913, p =0.0006) Controls walking at 0.6 m/s also showed the same significant relationship as they did at the slower speed (r=0.804, p =0.009) but additionally exhibited a significant correlation for step width and ML foot placement (r=0.632, p =0.050) The analysis partially supported our hypothesis for associations among outcome variability as was noted similarly for outcome magnitudes. Both persons with SCI and controls at both speeds exhibited significant positive relationships across measures as well as weaker, non significant associations. Discussion S pinal Cord Injury Alters Movement V ariability To our knowledge, this is the first study to explore the intrinsic capacity for walking balance (i.e. without assistive devices) in person s with SCI using variability as an indicator of balance. Understanding the way individuals avoid falling by challenging them to engage paretic trunk and limb muscles is compatible with the current shift in neuro rehabilitation toward recovery of function. O ur comparison of persons post SCI and healthy individuals revealed that an injury significantly changes the variability of step width, step length, ML and AP foot placements relative to the CoM, and MoS from normal values Regardless of whether measurement s were recorded simply between feet or whether they included the CoM and its velocity variability across steps was abnormal after injury Although variability was primarily higher than normal for most persons with SCI, a select few demonstrated variabilit y lower than the control mean. A debate exists in the scientific literature attempt ing to discern between good or bad movement variability (van Emmerik and van Wegen 2002) This discussion extends to the literature on balance (also referred to as stability ) with efforts to determine if low variability equates with stability and high variability with instability or
122 vice versa (van Emmerik and van Wegen 2002) Research of physiological responses such as heart rhythms indicates that a certain degree of variability is normal and a lack of variability, or conversely too much variability, is pathologic (Glass 2001) The same interpretation of variability can be applied to walking balance Balance is based similarly on underlying physiological mechanisms such as afferent feedback loops; only the result is motor output of the head/trunk/pelvis and extremities (Horak 2006) While a certain level of variability is inherent in healthy individuals, r esearch has described relationships of high and low stepping variability with balance deficits in elderly populations (Granata and Lockhart 2008) a nd those with certain neurological disorders, such as normal pressure hydrocephalus (Stolze et al. 2000 ) respectively T he analysis selected to evaluate our first hypothesis allowed for consideration of a normal range of variability (i .e. that even healthy individuals would not have perfectly consistent steps across a walking trial) By standardizing each person with SCI to the healthy group and controlling for treadmill speed, deviations from a normal range (i.e. beyond 2 standardized scores) were detected. In order to control CoM motion for walking balance, Dietz (2002) states that the afferent inputs weighted and selected by the central nervous system must meet the equilibrium requirements of that task. Furthermore, since SCIs interrupt the flow of sensory feedback to the supraspinal centers required for integration of information responsible for balance (e.g. brain stem, cerebellum, motor cortex) (Macpherson et al. 1997) the ability of the nervous syst em to appropriately select and utilize sensory information may be impaired. As a result, the motor output may contain errors, which continually will require corrections. Additionally, Barbeau et al. (1999) reported sev eral
123 sensorimotor factors post SCI that impact walking recovery, including balance. Muscular weakness and dyscoordination as well as hyperactive spinal reflexes were proposed factors. The presence of neuromuscular impairments such as these could feasibly alter the ability of an individual to produce a consistent series of steps, thus creating differences in variability from normal. Var iability Differences w ithin Groups In addition to differences observed between SCI and healthy groups, differences also were shown within groups across measurements of variability. R egardless of group (and regardless of speed for controls), spatial foot parameters revealed significantly gre ater variability than measures involving the CoM and MoS although the specific measures showing these differences varied depending on the group. Because these findings were based up on steps without devices during which persons with SCI were capable of sta ying upright without falling suggests that these individuals maintained balance control via more consistent foot placements relative to the CoM, particularly the XcoM, than via foot to foot distances. Healthy individuals responded similarly to those with SCI even though as indicated by the between groups comparison, persons with SCI possessed more variable motion than normal overall. Consequently it may not always be possible to simply examine one measure of variability and pre sume that va riability is occurring to a comparable degree i n another measure. For example, the amo unt of variability in step width across a persons walking trial may not possess the same amount of variability in MoS Based on our results, this specific example would be true for persons with SCI at their SS speeds as well as for healthy individuals walking at slower than normal speeds of 0.3 and 0.6 m/s. Individuals exhibit more variability in their
124 step widths than in their MoS thus implying that the CoM motion provi des us with different information about balance Variability May Be a Better Clinical Correlate for Balance than Magnitude Our investigation also sought to understand the potential of spatial parameters to function as clinical correlates for control of CoM motion Based on our findings, we suggest that outcome variability is a more robust quantification of balance control than ma gnitude. O utcome magnitudes showed strong positive associations which were common among the SCI group and controls alike. In part icular outcomes measured along the same plane (step width and ML foot placement as well as step length and AP foot placement) demonstrated these commonalities. For all groups, these relationships may be one strategy by which individuals regulate control o f the CoM. The size of spatial parameters further reflect s the magnitude of the respective AP or ML foot placement. These results are partially consistent with literature correlating measures of foot to foot distances with foot placement relative to the Co M in other populations (Hof et al. 2007) Persons post stroke who produced asymmetric step lengths also placed their feet asymmetrically relative to their pelvic CoM (Bal asubramanian et al. 2010) ; however, in contrast to findings in those with stroke our sample of persons with SCI also showed that step width and ML foot placement rela tive to the pelvic CoM were significantly positively related. Since the magnitude rela tionships seen in our study we re similar across SCI and control groups, the challenge remained to determine why and quantify how persons with SCI we re unstable as reflected on standardized balance tests and assistive device s use d for walking support (Table 5 1). Our analysis of variability began to address this challenge.
125 While literature exists that correlates the magnitudes of measures similar to those in our study (Hof et al. 2007; Balasubramanian et al. 2010) correlations of variabilities have not been examined until now. We showed that m easures of v ar iability exhibited a more extensive set of significant correlation s for persons with SCI than for healthy controls at 0.3 or 0.6 m/s While controls at either speed showed only one or two strong correlations, which incidentally were among the same measures as those seen in magnitudes, persons with SCI revealed strong posit i ve associations in all measures with the exception of a marginal significance for step length and MoS These associations were even present between step widths and step lengths themselves. The wide array of associations irrespective of direction, s ugges t s that individuals with SCI avoid falling via a continuous pattern of step adaptations within a walking trial. A sequence of recovery steps occurs in healthy persons also when the lower extremities are perturbed. Purportedly, this recovery strategy is due to the nervous systems automatic control to recapture CoM within the base of support (Oddsson et al. 2004) If each step in a person with SCI produces its own perturbation caused by step error, individuals could be attempting to correct for errors one step after the next Yet the series of steps may be quite different in the combination of step l engths and step widths. On the contrary t he lack of multiple significant associations across variability measures in controls may indicate that they have the ability to control parameters in different directions independently (i.e. AP foot placement and s tep width) Suggestions for Clinical Translation Using our analysis of relationships among outcomes of variability, a clinician could plausibly measure the variability of step widths and step lengths in a patients walking pattern and cautiously draw some general conclusions ab out the patients balance
126 abilities (assuming the individual is walking at a slow speed comparable to those that were studied here i.e. 0.6 m/s or slower ). I f the patients level of variability is abnormal in a particular spatial foot parameter, then based on our correlation analysis, the amount of variability for measures including control of the CoM and XcoM is most likely abnormal and similarly increased or decreased. These abnormalities would suggest walking balance deficits and c ould be examined over time. Because of time constraints in the clinical environment, performing calculations and comparisons with the literature is frequently not possible. Therefore, adapt ing equipment that collects spatial parameters, such as a GaitRite walkway or the more recently evolving inertial sensors, and modifying software to include literature based normal values of variability could assist clinicians in deriving quick interpretations about a patients balance. At this time, a dditional studies ar e necessary that 1) evaluate controls at a greater range of speeds, 2) evaluate larger samples of both controls and persons with SCI, and 3) determine specific cut off values of normal variability for step widths and lengths to create a more efficient and clinically friendly interpretation of this measurement tool. Further Clinical Considerations within a RecoveryBased Framework Quantifying walking balance following SCI is essential to understand progression of an individuals function over time. However the interpretation of the measurement tools implemented needs to be considered in light of the conditions under which an individual was evaluated. A unique feature of this study was the testing environment utilized to investigate walking balance. By having individuals with SCI walk in a safety harness while on a treadmill they were afforded the opportunity to explore any immediate balance responses without the constraint of an assistive device Thus, this testing condition creates a recovery or activity b ased evaluation environment, which
127 parallels the rehabilitation paradigm shift toward activity based therapies (Dromerick et al. 2006) This is in contrast to conventional evaluation and treatment approaches, which allow an individual to compensate for impairments with assistive devices and produce different movement patterns than they would if devices were removed and upper extremities were unloaded. Leroux and colleagues (2006) also examined individuals with SCI without their assistive devices during walking on a treadmill in a harness. While they assessed the ability of an individual to adapt to walking on a treadmill at different inclines, our intentions were to examine balance responses during level ground walking and only to the perturbations induced by a persons individual walking pat tern (i.e. without external perturbations or environmental changes during the task) Given this testing environment, the findings presented in this study should be carefully considered when attempting to translate to a clinical arena without such equipment capabilities. The potential exists that outcomes might have been different if individuals were walking with their customary assistive devices or overground rather than on a treadmill. However, in remaining consistent with the paradigm shift in neurorehabi litation, the outcomes presented here complement the transition toward activity based therapies and can assist in understanding the effects of these therapeutic strategies. Conclusions Whether examining only spatial foot parameters or foot placement relat ive to CoM and XcoM, persons after SCI exhibit a significantly different amount of variability (usually greater) compared to normal levels of variability. However, it appears that even in the presence of this variability, individuals retain the ability to stay upright and avoid falling via a continuous pattern of adaptive foot placement strategies. Future
128 investigations should continue to examine the variability of movements that individuals possess post SCI as an indicator of balance control during walking Understanding the repertoire of balance strategies in this population will assist clinicians in targeting therapies that address specific balance deficits.
129 Table 5 1. Participant demographics Participant Age (yrs) Sex Injury site Time post SCI (mos) Assistive device LEMS (max:50) AIS BBS (max:56) DGI (max:24) Self selected treadmill speed (m/s) SCI1 45 M C5 6 10 RW 43 D 46 17 0.3 SCI2 55 M C4 45 SPC 45 D 31 14 0.25 SCI3 48 F C5 25.5 SPC 46 D 51 12 0.3 SCI4 26 M T3 4 11 RW 40 D 21 15 0.15 SCI5 66 M C7 78 SPC 49 D 48 17 0.5 SCI6 47 F C4 6.5 RW 43 D 19 12 0.12 SCI7 40 F T2 3 11 RW 38 D 10 9 0.2 SCI8 21 M C6 7 RW 45 D 17 8 0.2 SCI9 27 M T6 12 RW 40 D 12 11 0.03 SCI10 51 F C4 5 7.5 SPC 45 D 42 14 0.25 LEMS: Lower Extremity Motor Score, BBS: Berg Balance Scale, DGI: Dynamic Gait Index, RW: Rolling Walker, SPC: Single Point Cane
130 Figure 51. Spatial foot parameters using feet center of mass (CoM) as body reference points. White arrows represent (A) step width and (B) s tep Length. This representative motion analysis graphic depicts SCI10 walking without a device with a left forefoot initial contact. Figure 52 Foot placement relative to center of mass (CoM) distances. Body reference points are the leading foot CoM and the pelvis CoM. White arrows represent (A) m e diolateral foot placement, (B) a nteroposterior foot placement This image depicts SCI10 as in Figure 51.
131 Figure 53 Raw center of pressure (CoP), center of mass (CoM), and extrapolated center of mass (XcoM) trajectories. The vertical black line connecting the peak of the XcoM with the CoP represents the margin of stabili ty (MoS) for a single step. The gray vertical line extending beyond the MoS distance illustrates the time in the gait cycle (end of double limb stance) when the shortest MoS occurred for that particular step.
132 Figure 5-4 Variability of standardized outcomes for individual participants with SCI. Values represent direction of deviation from the control mean of zero (variance=1). (A) Step width and mediolateral (ML) foot placement; (B) Step length and anteroposterior (AP) foot placement; (C) Margin of stability (MoS) (*) indicates significant difference from controls at p.
133 Figure 5-5 Variability of outcome measures by participant group. (A) SCI at self -selected speeds ; (B) Controls at 0.3 m/s and (C) at 0.6 m/s ML: mediolateral, AP : anteroposterior MoS: Margin of stability (*) indicates trend toward significance. (* ) indicates significant difference at p
134 T able 5-2 Associations between mean outcome magnitudes for participants with SCI at SS speeds. MoS: margin of stability, ML: mediolateral, AP: anteroposterior. (*) indicates trend toward significance. (**) indicates significant difference within group at p Table 5-3 Associations between mean outcome magnitudes f or healthy controls at 0.3 m/s. Table 5-4 Associations between mean outcome magnitudes for healthy controls at 0.6 m/s. Outcome Step width Step length Step width r -0.157 p -0.687 Step length r 0.157 -p 0.687 -ML Foot placement r 0.614 0.695 p 0.059 ** 0.038 AP Foot placement r 0.423 0.663 p 0.256 ** 0.051 MoS r 0.751 0.014 p ** 0.020 0.972 Outcome Step width Step length Step width r -0.241 p -0.502 Step length r -0.241 -p 0.502 -ML Foot placement r 0.901 0.088 p ** 0.0004 0.808 AP Foot placement r -0.155 0.649 p 0.669 ** 0.042 MoS r 0.392 0.072 p 0.262 0.844 Outcome S tep width Step length Step width r --0.167 p -0.668 Step length r 0.167 -p 0.668 -ML Foot placement r 0.772 0.242 p ** 0.015 0.531 AP Foot placement r 0.172 0.942 p 0.657 ** 0.0001 MoS r 0.513 0.792 p 0.158 ** 0.011
135 Table 5-5 Associations between outcome variabilities for participants with SCI at SS speeds. Outcome Step width Step length Step width r -0.802 p -** 0.005 Step length r 0.802 -p ** 0.005 -ML Foot placement r 0.888 0.822 p ** 0.001 ** 0.004 AP Foot placement r 0.857 0.920 p ** 0.002 ** 0.0002 MoS r 0.915 0.597 p ** 0.0002 0.068 MoS: margin of stability, ML: mediolateral, AP: anteroposterior. (*) indicates trend toward significance. (**) indicates significant difference within group at p Table 5-6 Associations between outcome variabilities for healthy controls at 0.3 m/s. Table 5-7 Associations between outcome variabilities for healthy controls at 0.6 m/s. Outcome Step width Step length Step width r -0.169 p -0.665 Step length r 0.169 -p 0.665 -ML Foot placement r 0.632 0.131 p ** 0.050 0.738 AP Foot placement r -0.172 0.804 p 0.658 ** 0.009 MoS r 0.274 0.230 p 0.475 0.552 Outcome Step width Step length Step width r -0.234 p -0.545 Step length r 0.234 -p 0.545 -ML Foot placement r 0.459 0.509 p 0.214 0.162 AP Foot placement r 0.277 0.913 p 0.471 ** 0.0006 MoS r 0.044 0.458 p 0.910 0.215
136 CHAPTER 6 EXPERIMENT 3: DIFFER ENTIAL EFFECTS OF MA NUAL ASSISTED VERSUS ROBOTIC ASSISTED LOCOMOTOR T RAINING ON DYNAMIC S TABILITY POST SCI Individuals with spinal cord injuries (SCI) often possess disruptions in d ynamic stability defined as the ability to control the center of mass during walking (Barbeau et al. 2006; Brotherton et al. 2007a) In an effort to prevent falls from dynamic stability dysfunction and yet promote mobility clinicians treating patient s with SCI commonly prescribe assistive devices to provide support and/or to compensate for impairments in the trunk and lower extremities. Unfortunately, use of assistive device s post SCI reduce s lower extremity muscle activity, increase s upper extremity weight bearing and consequently induce s trunk flexion (Melis et al. 1999) Furthermore, removal of these devices reveals chronic postural limitations, as t he trunk appears to adapt to a flexed position over time. (Leroux et al. 2006) Maintenance of such abnormal postures may prevent acquisition of normal balance strategies particularl y given evidence that h ealthy persons who voluntarily adopt a flexed trunk during walking exhibit altered muscle synergy patterns, ground reaction forces, kinematics and kinetics (Grasso et al. 2000) Therefore, therapies are necessary that can address functional recovery of dynamic stability rather than emphasize comp ensatory strategies and induce chronic deficits by prolonged tra ining with assistive devices. Locomotor training (LT) using a bodyweight support (BWS) and treadmill system may be one rehabilitation technique that provides an optimal environment for persons with SCI to maximize function al recovery of dynamic stability (Barbeau 2003; Behrman et al. 2005; Dobkin et al. 2006) R emoval of upper extremity weightbearing, and in turn, promotion of loading through the low er extremities, upright posture and appropriate stepping kinematics allows individuals to achieve a sensorimotor experience consistent
137 with normal bipedal walking (Dietz and Harkema 2004; Behrman et al. 2005; Wirz et al. 2005; Behrman and Harkema 2007; Dietz 2008) At this time, two primary LT environments are available: 1) manual assisted LT (MLT) with traine rs at the lower extremities, pelvis, and trunk physically facilitating the best possible walking pattern through sen sory cueing and 2) robotic assisted LT (RLT) with an exoskeleton and computer system driving the walking pattern. Both training scenarios allow for an intense, repetitive stepping experience as they attempt to engrain the nervous system with the constituents of walking through principles of activity dependent plasticity (Kleim and Jones 2008) M LT creates an environment of increased walking variability due to the physical fatigue and frequent rotation of trainers on a bout to bout and day to day basis. D ifferent trainers applying variable forces and amounts of assistance does not assure reproducibility of where the foot is placed or how the pelvis is rotated or l aterally shifted (Galvez et al. 2007) Although variability may develop, motor learning theory purports that such a training scenario should increase retent ion and transfer of walking to real world contexts (Riolo Quinn 1999) Conve rsely, RLT promotes an environment in which a computer system creates consistent and relatively symmetrical stepping based on hip and knee range parameters established for the exoskeleton motion. Trochanteric pads immobilize the pelvis to prevent rotation or shifting, trunk straps enable an upright posture as needed, and toe lifters are available to dorsiflex the ankle passively when active motion is minimal or absent. The lateral bulk of the robotic exoskeleton may hinder a natural arm swing normally observed in healthy walking, whereas trainers may physically facilitate arm swing with bilateral, horizontal poles in the MLT setting
138 (Behrman et al. 2005; Hidler et al. 2009) Although the robotic system was developed with the int ention of simulating the same training situation as in the MLT environment, the need exists to understand whether these two environments do indeed have the same training effect. A wareness of each environments benefits in altering the walking stability cap acity of an individuals nervous system has potential to direct h is/her rehabilitation plan more effectively and efficiently. Theref ore, the purpose of this study was to examine the differential effects of MLT versus R LT on dynamic stability during walking in persons with chronic, motor incomplete SCI. More specifically, we aimed to understand how closely persons with SCI approximated normal balance abilities before they train ed and how each mode of LT sh ifted those abi liti es relative to normal walking. Since recovery of dynamic stability is the goal, normal walking is the reference by which recovery can be assessed. We hypothesized that compared to a healthy control group, persons with SCI prior to their respective LT i ntervention would exhibit significant differences in measures of dyna mic stability based on the literature describing a high falls incidence and balance deficits in thi s population (Brotherton et al. 2007b) However, post training, those who underwent R LT would show outcomes similar to control values whereas those who underwent M LT would conti nue to demonstrate significant differences. We anticipated these outcomes because RLT prov ides a consistent walking pattern for individuals with SCI who are newly learning to walk without their assistive devices. Si nce functional walking requires one to resist both internal ly generated and external ly provided perturbations, we believe an individual might need to learn how to consistently resist their own internal perturbations (e.g. neuromuscular errors resulting from treadmill speed or bodyweight
139 support changes), prior to being introduced to a variable environment that creates external perturbations (e.g. variability in trainer forces). Also, testing was conducted on a level ground treadmill without assistive devices or bodyweight support S ince external perturbations were not introduced during testing (i.e. only internal perturbation s were present ) those who went through RLT would have been trained in such a context. Finally, b ecause of the anticipated injury heterogeneity among participants, any differences hypothesized before and after LT were expected to present bi directionally (i.e. more or less of a given motion) relative to controls. Methods P articipants Ten individuals with chronic, motor incomplete SCI (6 males; mean age=42.6 years, SD= 1 4.2 A IS D ) (Table 61) and 10 healthy persons comprising a control group (3 males; mean age=56.1 years, SD=3.3) were recruited to participate in this study. Participants with SCI were involved in a larger randomized controlled trial with the following inclusion criteria: 1) at least 18 years of age, 2) injury sustained at least 6 months prior to the study, 3) upper motor neuron, motor incomplete sp inal cord lesion, 4) ability to ambulate at least 10m with or without an assistive device, and 5) injury of traumatic or non traumatic origin, excluding those of congenital etiology. The subset of individuals selected for this study from the larger trial included those who could generate at least three steps without an assistive device BWS or physical assistan ce. The healthy adult controls were a convenience sample from a larger ongoing cross sectional study. All controls were18 years of age or older and full time ambulators without assistive devices or physical assistance. All experimental procedures were cond ucted at the Brain Rehabilitation Research Center, Malcom G. Randall Veteran
140 Affairs (VA) Medical Center in Gainesville, Florida. Each participant signed a written informed consent approved by both the VA Subcommittee for Clinical Investigation and the Uni versity of Florida Health Science Cent er Institutional Review Board. Locomotor Training Intervention Participants with SCI were randomized to the MLT or RLT environment as part of the parent clinical trial. Five participants trained via the manual interve ntion and five via the robotic intervention. Training occurred 5 days per week for 45 sessions total At least one licensed physical therapist w as present for each training session. The intervention goal was for participants to achieve 30 minutes of qualit y stepping per session; however, the length and number of individual bouts necessary to obtain this goal varied daily depending on each participants functional ability and trainer fatigue. Rest periods, typically standing with assistance for posture and with BWS lowered to promote loading, w ere provided between s tepping bouts. Vital sign s were obtained before, after, and at regular intervals during training sessions to assess participants physiological responses to exercise and ensure possible episodes of autonomic dysreflexia or postural hypotension we re avoided or addressed as necessary. Over the 45 sessi ons, participants were challenged with progression of training parameters while maintaining optimal stepping patterns, including an upright trunk. In p articular BWS was decreased, treadmill speed increased and either robotic guidance force or trainer manual assistance decreased. Over the course of nine weeks, the MLT group increased mean treadmill speeds from 0.71 m/s (SD=0.14) to 1.0 m/s (SD=0.15) and number of training bouts changed from 8.43 (SD=2.51) to 9.27 (SD=9.74). The RLT group increased mean treadmill speeds from 0.69 m/s (SD=0.06) to 0.87 m/s (SD=0.11) and number of training bouts changed from 4.71 (SD=1.16) to 4.30 (SD=1.43). BWS
141 decreased in t he MLT group from 35.98% (SD=8.01) to 18.29% (SD=8.77) while the RLT group decreased BWS from 34.84% (SD=4.70) to 13.24% (SD=3.91) Experimental Procedures T he following sequence of experim ental procedures occurred prior to initiating the LT protocol A licensed physical therapist assess ed bila teral upper and lower extremity motor and sensory function in persons with SCI based on the American Spinal Injury Association (ASIA) International Standards for Neurological and Functional Classification of Spinal Cord Injury ( American Spinal Injury Association 2002). This assessment establish ed SCI severity and categorized injuries according to the ASIA Impairment Scale (AIS). A physical therapist also assessed performance on standardized balance assessments, the Berg Balance Scale (BBS) (Berg et al. 1992) and the Dynamic Gait Index (DGI) (Shumway Cook and Woollacott 2001) to provide further clinical descriptions of each individual s ability to balance The following motion analysis collections were conducted pre and post LT for persons with SCI and once for healthy controls. Both participants with SCI and healthy controls had reflective marker balls and rigid body clusters positioned on specified body landm arks to acquire 3D motion data. Marker positions we re based on the Vicon PlugInGait marker set (modif ied Helen Hayes set). All i ndividuals wore a safety harness attached to an overhead cable and track system that was suspended from the laboratory ceiling. Walking trials lasted a maximum of 30 seconds and were performed over a split b elt instrumented treadmill (Tec machine, Inc.). Walking practice was permitted to become accustomed to walking on the treadmill and to obtain the best possible steady state walki ng speed. When comfortable, data collections commence d Those individuals with SCI perform ed one walking trial during which they were
142 instructed to walk to the best of their abilities at their self selected (SS) t readmill speed. All walked without assistive devices or braces. In addition, although participants with SCI wore a safety harness for each trial should they stumble or fall, neither BWS nor manual assistance was provided during any trial. This testing condition was intended to capture true walking capacity post injury Either sitting o r standing rest periods were provided as needed between bouts of activity. Additionally, healthy controls performed three s eparate walking trials at randomly introduced treadmill speeds : 0.3, 0.6, and 0.9 m/s. These speeds were selected to represent a likely range within which persons with SCI would elect to walk for normal speedmatched comparisons Data Acquisition and Processing Twelve camera passive motion analysis (Vicon Motion Systems) and 3D grou nd rea ction forces (GRFs) from four piezoelectric force transducers (Advanced Medical Technology, Inc.) located beneath each half of the treadmill were acquired continuously during walking trials. The split be lt treadmill system collected GRFs for each stance ph ase over multiple steps of the gait cycle. Raw kinematic data w ere collected at 100 Hz, then low pass filt ered using a fourth order, zero lag Butterworth filter with a 6 Hz cut off frequ ency. GRFs were acquired at a sampling rate of 2000 Hz, and low pass f iltered using a fourthorder, zero lag Butterworth filter wi th a 20 Hz cut off frequency. A 13segment musculoskeletal model was created using Visual 3D (V3D) processing that fit the model to marker trajectories. Anthropometric and inertial values defined within V3D w ere applied for segment modeling and segmental center of mass (CoM) calculations. V3D models were used to conduct inverse dynamics analyses for calculation of intersegmental joint kinetics. Custom Matlab programming (Mathworks, Inc.) was develo ped to calculate the outcome measures described below.
143 Dynamic Stability Biomechanical Outcome Measures Primary outcome Center of mass trajectory length. Motion of the whole body CoM particularly in the transverse plane, is an indicator of dynamic stabili ty during walking as the CoM must be controlled within a continually moving base of support or a fall will result (Winter 1995) Normal CoM movement over several gait cycles follows a smooth, sinusoidal path that varies minimally from one stride to the nex t. Thus, the mean length of this path over a series of strides as well as its variability from stride to stride relative to normal values provides a gauge of how well an individual is controlling his/her CoM For each st ride in a walking trial, the length of the whole body C oM trajectory in the transverse plane was calculated and normalized to stride length. Secondary outcomes Trunk excursions. Rotational movements of the trunk have been used to quantify balance strategies in elderly with and without balance deficits in response to walking perturbations (Grabiner et al. 2008). Similarly, in this study 3D trunk angular displacement ranges were det ermined for each step in a walking trial. Trunk displacements were defined as the angl es of rotation about each axis. The established reference axes we re positioned orthogonal to one another with the x axis directed mediolaterally, the y axis anteroposteri orly, and the z axis vertically. Proximal and distal ends of the modeled trunk segment were used to calculate movement about the x and y axes, while markers positioned on bilateral acromion processes were used to calcula te rotation about the z axis. Both mean displacement ranges and variability over a walking trial were calculated in each direction.
144 Spatial foot p arameters. Two spatial parameters were calculated separately for each step: step width and step length. Step width was defined as the mediolatera l distance between the heel marker of the leading foot at initial contact and the heel marker of the trailing limb at that same point in time. Step length was defined as the anteroposterior distance between the heel marker of the leading foot at initial co ntact and the heel marker of the trailing limb at that same point in time (Perry 1992) Both means and variability of each parameter over a walking trial were calculated. Data Analysis All statistical analyses were conducte d using SAS 9.13 software. Outcomes for each person with SCI were standardiz ed using standard differences from the control group walking at a speed similar to each persons self selected speed. Thus, the cont rol group mean equaled zero, varian ce equaled one, and SCI values beyond 2 standardardized scores were considered out liers Speed matching was established as follows: SCI 0.3 m/s matched to controls at 0.3 m/s; SCI >0.3 m/s to matched to controls at 0.6 m/s; and SCI > 0.6 m/s to controls at 0.9 m/s. V isual analysis of standardized data demonstrated heterogeneity in the direction of outcome measures relative to controls. Therefore, i n order to avoid masking true differe nces in the SCI sample and regress ing data toward a mean value which might demonstrate no difference from controls a bsolute values were calculated to determine a participants distance from a central point of zero. The mean of all participants absolute values were used for group comparisons using a permutation test Alpha level for significance was set at .05.
145 Results Pre Locomotor Training Biomechanical Outcomes M ean m agnitude of outcomes. When self selecting treadmill speeds for biomechanical testing prior to intervention, all participants elected to walk at extremely slow speeds with comparable speeds observed in both groups (RLT: mean= 0.23 m/s SD=0.1 ; MLT: mean=0.25 m/s SD=0.2 ) (Table 62 ) Although we had hypothesized both groups overall would demonstrate differences from controls in all outcomes, the RLT group was significantly different o nly in CoM trajectory length and trunk rotation ( p 0.05) However, all outcomes were significantly different from controls in the MLT group ( p 0.01) with the except ion of trunk rotation ( p =0.111) (Table 63) Variability of o utcomes. All outcom e measures except trunk rotation and step length in the RLT group showed significantly different variability from controls ( p 0.0 5) The MLT group demonstrated significantly different variability in all outcome measures before training ( p 0.01) (Table 64) Post Locomotor Training Biomechanical Outcomes Mean m agnitude of outcomes. Post LT, all participants in the RLT group and three out of five in the MLT group elected to walk at higher treadmill speeds than during pre testing; on the whole, both groups improved speed similar ly (RLT =+0.25 m/s; MLT ) (Tabl e 6 2 ) In the RLT group, CoM trajectory length was no longer significa ntly different from controls indicating an overall shift to within a normal distribution as originally hypothesized ( p =0.09 ) yet trunk rotation remained sign i f i cantly different ( p 0.0 1 ) M oreover, m ediolateral trunk excursions step length, and step width altered to become significantly different from controls ( p 0.0 5 ) even though these measures indicat ed normal movements prior to training. As anticipated, t he MLT group
146 exhibit ed significant differences from controls in all outcomes post in tervention ( p 0.01) which includes a shift in trunk rotation outside of the normal distribution, although it had been normal pretraining (Table 63) Figure 61 illustrates individual prepost data comparisons for all participants Althoug h group analyses show overall deviations from a central point of zero, the individual data depict the heterogeneity of responses within the sample including directional and magnitude differences. Variability of o utcomes. Contrary to our hypotheses, t he RLT group post training remained significa ntly different in variability from controls in the same outcomes as pre RLT ( p 0.0 001) except trunk rotation still show ed no difference ( p = 0.988) Conversely step length variability changed such that it bec ame significantly different ( p 0.0 001) even though it was not prior to RLT As expected, t he MLT group continued to display significantly different levels of variability from controls in all outcomes ( p 0.0 001) (Table 6 4) Figure 6 2 highlights the individua l participant changes in variability following either RLT or MLT. Discussion This study is the first to examine the impact of LT environments on dynamic stability recovery post SCI. Restitution of walking function for those with incomplete SCI as opposed to sole utilization of spared musculature is consistent with our current rehabilitation paradigm shift. To d ate, the activity based LT literature usi ng manual assisted and/or robot ic assisted environments for humans with SCI has emphasized prima rily the generation of stepping patterns or overground walking with customary assistive devices as outcomes (Behrman and Harkema 2000; Field Fote and Tepavac 2002; Wirz et al. 2005; Nooijen et al. 2009) O ur study expanded the focus of walking recovery in this population to address dynamic stability as a nother essential walking
147 subtask (Barbeau 2003) This subtask has been studied only i n animal models of SCI (Howland et al. 1995; Bolton and Misiaszek 2009; Karayannidou et al. 2009) a nd in healthy and other human patient populations such as persons with lower e xtremity amputations and peripheral neuropathy (Dingwell et al. 2000; Hof et al. 2007) Additionally w e chose to evaluate dynamic stability during walking in a novel manner on a treadmill Evaluation in th is environment paralleled the task spec ific recovery based interventions implemented. This aspect of the evaluation was intended to detect more purely what each intervention specifically trained (i.e. if particular aspects of stability that were trained in each environment translated to walking without assistance). This is in contrast to th e testing in the overground environment with assistive devices, which was not component of the training. R emoval of all assistance including BWS and devices allowed individuals to reveal the ways in which their nervous systems solved the problem of dynam ic stability. MLT and RLT Both Trained Dynamic Stability P ost SCI The effectiveness of each persons solution to dynamic stability manifests in the control of his/her CoM, i.e. the balance point of the body (Winter 1995). The refore, the improved or maintained control of the CoM trajectory lengths relative to normal in eight of ten SCI participants suggests that both the manual assisted and robotic assisted environments have benefits for training dynamic stability. The significant differences from nor mal values that remained in the MLT group as a whole may be a result of the more impaired CoM control that this group presented with preLT. In addition to pre training deficits the number of training sessions in which participants engaged may have impact ed the degree of change in CoM control Currently, the optimal parameters for LT, including duration and frequency, are unknown and vary widely among studies
148 (Field Fote and Tepavac 2002; Hidler et al. 2009) T he training protocol for this study was established at 45 training sessions If persons in the MLT group possessed greater impairments initially but their trends toward improvement were too small to be considered normal, the number of sessions may have been the limiting factor. Adaptive Movement Strategies Developed to Maintain Dynamic Stability Post LT Secondary outcomes of trunk excursions and s patial par ameters demonstrated heterogeneity among and within groups as well; in many cases individuals who initially showed normal values shifted beyond normal post LT, even if their CoM trajectory exhibited improvements. Post hoc video analysis confirmed the apparent generation of new adaptive balance strategies and exaggerated movements (e.g. increased arm swing resulting in increased trunk rotation) Our laboratory previously observed this development of increased arm swing in persons with SCI at slow treadmill speeds immediately post M LT and several months beyond, even though healthy controls possessed much lower amplitude arm swing at the same speeds (NJ Tester, unpublished data). Thus our current findings suggest that amplified trunk rotation may be propagated by increased arm swing (or vice versa). Literature in healthy individuals supports this linkage between arm swing and upper trunk rotation which move concomitantly to counteract pelvic rot ation when walking speeds become increasingly normal (van Emmerik and Wagenaar 1996; Wagenaar and van Emmerik 2000) Moreover, Pijnappels et al. (2010) reported enhanced asymmetric arm swinging with associated increases in trunk rotation as a recovery strategy when healthy subjects were tripped during walking. Interestingly, the most dramatic representation of excessive trunk rotation was evident in four of five persons trained in the MLT environment (SCI 6, 7, 9, 10) with the fifth individual in the group (SCI 8) demonstrating
149 a trend toward the same pattern. Unlike the RLT environment that hinders arm swing with the bulk of its exoskeleton, MLT promotes arm swing. A t times trainers provide arm swing deliberately using parallel poles to assist the alternating arm motion. Thus, the arm swing emphasized in MLT may be responsible for the pro minent trunk rotation observed. One additional, potential e xplanation for those individuals who displayed excessive trunk motion as well as increased variability among step lengths and step widths is that p ost testing often was the first op portunity t o attempt walking without BWS or a device. Up until this point, with the exception of pre testing conditions, BWS was provided during training or customary assistive devices were used outside of the study. Both afford external, supportive influences (Melis et al. 1999) and ultimately were remo ved post testing. This tes ting condition destabilized the body, which then had to achieve stabilization using strategies such as increased trunk movements and/or altered foot placements An intriguing factor is that the majority of participants from both groups also demonstrated a preference for higher treadmill speeds at post testing, likely influenced by walking at closer to normal speeds during training in a similar treadmill environment However, most individuals never had the opportunity to practice walking at those higher speeds without BWS or devices Thus, the exaggerated trunk motion and variability in stepping parameters perhaps reflects the persons immediate balance responses to manage their increase in functional capacity and prefe rence for faster walking speeds The var iability of step length and width among individuals in both training groups require s additional attention based on their pattern of outcome s. Some individuals in the manual assisted group, who demonstrated values outside of normal pre LT shifted in
150 the direction of normal; however, several individuals in the RLT group who showed more normal pre training values became more variable after training Improvements in degree of variability in the MLT group suggest retention and motor learning, which the va riability of this specific intervention is believed to promote. Furthermore, MLT allowed for occasional bout s of independence in which the speed was decreased, BWS increased if needed, and body segments one at a time permitted to move w ithout trainer assistance (although other body segments continued to have assistance). In contrast, although BWS and speed were altered in RLT the computerized exoskeleton provided a relatively error free experience, lacking opportunities to adapt to errors in foot placement. These findings are consistent with recent literature in the stroke population that also compared the manual assisted and robotic assisted interventions (Lewek et al. 2009) Lewek et al. reported improved coordination and kinematic consistency at the hip and knee following four w eeks of training in the manual assisted but not in the robotic assisted environment. In addition, similar studies have been con ducted in the animal literature. Cai et al. (2006) found that spinal mice produced more consistent stepping patterns following robotic assisted trai ning with an assist as needed capability than with robotic assisted training using fixed stepping trajectories. Furthermore, Edgerton and Roy (2009) indicate that after minutes of repetitive stepping along the same path with no vari ation, the nervous system becomes less responsive to sensory stimuli for stepping and muscle activation dimin i shes. AIS Categorization Does Not Reflect Dynamic Stability Of final note are the variations in dynamic stability outcomes in consideration of part icipants AIS categorizations. Following a physical therapist s assessment of upper and lower extremity sensory and motor function, each person included in this study had
151 SCIs classif ied as AIS D and all with lower extremity motor scores 38 out of a possible 50. Previous liter ature regarding the AIS suggested that this classification system and specifically the motor score, correlated well with walking function post SCI (Waters et al. 1994) Thus, individuals with injuries categorized as AIS D purportedly have the highest degree of walking ability I n this study, both primary and secondary measures of dynamic stability during walking revealed great heterogeneity, not only following interventions, but also prior to training. Pre LT, p articipants i n the MLT group were significantly different from cont rols in five measures of mean magnitude and all measures of variability while the RLT group was only different in two measures of mean magnitude and four measures of variability. Therefore, clinicians and researchers should reflect upon the value of grouping all persons together solely with this classification system w hen evaluating dynamic stability. Tests of isolated, muscle strength in a supine may not fully characterize the requirements necessary for dynamic stability in a weight bearing position. Alternative means of stratifying individuals on this walking subtask perhaps via standardized balance assessments, may be necessary ultimately to direct clinicians toward the most e ffective intervention strategy. Conclusions Bo th MLT and RLT demonstrated benefits for training dynamic stability post SCI with the ma jority of participants exhibiting increased CoM control. However, the evolution of adaptive strategies, resulting at the trunk and feet, differed between interventions and within participants. Although values for trunk motion and spatial parameters at the feet often shifted outside of normal ranges post LT, even if individuals demonstrated normal values pretraining, they continued to increase in walking speed overall. T hus, the seemingly abnormal movements may have been solutions to
152 maintain stability as speed capacity increased. Future work should investigate methods of categorizi ng persons post with SCI in alternative ways beyond AIS prior to examining dynamic stabil ity. Additionally, investigations into optimal training parameters, such as duration or frequency, for achieving dynamic stability in either the MLT or RLT environment would be beneficial in directing both researchers and clinicians
153 Table 6 1. Pa rticipant demographics separated by intervention group RLT: Robotic Locomotor Training, MLT: Manual Locomotor Training, LEMS: Lower Extremity Motor Score, AIS : American Spinal Injury Association Impairment Scale, RW: Rolling Walker, SPC: Single Point Cane Participant Training group Age (yrs) Sex Injury site Time post SCI (mos) Assistive device LEMS (max:50) AIS BBS (max:56) DGI (max:24) SCI1 RLT 45 M C5 6 10 RW 43 D 46 17 SCI2 RLT 55 M C4 45 SPC 45 D 31 14 SCI3 RLT 26 M T3 4 11 RW 40 D 21 15 SCI4 RLT 47 F C4 6.5 RW 43 D 19 12 SCI5 RLT 51 F C4 5 7.5 SPC 45 D 42 14 Mean (SD) 44.8 (11.2) 16.0 (16.3) 43.2 (2.0) 31.8 ( 12.1) 14.4 (1.8) SCI6 MLT 48 F C5 25.5 SPC 46 D 51 12 SCI7 MLT 66 M C7 78 SPC 49 D 48 17 SCI8 MLT 40 F T2 3 11 RW 38 D 10 9 SCI9 MLT 21 M C6 7 RW 45 D 17 8 SCI10 MLT 27 M T6 12 RW 40 D 12 11 Mean (SD) 40.4 (17.8) 26.7 (29.5) 43.6 (4.5) 27.6 (20.2) 11.4 (3.5)
154 Table 6 2. SCI self selected treadmill speeds before and after intervention RLT: Robotic -assisted Locomotor Training, MLT: Manual -assisted Locomotor Training Participant Training group Treadmill speed (m/s) Pre Post SCI1 RLT 0.3 0.63 SCI2 RLT 0.25 0.4 SCI3 RLT 0.15 0.25 SCI4 RLT 0.12 0.3 SCI5 RLT 0.25 0.7 Mean (SD) 0.21 (0.1) 0.46 (0.2) SCI6 MLT 0.3 0.9 SCI7 MLT 0.5 0.9 SCI8 MLT 0.2 0.2 SCI9 MLT 0.2 0.15 SCI 10 MLT 0.03 0.15 Mean (SD) 0.25 (0.2) 0.46 (0.4)
155 T able 6 3. Magnitude of standardized outcomes separated by intervention Training Group Testing Session Variable Mean SD Mean(ABS) p value RLT Pre CoM trajectory length 0.633 1.518 1.383 0.027 Post 0.144 1.606 1.606 0.091 Pre Trunk ML 0.712 1.366 1.059 0.166 Post 2.258 1.638 2.258 <0.0001 Pre Trunk AP 0.100 0.770 0.584 0.771 Post 0.705 0.733 0.876 0.364 Pre Trunk Rotation 0.537 2.063 1.810 0.002 Post 0.541 2.294 1.827 *0.001 Pre Step length 0.028 1.207 0.914 0.313 Post 0.621 1.705 1.441 0.018 Pre Step width 0.413 1.200 0.998 0.220 Post 0.619 2.760 2.310 <0.0001 MLT Pre CoM trajectory length 1.777 1.606 1.777 0.002 Post 1.514 1.116 1.514 0.010 Pre Trunk ML 1.566 1.516 1.566 0.007 Post 4.426 2.189 4.426 <0.0001 Pre Trunk AP 1.385 1.243 1.667 0.004 Post 1.902 2.12 2.065 0.0001 Pre Trunk Rotation 1.125 0.988 1.136 0.111 Post 2.855 0.917 2.855 <0.0001 Pre Step length 1.409 2.593 1.869 0.001 Post -0.501 2.503 1.650 0.004 Pre Step width 1.398 4.318 3.966 <0.0001 Post 1.379 3.897 2.876 <0.0001 RLT: Robotic -assisted Locomotor Training, MLT: Manual -assisted Locomotor Training CoM: center -of -mass, ML: mediolateral, AP: anteroposterior, SD: standard deviation, Mean(ABS): mean of absolute values representing overall distance from control mean of zero. (*) indicates significantly different from controls at p
156 Table 6 4. Variability of standardiz ed outcomes separated by intervention Training Group Testing Session Variable Mean SD Mean(ABS) p value RLT Pre CoM trajectory length 1.662 1.990 1.772 0.002 Post 4.256 3.554 4.256 <0.0001 Pre Trunk ML 0.611 2.084 1.426 0.021 Post 3.524 3.497 3.653 <0.0001 Pre Trunk AP 1.136 2.870 2.166 <0.0001 Post 1.990 1.803 1.990 <0.0001 Pre Trunk Rotation 0.416 0.926 0.723 0.582 Post 0.107 0.404 0.286 0.988 Pre Step length 0.618 0.790 0.772 0.514 Post 2.278 2.830 2.374 <0.0001 Pre Step width 2.927 2.011 2.927 <0.0001 Post 3.315 2.248 3.490 <0.0001 MLT Pre CoM trajectory length 6.498 9.169 7.756 <0.0001 Post 3.688 3.853 3.779 <0.0001 Pre Trunk ML 3.658 3.179 3.718 <0.0001 Post 4.379 4.917 4.410 <0.0001 Pre Trunk AP 5.749 6.412 5.749 <0.0001 Post 6.908 8.433 7.552 <0.0001 Pre Trunk Rotation 1.752 1.655 1.768 0.002 Post 2.242 1.955 2.293 <0.0001 Pre Step length 4.728 5.334 4.728 <0.0001 Post 2.503 3.759 2.528 <0.0001 Pre Step width 7.319 8.503 7.632 <0.0001 Post 4.507 7.008 4.891 <0.0001 RLT: Robotic -assisted Locomotor Training, MLT: Manual -assisted Locomotor Training, CoM: center -of -mass, ML: mediolateral, AP: anteroposterior, SD: standard deviation, Mean(ABS): mean of absolute values represent ing overall distance from control mean of zero. (*) indicates significantly different from controls at p
157 Figure 61. Mean o utcome magnitudes standardized to control data pre vs. post locomotor training for 10 SCI participants (separated by training group) Vertical blue shading at 2 standardized scores indicates normal distribution about a c ontrol mean of zero. Shorter Longer Narrower Wider Decrease Increase Decrease Increase Decrease Increase Decrease Increase
158 Figure 62. Variability of outcome measures standardized to control data pre vs. post locomotor training for 10 SCI participant s (separated by training group). Vertical blue shading at 2 standardized scores indicates normal distribution about a control mean of zero. Decrease Increase Decrease Increase
159 CHAPTER 7 CONCLUSIONS Experimental Limitations When examining the results of these three experiments, limitations in met hodology should be considered. First, participants with SCI exhibited a continuum of stepping ability; particularly when walking without assistive devices, which was a condition th at occurred in each experiment. Consequently, participants produced va rying numbers of steps for analysis. Furthermore, during pre analysis data processing certain steps were removed secondary to noise in the acceleration profiles. Ultimately, the number of steps included in analysis therefore was limited for some participan ts. This may have had an impact on the standard deviations used to determine variability during a walking trial. Another avenue for increasing the number of available steps might be to eliminate only certain phases of the gait cycle with unusable data whil e retaining other phases. A second limitation was the manual selection of head and pelvis acceleration thresholds as a method to eliminate spikes or noise in the data and to select only valid steps for analysis. The potential for human error existed in t his process such that the thresholds may have been too low or high, thus inadvertently eliminating valid steps or keeping steps which should have been removed. However, only one individual conducted the manual selection, thereby creating relative consisten cy in the process. A third potential confound across all three experiments was the use of a harness during walking trials. In these studies, which evaluated dynamic stability, the harness may be viewed as an assistive device even though no bodyweight s up port was given at any point. The harness surrounding the trunk and pelvis provided sensory input to the
160 nervous system, and the possibility exists that this input may have influenced stability. Given the safety issues and fall risks of individuals with SCI walking without their customary assistive devices, impleme nting a harness was necessary. However, healthy individuals across these experiments also walked in harnesses, which created a better controlled environment for comparison among groups. Overall Conclusions Spinal cord injury (SCI) impairs dynamic stability during walking. As a result, individuals who sustain SCIs frequently fall and are prone to secondary injuries (Brotherton et al. 2007). Clinicians treating this population conventionally instruct patients on the use of assistive devices and residual motor function to compensate for dynamic stability deficits. More recently, however, SCI rehabilitation has been shifting away from this traditional approach toward one that encourages the use of weak, and previously believed unusable, musculature for the restitution of walking (Behrman and Harkema 2007). However, this transition in interventions also necessitates a concomitant shift in measurement strategies. Thus, the main objective of this dissertati on was to system at ically examine dynamic stability in persons post SCI via a series of experiments, which reflect the SCI rehabilitation paradigm shift. Embedded within these experiments were measurement tools that quantify biomechanical movement strategies, deemed by existing literature as nervous system priorities, essential for maintaining stability. The first experiment investigated both the influence of assistive devices (ADs) on head stability during walking post SCI as well as the effect of injury on head stability when ADs were eliminated. These two aspects of the experiment reflect functionally compensated and uncompensated evaluations. Although results demonstrated a large
161 degree of heterogeneity among participants with SCI, they suggest that ADs pla y a role in reducing mediolateral head motion. Remarkably, when ADs were removed, individuals displayed the ability to decrease anteroposterior head motion to the extent that it was moving less than the pelvis overall. Stability in the AP direction appeared to be a priority. Furthermore, when compared to healthy individuals, those with SCI walking without ADs exhibited highly variable head displacements in all directions. This lack of consistent movement relative to normal levels of variability is purported as an indicator of impaired dynamic stability. In the second experiment, dynamic stability during walking was examined in persons post SCI walking without ADs using three measurement strategies: spatial foot parameters (i.e. step width and step length), a nteroposterior and mediolateral foot placements relative to the pelvic center of mass (CoM), and margin of stability, which accounts for CoM velocity. Findings indicated that over a walking trial, persons with SCI produce greater variability in all measure ments than healthy individuals. Moreover, within groups (SCI and controls), measures demonstrated different amounts of variability suggesting that these outcomes provide different kinds of information about dynamic stability, regardless of whether one has had an injury or not. Lastly, correlations between variability of spatial foot parameters and measures involving the CoM were strong and positive. These relationships may have revealed not only the continuous stepby step adaptations that persons with SCI generate to avoid falling, but also the potential for spatial foot parameters to function as clinical correlates. The final experiment remained within the paradigm shift toward recovery by examining the effects of activity based walking interventions on persons post SCI.
162 Specifically, we studied the differential effects of manual assisted and robotic assisted locomotor training (MLT and RLT) on dynamic stability after SCI to determine how interventions altered balance abilities relative to normal. Results suggested that both MLT and RLT possess training features that positively impact dynamic stability as quantified by the length of the CoM trajectory (i.e. indicative of CoM control). However, responses among individuals were quite variable, even though all were categorized by sensorimotor impairment in the same way (AIS D) prior to training. The most compelling findings were the evolution of movement strategies, such as trunk rotation, that increased beyond normal values, particularly after MLT. These movem ents appeared to be the result of motions including arm swing and pelvic rotation, which are emphasized in the MLT environment. Regardless of intervention approach, individuals post SCI demonstrated newly generated adaptive balance strategies which appear consistent with each environments training properties and that further allowed for simultaneous increases in treadmill speed. Summary and Future Work F ollowing SCI, a subset of the population demonstrates the ability to maintain dynamic stability during w alking when assistive devices are removed. However, the manner in which they move may differ from normal and also may differ from other persons with SCI. The three experiments presented here illustrate the variety of solutions individuals generate at the h ead, trunk, pelvis and feet to avoid falling. Since the same individuals were included as participants across all three studies, a collective picture of each persons balance ability can be gained from the different measurement tools implemented. Another a venue for analysis with this current data set would be to examine the relationships between biomechanical outcome measu res from these
163 experiments and clinical measures of balance /falls (i.e. Berg Balance Scale and Dynamic Gait Index), which were collected as well. Future studies should continue to evaluate measures of dynamic stability, including t hose utilized in these studies, in larger samples of individuals with SCI. In doing so, alternative classification strategies for participants may need to be cons idered beyond AIS classification only. Additional measures of dynamic stability might include a biomechanical evaluation of the upper extremities, muscle synergies in the upper and lower limbs, or neurophysiological testing, such as H reflexes, to understa nd their roles in dynamic stability with this population. Additionally, small accelerometers that attach to body segments might be useful tools to explore dynamic stability in the laboratory and be feasible for use in the clinic as well, thus easing transl ation of information from the literature to the clinic. Investigations should continue to evaluate dynamic stability post SCI within the framework of activity based therapies and restitution of preinjury walking. The continuation of treadmill based evaluations would be consistent with this framework and would allow clinicians to both examine and train a patient in a single environment while gaining essential information about an individuals progress toward dynamic stability recovery during walking.
164 REFE RENCES Allum JH, Carpenter MG (2005) A speedy solution for balance and gait analysis: angular velocity measured at the centre of body mass. Curr Opin Neurol 18: 1521 Anderson KD (2004) Targeting recovery: priorities of the spinal cord injured population. J Neurotrauma 21: 13711383 American Spinal Injury Association (2002) American Spinal Injury Association: International Standards for Neurological Classification of Spinal Cord Injuy. Revised 2002, Chicago, IL Atrice MB, Morrison, SA, McDowell, SL, Ackerman, PM, Foy, TA (2005) Traumatic Spinal Cord Injury. In: Umphred DA (ed) Neurological Rehabilitation. Mosby Inc., St.Louis, pp 605657 Bakker M, Allum JH, Visser JE, Gruneberg C, van de Warrenburg BP, Kremer BH, Bloem BR (2006) Post ural responses to multidirectional stance perturbations in cerebellar ataxia. Exp Neurol 202: 2135 Balasubramanian CK, Neptune RR, Kautz SA (2010) Foot placement in a body reference frame during walking and its relationship to hemiparetic walking performa n ce. Clin Biomech: in press Barbeau H (2003) Locomotor training in neurorehabilitation: emerging rehabilitation concepts. Neur orehabil Neural Repair 17: 311 Barbeau H, Fung J (2001) The role of rehabilitation in the recovery of walking in the neurological population. Curr Opin Neurol 14: 735740 Barbeau H, Ladouceur M, Norman KE, Pepin A, Leroux A (1999) Walking after spinal cord injury: evaluation, treatment, and functional recovery. Arch Phys Med Rehabil 80: 225235 Barbeau H, Nadeau S, Garneau C (2006) Physical determinants, emerging concepts, and training approaches in gait of individuals with spinal cord injury. J Neurotrauma 23: 571585 Barbeau H, Rossignol S (1987) Recovery of locomotion after chronic spinalization in the adult cat. Brain Res 412: 8495 Bastian AJ (2006) Learning to predict the future: the cerebellum adapts feedforward movement control. Curr Opin Neurobiol 16: 645649 Bateni H, Maki BE (2005) Assistive devices for balance and mobility: benefits, demands, and adverse conseq uences. Arch Phys Med Rehabil 86: 134145
165 Behrman AL, Bowden MG, Nair PM (2006) Neuroplasticity after spinal cord injury and training: an emerging paradigm shift in rehabilitation and walking recovery. Phys Ther 86: 14061425 Behrman AL, Harkema SJ (2000) Locomotor training after human spinal cord injury: a series of case studies. Phys Ther 80: 688700 Behrman AL, Harkema SJ (2007) Physical rehabilitation as an agent for recovery after spinal cord injury. Phys Med Rehabil Clin N Am 18: 183202 Behrman AL, L awless Dixon AR, Davis SB, Bowden MG, Nair P, Phadke C, Hannold EM, Plummer P, Harkema SJ (2005) Locomotor training progression and outcomes after incomplete spinal cord injury. Phys Ther 85: 13561371 Behrman AL, Nair PM, Bowden MG, Dauser RC, Herget BR, Martin JB, Phadke CP, Reier PJ, Senesac CR, Thompson FJ, Howland DR (2008) Locomotor Training Restores Walking in a Nonambulatory Child With Chronic, Severe, Incomplete Cervical Spinal Cord Injury. Phys Ther 88: 580 590 Bekoff A, Sabichi AL (1987) Sensory control of the initiation of hatching in chicks: effects of a local anesthetic injected into the neck. Dev Psychobiol 20: 489495 Beloozerova IN, Sirota MG, Orlovsky GN, Deliagina TG (2005) Activity of pyramidal tract neurons in the cat during postural co rrections. J Neurophysiol 93: 18311844 Beres Jones JA, Johnson TD, Harkema SJ (2003) Clonus after human spinal cord injury cannot be attributed solely to recurrent muscle tendon stretch. Exp Brain Res 149: 222236 Berg K, WoodDauphinee S, Williams J, Gay ton D (1989) Measuring balance in the elderly: preliminary development of an instrument. Physiotherapy Canada 41: 304311 Berg KO, WoodDauphinee SL, Williams JI, Maki B (1992) Measuring balance in the elderly: validation of an instrument. Can J Public Health 83 Suppl 2: S711 Berg WP, Alessio HM, Mills EM, Tong C (1997) Circumstances and consequences of falls in independent community dwelling older adults. Age Ageing 26: 261268 Bernhard M, Gries A, Kremer P, Bottiger BW (2005) Spinal cord injury (SCI) -pr ehospital management. Resuscitation 66: 127139 Bjerkefors A, Carpenter MG, Cresswell AG, Thorstensson A (2009) Trunk muscle activation in a person with clinically complete thoracic spinal cord injury. J Rehabil Med 41: 390392
166 Bolton DA, Misiaszek JE (2009) Contribution of hindpaw cutaneous inputs to the control of lateral stability during walking in the cat. J Neurophysiol 102: 17111724 Borghuis J, Hof AL, Lemmink KA (2008) The importance of sensory motor control in providing core stability: implications for measurement and training. Sports Med 38: 893916 Bowden MG, Hannold EM, Nair PM, Fuller LB, Behrman AL (2008) Beyond gait speed: a case report of a multidimensional approach to locomotor rehabilitation outcomes in incomplete spinal cord injury. J Neurol Phys Ther 32: 129138 Brach JS, Berthold R, Craik R, VanSwearingen JM, Newman AB (2001) Gait variability in community dwelling older adults. J Am Geriatr Soc 49: 16461650 Brooks DJ (2001) Functional imaging studies on dopamine and motor control. J Neural Transm 108: 1283 1298 Brotherton SS, Krause JS, Nietert PJ (2007a) Falls in individuals with incomplete spinal cord injury. Spinal Cord 45: 3740 Brotherton SS, Krause JS, Nietert PJ (2007b) A pilot study of factors associated with falls in individu als with incomplete spinal cord injury. J Spinal Cord Med 30: 243250 Burke RE, Degtyarenko AM, Simon ES (2001) Patterns of locomotor drive to motoneurons and last order interneurons: clues to the structure of the CPG. J Neurophysiol 86: 447462 BurleighJacobs A, Horak FB, Nutt JG, Obeso JA (1997) Step initiation in Parkinson's disease: influence of levodopa and external sensory triggers. Mov Disord 12: 206215 Cai LL, Fong AJ, Otoshi CK, Liang Y, Burdick JW, Roy RR, Edgerton VR (2006) Implications of assi st as needed robotic step training after a complete spinal cord injury on intrinsic strategies of motor learning. J Neurosci 26: 1056410568 Calancie B, Needham Shropshire B, Jacobs P, Willer K, Zych G, Green BA (1994) Involuntary stepping after chronic sp inal cord injury. Evidence for a central rhythm generator for locomotion in man. Brain 117 ( Pt 5): 11431159 Cattaneo D, Jonsdottir J, Repetti S (2007) Reliability of four scales on balance disorders in persons with multiple sclerosis. Disabil Rehabil 29: 19201925 Chong RK, Horak FB, Woollacott MH (2000) Parkinson's disease impairs the ability to change set quickly. J Neurol Sci 175: 5770
167 Chou LS, Kaufman KR, Walker Rabatin AE, Brey RH, Basford JR (2004) Dynamic instability during obstacle crossing following traumatic brain injury. Gait Posture 20: 245254 Colombo G, Joerg M, Schreier R, Dietz V (2000) Treadmill training of paraplegic patients using a robotic orthosis. J Rehabil Res Dev 37: 693700. Cordo PJ, Nashner LM (1982) Properties of postural adjus tments associated with rapid arm movements. J Neurophysiol 47: 287302 de Leon RD, Hodgson JA, Roy RR, Edgerton VR (1998) Locomotor capacity attributable to step training versus spontaneous recovery after spinalization in adult cats. J Neurophysiol 79: 1329 1340 Deliagina TG, Beloozerova IN, Zelenin PV, Orlovsky GN (2008) Spinal and supraspinal postural networks. Brain Res Rev 57: 212221 Diener HC, Dichgans J, Bootz F, Bacher M (1984) Early stabilization of human posture after a sudden disturbance: influen ce of rate and amplitude of displacement. Exp Brain Res 56: 126134 Dietz V (2002) Proprioception and locomotor disorders. Nat Rev Neurosci 3: 781790 Dietz V (2003) Spinal cord pattern generators for locomotion. Clin Neurophysiol 114: 13791389 Dietz V (2008) Body weight supported gait training: from laboratory to clinical setting. Brain Res Bull 76: 459463 Dietz V, Colombo G, Jensen L (1994) Locomotor activity in spinal man. Lancet 344: 12601263 Dietz V, Colombo G, Jensen L, Baumgartner L (1995) Locom otor capacity of spinal cord in paraplegic patients. Ann Neurol 37: 574582. Dietz V, Harkema SJ (2004) Locomotor activity in spinal cordinjured persons. J Appl Physiol 96: 19541960 Dietz V, Muller R, Colombo G (2002) Locomotor activity in spinal man: si gnificance of afferent input from joint and load receptors. Brain 125: 26262634 Dietz V, Wirz M, Jensen L (1997) Locomotion in patients with spinal cord injuries. Phys Ther 77: 508516 Dimitrijevic MR, Gerasimenko Y, Pinter MM (1998) Evidence for a spinal central pattern generator in humans. Ann N Y Acad Sci 860: 360376
168 Dimitrijevic MR, Persy I, Forstner C, Kern H, Dimitrijevic MM (2005) Motor control in the human spinal cord. Artif Organs 29: 216219 Dingwell JB, Cusumano JP (2000) Nonlinear time series analysis of normal and pathological human walking. Chaos 10: 848863 Dingwell JB, Cusumano JP, Sternad D, Cavanagh PR (2000) Slower speeds in patients with diabetic neuropathy lead to improved local dynamic stability of continuous overground walking. J Biomech 33: 1269 1277 Ditunno PL, Patrick M, Stineman M, Ditunno JF (2008) Who wants to walk? Preferences for recovery after SCI: a longitudinal and cross sectional study. Spinal Cord 46: 500506 Dobkin B, Apple D, Barbeau H, Basso M, Behrman A, Deforge D, Ditunno J, Dudley G, Elashoff R, Fugate L, Harkema S, Saulino M, Scott M (2006) Weight supported treadmill vs over ground training for walking after acute incomplete SCI. Neurology 66: 484 493 Doyle RJ, Hsiao Wecksler ET, Ragan BG, Rosengren KS (2007) Gen eralizability of center of pressure measures of quiet standing. Gait Posture 25: 166171 Drew T, Prentice S, Schepens B (2004) Cortical and brainstem control of locomotion. Prog Brain Res 143: 251261 Dromerick AW, Lum PS, Hidler J (2006) Activity based th erapies. NeuroRx 3: 428438 Duysens J, Van de Crommert HW (1998) Neural control of locomotion; The central pattern generator from cats to humans. Gait Posture 7: 131141 Edgerton VR, Roy RR (2009) Robotic training and spinal cord plasticity. Brain Res Bull 78: 412 Edgerton VR, Tillakaratne NJ, Bigbee AJ, de Leon RD, Roy RR (2004) Plasticity of the spinal neural circuitry after injury. Annu Rev Neurosci 27: 145 167 Edgley SA, Jankowska E, Shefchyk S (1988) Evidence that midlumbar neurones in reflex pathway s from group II afferents are involved in locomotion in the cat. J Physiol 403: 5771 Eidelberg E, Story JL, Meyer BL, Nystel J (1980) Stepping by chronic spinal cats. Exp Brain Res 40: 241246 England SA, Granata KP (2007) The influence of gait speed on l ocal dynamic stability of walking. Gait Posture 25: 172178 Faber MJ, Bosscher RJ, van Wieringen PC (2006) Clinimetric properties of the performance oriented mobility assessment. Phys Ther 86: 944954
169 Field Fote EC, Fluet GG, Schafer SD, Schneider EM, Smit h R, Downey PA, Ruhl CD (2001) The Spinal Cord Injury Functional Ambulation Inventory (SCI FAI). J Rehabil Med 33: 177181 Field Fote EC, Lindley SD, Sherman AL (2005) Locomotor training approaches for individuals with spinal cord injury: a preliminary report of walking related outcomes. J Neurol Phys Ther 29: 127137 Field Fote EC, Tepavac D (2002) Improved intralimb coordination in people with incomplete spinal cord injury following training with body weight support and electrical stimulation. Phys Ther 82: 707715 Forssberg H (1979) Stumbling corrective reaction: a phase dependent compensatory reaction during locomotion. J Neurophysiol 42: 936953 Forssberg H, Grillner S, Halbertsma J (1980a) The locomotion of the low spinal cat. I. Coordination within a hindlimb. Acta Physiol Scand 108: 269281 Forssberg H, Grillner S, Halbertsma J, Rossignol S (1980b) The locomotion of the low spinal cat. II. Interlimb coordination. Acta Physiol Scand 108: 283295 Forssberg H, Grillner S, Rossignol S (1975) Phase dependent reflex reversal during walking in chronic spinal cats. Brain Res 85: 103107 Frank JS, Horak FB, Nutt J (2000) Centrally initiated postural adjustments in parkinsonian patients on and off levodopa. J Neurophysiol 84: 24402448 Galvez JA, Budovitch A, Ha rkema SJ, Reinkensmeyer DJ (2007) Quantification of therapists' manual assistance on the leg during treadmill gait training with partial body weight support after spinal cord injury. Conf Proc IEEE Eng Med Biol Soc 2007: 40284032 Giuliani CA, Smith JL (19 85) Development and characteristics of airstepping in chronic spinal cats. J Neurosci 5: 12761282 Glass L (2001) Synchronization and rhythmic processes in physiology. Nature 410: 277284 Goldie PA, Bach TM, Evans OM (1989) Force platform measures for eval uating postural control: reliability and validity. Arch Phys Med Rehabil 70: 510517 Gorassini MA, Norton JA, Nevett Duchcherer J, Roy FD, Yang JF (2009) Changes in locomotor muscle activity after treadmill training in subjects with incomplete spinal cord injury. J Neurophysiol 101: 969979 Grabiner MD, Donovan S, Bareither ML, Marone JR, Hamstra Wright K, Gatts S, Troy KL (2008) Trunk kinematics and fall risk of older adults: translating biomechanical results to the clinic. J Electromyogr Kinesiol 18: 197204
170 Granata KP, Lockhart TE (2008) Dynamic stability differences in fall prone and healthy adults. J Electromyogr Kinesiol 18: 172178 Grasso R, Zago M, Lacquaniti F (2000) Interactions between posture and locomotion: motor patterns in humans walking with bent posture versus erect posture. J Neurophysiol 83: 288300 Gregory CM, Bowden MG, Jayaraman A, Shah P, Behrman A Kautz SA, Vandenborne K (2007) Resistance training and locomotor recovery after incomplete spinal cord injury: a case series. Spinal Cord 45: 522530 Grillner S, Deliagina T, Ekeberg O, el Manira A, Hill RH, Lansner A, Orlovsky GN, Wallen P (1995) Neural networks that co ordinate locomotion and body orientation in lamprey. Trends Neurosci 18: 270279 Grillner S, Wallen P (1985) Central patter n generators for locomotion, with special reference to vertebrates. Annu Rev Neurosci 8: 233261 Grillner S, Zangger P (1979) On the central generation of locomotion in the low spinal cat. Exp Brain Res 34: 241261 Hahn ME, Chou LS (2003) Can motion of individual body segments identify dynamic instabili ty in the elderly? Clin Biomech 18: 737744 Harkema SJ (2001) Neural plasticity after human spinal cord injury: application of locomotor training to the rehabilitation of walking. Neuroscientist 7: 455468 Harkema SJ, Hurley SL, Patel UK, Requejo PS, Dobkin BH, Edgerton VR (1997) Human lumbosacral spinal cord interprets loading during stepping. J Neurophysiol 77: 797811 Harris Love ML, Forrester LW, Macko RF, Silver KH, Smith GV (2001) Hemiparetic gait parameters in overground versus treadmill walking. Neurorehabil Neural Repair 15: 105112 Hidler J, Nichols D, Pelliccio M, Brady K, Campbell DD, Kahn JH, Hornby TG (2009) Multicenter randomized clinical trial evaluating the effectiveness of the Lokomat in sub acute stroke. Neurorehabil Neural Repair 23: 513 Hodgson JA, Roy RR, de Leon R, Dobkin B, Edgerton VR (1994) Can the mammalian lumbar spinal cord learn a motor task? Med Sci Sports Exerc 26: 14911497 Hof AL (2008) The 'extrapolated center of mass' concept suggests a simple control of balance in walking. Human Movement Science 27: 112125 Hof AL, Gazendam MG, Sinke WE (2005) The condition for dynamic stability. J Biomech 38: 1 8
171 Hof AL, van Bockel RM, Schoppen T, Postema K (2007) Control of lateral balance in walking. Experimental findings in normal subjects and aboveknee amputees. Gait Posture 25: 250258 Holt KG, Ratcliffe R, Jeng SF (1999) Head stability in walking in children with cerebral palsy and in children and adults without neurological impairment. Phys Ther 79: 11531162 Hong M, Earhart GM (2010) Effects of medication on turning deficits in individuals with Parkinson's disease. J Neurol Phys Ther 34: 1116 Horak FB (2006) Postural orientation and equilibrium: what do we need to know about neural control of balance to prevent falls? Age Ageing 35 Suppl 2: ii7ii11 Horak FB, Diener HC (1994) Cerebellar control of postural scaling and central set in stance. J Neurophysiol 72: 479493 Horak FB, Frank J, Nutt J (1996) Effects of dopamine on postural co ntrol in parkinsonian subjects: scaling, set, and tone. J Neurophysiol 75: 23802396 Horak FB, Macpherson JM (1996) Postural orientation and equilibrium. In: Handbook of Physiology and Exercise: regulation and integration of multiple systems. Oxford, New Y ork, pp 255 292 Horak FB, Nashner LM (1986) Central programming of postural movements: adaptation to altered support surface configurations. J Neurophysiol 55: 13691381 Horak FB, Wrisley DM, Frank J (2009) The Balance Evaluation Systems Test (BESTest) t o differentiate balance deficits. Phys Ther 89: 484498 Howland DR, Bregman BS, Tessler A, Goldberger ME (1995) Development of locomotor behavior in the spinal kitten. Exp Neurol 135: 108122 Hughey LK, Fung J (2005) Postural responses triggered by multidi rectional leg lifts and surface tilts. Exp Brain Res 165: 152166 Huxham FE, Goldie PA, Patla AE (2001) Theoretical considerations in balance assessment. Aust J Physiother 47: 89100 Illis LS (1995) Is there a central pattern generator in man? Paraplegia 33: 239240 Ivanenko YP, Poppele RE, Lacquaniti F (2009) Distributed neural networks for controlling human locomotion: lessons from normal and SCI subjects. Brain Res Bull 78: 1321 Jacobs JV, Fujiwara K, Tomita H, Furune N, Kunita K, Horak FB (2008) Changes in the activity of the cerebral cortex relate to postural response modification when warned of a perturbation. Clin Neurophysiol 119: 14311442
172 Jacobs JV, Horak FB (2007a) Cortical control of postural responses. J Neural Transm 114: 13391348 Jacobs JV, Horak FB (2007b) External postural perturbations induce multiple anticipatory postural adjustments when subjects cannot preselect their stepping foot. Exp Brain Res 179: 2942 Jacobs PL, Mahoney ET (2002) Peak exercise capacity of electrically induced amb ulation in persons with paraplegia. Med Sci Sports Exerc 34: 15511556 Jacobs PL, Nash MS (2004) Exercise recommendations for individuals with spinal cord injury. Sports Med 34: 727751 Jayaraman A, Gregory CM, Bowden M, Stevens JE, Shah P, Behrman AL, Van denborne K (2006) Lower extremity skeletal muscle function in persons with incomplete spinal cord injury. Spinal Cord 44: 680687 Jeka JJ (1997) Light touch contact as a balance aid. Phys Ther 77: 476487 Jeka JJ, Lackner JR (1994) Fingertip contact influe nces human postural control. Exp Brain Res 100: 495 502 Jonsdottir J, Cattaneo D (2007) Reliability and validity of the dynamic gait index in persons with chronic stroke. Arch Phys Med Rehabil 88: 14101415 Jonsson E, Seiger A, Hirschfeld H (2004) Oneleg stance in healthy young and elderly adults: a measure of postural steadiness? Clin Biomech (Bristol, Avon) 19: 688694 Jordan LM, Brownstone RM, Noga BR (1992) Control of functional systems in the brainstem and spinal cord. Curr Opin Neurobiol 2: 794 801 Juvin L, Simmers J, Morin D (2007) Locomotor rhythmogenesis in the isolated rat spinal cord: a phase coupled set of symmetrical flexion extension oscillators. J Physiol 583: 115128 Kang HG, Dingwell JB (2006) A direct comparison of local dynamic stability during unperturbed standing and walking. Exp Brain Res 172: 3548 Karayannidou A, Zelenin PV, Orlovsky GN, Sirota MG, Beloozerova IN, Deliagina TG (2009) Maintenance of lateral stability during standing and walking in the cat. J Neurophysiol 101: 8 19 Karc nik T, Kralj A (1999) Stability and velocity in incomplete spinal cord injured subject gaits. Artif Organs 23: 421423 Kavanagh J, Barrett R, Morrison S (2006) The role of the neck and trunk in facilitating head stability during walking. Exp Brain Res 172: 454 463
173 Kavanagh JJ, Barrett RS, Morrison S (2004) Upper body accelerations during walking in healthy young and elderly men. Gait Posture 20: 291298 Kavanagh JJ, Barrett RS, Morrison S (2005a) Agerelated differences in head and trunk coordination during walking. Hum Mov Sci 24: 574587 Kavanagh JJ, Morrison S, Barrett RS (2005b) Coordination of head and trunk accelerations during walking. Eur J Appl Physiol 94: 468475 Kleim JA, Jones TA (2008) Principles of experience dependent neural plasticity: implic ations for rehabilitation after brain damage. J Speech Lang Hear Res 51: S225239 Krause JS (2004) Factors associated with risk for subsequent injuries after traumatic spinal cord injury. Arch Phys Med Rehabil 85: 15031508 Krauss EM, Misiaszek JE (2007) Phase specific modulation of the soleus H reflex as a function of threat to stability during walking. Exp Brain Res 181: 665672 Krebs DE, Goldvasser D, Lockert JD, Portney LG, Gill Body KM (2002) Is base of support greater in unsteady gait? Phys Ther 82: 138147 Krebs DE, Wong D, Jevsevar D, Riley PO, Hodge WA (1992) Trunk kinematics during locomotor activities. Phys Ther 72: 505514 Ladouceur M, Pepin A, Norman KE, Barbeau H (1997) Recovery of walking after spinal cord injury. Adv Neurol 72: 249255 Lam T Wolstenholme C, van der Linden M, Pang MY, Yang JF (2003) Stumbling corrective responses during treadmill elicited stepping in human infants. J Physiol 553: 319331 Larson CA, Tezak WD, Malley MS, Thornton W (2010) Assessment of postural muscle strength in sitting: reliability of measures obtained with handheld dynamometry in individuals with spinal cord injury. J Neurol Phys Ther 34: 2431 Latt LD, Sparto PJ, Furman JM, Redfern MS (2003) The steady state postural response to continuous sinusoidal galvan ic vestibular stimulation. Gait Posture 18: 6472 Laudani L, Casabona A, Perciavalle V, Macaluso A (2006) Control of head stability during gait initiation in young and older women. J Electromyogr Kinesiol 16: 603610 Lee HJ, Chou LS (2006) Detection of gait instability using the center of mass and center of pressure inclination angles. Arch Phys Med Rehabil 87: 569575
174 Lee HK, Cho SP, You JH, Lee KJ (2007) The concurrent validity of the body center of mass in accelerometric measurement. Conf Proc IEEE En g Med Biol Soc 2007: 659661 Leroux A, Fung J, Barbeau H (1999) Adaptation of the walking pattern to uphill walking in normal and spinal cord injured subjects. Exp Brain Res 126: 359368 Leroux A, Fung J, Barbeau H (2002) Postural adaptation to walking on inclined surfaces: I. Normal strategies. Gait Posture 15: 6474 Leroux A, Fung J, Barbeau H (2006) Postural adaptation to walking on inclined surfaces: II. Strategies following spinal cord injury. Clin Neurophysiol 117: 12731282 Levin MF, Kleim JA, Wolf S L (2009) What do motor "recovery" and "compensation" mean in patients following stroke? Neurorehabil Neural Repair 23: 313319 Lewek MD, Cruz TH, Moore JL, Roth HR, Dhaher YY, Hornby TG (2009) Allowing intralimb kinematic variability during locomotor train ing poststroke improves kinematic consistency: a subgroup analysis from a randomized clinical trial. Phys Ther 89: 829839 Liechti M, Muller R, Lam T, Curt A (2008) Vestibulospinal responses in motor incomplete spinal cord injury. Clin Neurophysiol 119: 28042812 Lovely RG, Gregor RJ, Roy RR, Edgerton VR (1986) Effects of training on the recovery of full weight bearing stepping in the adult spinal cat. Exp Neurol 92: 421435 MacKinnon CD, Winter DA (1993) Control of whole body balance in the frontal plane d uring human walking. J Biomech 26: 633 644 MacLellan MJ, Patla AE (2006) Adaptations of walking pattern on a compliant surface to regulate dynamic stability. Exp Brain Res 173: 521530 Macpherson JM, Fung J (1999) Weight support and balance during perturbed stance in the chronic spinal cat. J Neurophysiol 82: 30663081 Macpherson JM, Fung J, Jacobs R (1997) Postural orientation, equilibrium, and the spinal cord. Adv Neurol 72: 227232 Maki BE, McIlroy WE (1996) Postural control in the older adul t. Clin Geri atr Med 12: 635658 Mamoto Y, Yamamoto K, Imai T, Tamura M, Kubo T (2002) Threedimensional analysis of human locomotion in normal subjects and patients with vestibular deficiency. Acta Otolaryngol 122: 495500
175 Marder E, Calabrese RL (1996) Principles of r hythmic motor pattern generation. Physiol Rev 76: 687717 Martin M, Shinberg M, Kuchibhatla M, Ray L, Carollo JJ, Schenkman ML (2002) Gait initiation in community dwelling adults with Parkinson disease: comparison with older and younger adults without the disease. Phys Ther 82: 566577 Matsuyama K, Drew T (2000) Vestibulospinal and reticulospinal neuronal activity during locomotion in the intact cat. II. Walking on an inclined plane. J Neurophysiol 84: 22572276 Mazza C, Iosa M, Pecoraro F, Cappozzo A (2008) Control of the upper body accelerations in young and elderly women during level walking. J Neuroeng Rehabil 5: 30 Mazza C, Iosa M, Picerno P, Cappozzo A (2009) Gender differences in the control of the upper body accelerations during level walking. Gait P osture 29: 300303 McClellan AD, Sigvardt KA (1988) Features of entrainment of spinal pattern generators for locomotor activity in the lamprey spinal cord. J Neurosci 8: 133145 McConvey J, Bennett SE (2005) Reliability of the Dynamic Gait Index in individuals with multiple sclerosis. Arch Phys Med Rehabil 86: 130133 Melis EH, Torres Moreno R, Barbeau H, Lemaire ED (1999) Analysis of assistedgait characteristics in persons with incomplete spinal cord injury. Spinal Cord 37: 430439. Mentel T, Cangiano L, Grillner S, Buschges A (2008) Neuronal substrates for statedependent changes in coordination between motoneuron pools during fictive locomotion in the lamprey spinal cord. J Neurosci 28: 868879 Menz HB, Lord SR, Fitzpatrick RC (2003a) Acceleration patter ns of the head and pelvis when walking are associated with risk of falling in community dwelling older people. J Gerontol A Biol Sci Med Sci 58: M446452 Menz HB, Lord SR, Fitzpatrick RC (2003b) Agerelated differences in walking stability. Age Ageing 32: 137142 Minassian K, Jilge B, Rattay F, Pinter MM, Binder H, Gerstenbrand F, Dimitrijevic MR (2004) Steppinglike movements in humans with complete spinal cord injury induced by epidural stimulation of the lumbar cord: electromyographic study of compound m uscle action potentials. Spinal Cord 42: 401416 Minassian K, Persy I, Rattay F, Pinter MM, Kern H, Dimitrijevic MR (2007) Human lumbar cord circuitries can be activated by extrinsic tonic input to generate locomotor like activity. Hum Mov Sci 26: 275295
176 Misiaszek JE (2006) Neural control of walking balance: if falling then react else continue. Exerc Sport Sci Rev 34: 128134 Mochizuki G, Ivanova TD, Garland SJ (2004) Postural muscle activity during bilateral and unilateral arm movements at different speeds. Exp Brain Res 155: 352361 Mori S (1987) Integration of posture and locomotion in acute decerebrate cats and in awake, freely moving cats. Prog Neurobiol 28: 161195 Mori S, Matsuyama K, Kohyama J, Kobayashi Y, Takakusaki K (1992) Neuronal constituents of postural and locomotor control systems and their interactions in cats. Brain Dev 14 Suppl: S109120 Morton SM, Bastian AJ (2004) Cerebellar control of balance and locomotion. Neuroscientist 10: 247259 Muir GD, Steeves JD (1995) Phasic cutaneous input f acilitates locomotor recovery after incomplete spinal injury in the chick. J Neurophysiol 74: 358368 Mummel P, Timmann D, Krause UW, Boering D, Thilmann AF, Diener HC, Horak FB (1998) Postural responses to changing task conditions in patients with cerebel lar lesions. J Neurol Neurosurg Psychiatry 65: 734742 Musselman KE, Yang JF (2007) Walking tasks encountered by urbandwelling adults and persons with incomplete spinal cord injuries. J Rehabil Med 39: 567574 Nadeau S, Amblard B, Mesure S, Bourbonnais D (2003) Head and trunk stabilization strategies during forward and backward walking in healthy adults. Gait Posture 18: 134142 Nashner LM (1976) Adapting reflexes controlling the human posture. Exp Brain Res 26: 5972 Nashner LM (1980) Balance adjustments of humans perturbed while walking. J Neurophysiol 44: 650664 National Spinal Cord Injury Statistical Center (2008) In. Department of Education, Washington, DC Nielsen JB, Conway BA, Halliday DM, Perreault MC, Hultborn H (2005) Organization of common synaptic drive to motoneurones during fictive locomotion in the spinal cat. J Physiol 569: 291304 Nooijen CF, Ter Hoeve N, FieldFote EC (2009) Gait quality is improved by locomotor training in individuals with SCI regardless of training approach. J Neuroeng R ehabil 6: 36
177 Noreau L, Proulx P, Gagnon L, Drolet M, Laramee MT (2000) Secondary impairments after spinal cord injury: a population based study. Am J Phys Med Rehabil 79: 526535 Oates AR, Frank JS, Patla AE, VanOoteghem K, Horak FB (2008) Control of dynam ic stability during gait termination on a slippery surface in Parkinson's disease. Mov Disord 23: 19771983 Oddsson LI, Wall C, McPartland MD, Krebs DE, Tucker CA (2004) Recovery from perturbations during paced walking. Gait Posture 19: 2434 Pai YC, Patto n J (1997) Center of mass velocity position predictions for balance control. J Biomech 30: 347 354 Parvataneni K, Ploeg L, Olney SJ, Brouwer B (2009) Kinematic, kinetic and metabolic parameters of treadmill versus overground walking in healthy older adults Clin Biomech (Bristol, Avon) 24: 95 100 Patla AE, Prentice SD (1995) The role of active forces and intersegmental dynamics in the control of limb trajectory over obstacles during locomotion in humans. Exp Brain Res 106: 499 504 Patla AE, Prentice SD, Rie tdyk S, Allard F, Martin C (1999) What guides the selection of alternate foot placement during locomotion in humans. Exp Brain Res 128: 441450 Patterson SL, Forrester LW, Rodgers MM, Ryan AS, Ivey FM, Sorkin JD, Macko RF (2007) Determinants of walking fun ction after stroke: differences by deficit severity. Arch Phys Med Rehabil 88: 115 119 Pepin A, Ladouceur M, Barbeau H (2003a) Treadmill walking in incomplete spinal cord injured subjects: 2. Factors limiting the maximal speed. Spinal Cord 41: 271279. Pepin A, Norman KE, Barbeau H (2003b) Treadmill walking in incomplete spinal cord injured subjects: 1. Adaptation to changes in speed. Spinal Cord 41: 257270. Perry J (1992) Gait Analysis : Normal and pathological function. Slack, Thorofare,NJ Perry J, Garre tt M, Gronley JK, Mulroy SJ (1995) Classification of walking handicap in the stroke population. Stroke 26: 982989 Pijnappels M, Kingma I, Wezenberg D, Reurink G, van Dieen JH (2010) Armed against falls: the contribution of arm movements to balance recover y after tripping. Exp Brain Res 201: 689 699 Pozzo T, Berthoz A, Lefort L (1990) Head stabilization during various locomotor tasks in humans. I. Normal subjects. Exp Brain Res 82: 97106
178 Redfern MS, Schumann T (1994) A model of foot placement during gait. J Biomech 27: 13391346 Reinkensmeyer D, Aoyagi D, Emken J, Galvez J, Ichinose W, Kerdanyan G, Nessler J, Maneekobkunwong S, Timoszyk B, Vallance K, Weber R, de Leon R, Bobrow J, Harkema S, Wynne J, Edgerton V (2004) Robotic gait training: toward more natural movements and optimal training algorithms. Conf Proc IEEE Eng Med Biol Soc 7: 48184821 Riolo Quinn L (1999) Principles of Motor Learning. Clinical Management 7: 611 Rochester L, Hetherington V, Jones D, Nieuwboer A, Willems AM, Kwakkel G, Van Wegen E (2004) Attending to the task: interference effects of functional tasks on walking in Parkinson's disease and the roles of cognition, depression, fatigue, and balance. Arch Phys Med Rehabil 85: 15781585 Saha D, Gard S, Fatone S (2008) The effect of trunk flexion on able bodied gait. Gait Posture 27: 653660 Scivoletto G, Romanelli A, Mariotti A, Marinucci D, Tamburella F, Mammone A, Cosentino E, Sterzi S, Molinari M (2008) Clinical factors that affect walking level and performance in chronic spinal cord lesion patients. Spine 33: 259264 Segal AD, Orendurff MS, Czerniecki JM, Shofer JB, Klute GK (2008) Local dynamic stability in turning and straight line gait. J Biomech 41: 14861493 Shah PK, Stevens JE, Gregory CM, Pathare NC, Jayaraman A, Bickel SC, Bowde n M, Behrman AL, Walter GA, Dudley GA, Vandenborne K (2006) Lower extremity muscle cross sectional area after incomplete spinal cord injury. Arch Phys Med Rehabil 87: 772778 Sherrington CS (1910) Flexion reflex of the limb, crossed extension reflex, and r eflex stepping and standing. Journal of Physiology 40: 28121 Sherwood AM, Dimitrijevic MR, McKay WB (1992) Evidence of subclinical brain influence in clinically complete spinal cord injury: discomplete SCI. J Neurol Sci 110: 9098 Shumway Cook A, Horak FB (1986) Assessing the influence of sensory interaction of balance. Suggestion from the field. Phys Ther 66: 15481550 Shumway Cook A, Patla AE, Stewart A, Ferrucci L, Ciol MA, Guralnik JM (2002) Environmental demands associated with community mobility in o lder adults with and without mobility disabilities. Phys Ther 82: 670 681. Shumway Cook A, Woollacott M (2001) Motor Control: Theory and Practical Applications. Lippincott Williams and Wilkins, Philadelphia
179 Slobounov S, Hallett M, Stanhope S, Shibasaki H ( 2005) Role of cerebral cortex in human postural control: an EEG study. Clin Neurophysiol 116: 315323 Smith JL, Smith LA, Zernicke RF, Hoy M (1982) Locomotion in exercised and nonexercised cats cordotomized at two or twelve weeks of age. Exp Neurol 76: 393413 Stolze H, Kuhtz Buschbeck JP, Drucke H, Johnk K, Diercks C, Palmie S, Mehdorn HM, Illert M, Deuschl G (2000) Gait analysis in idiopathic normal pressure hydrocephalus -which parameters respond to the CSF tap test? Clin Neurophysiol 111: 16781686 T akakusaki K, Saitoh K, Harada H, Kashiwayanagi M (2004) Role of basal gangliabrainstem pathways in the control of motor behaviors. Neurosci Res 50: 137151 Tang PF, Woollacott MH, Chong RK (1998) Control of reactive balance adjustments in perturbed human walking: roles of proximal and distal postural muscle activity. Exp Brain Res 119: 141152 Tesio L, Rota V (2008) Gait analysis on split belt force treadmills: validation of an instrument. Am J Phys Med Rehabil 87: 515526 Thigpen MT, Cauraugh J, Creel G, Day K, Flynn S, Fritz S, Frost S, Respess R, Gardner Smith P, Brack M, Behrman A (2009) Adaptation of postural responses during different standing perturbation conditions in individuals with incomplete spinal cord injury. Gait Posture 29: 113118 Timmann D Horak FB (1997) Prediction and set dependent scaling of early postural responses in cerebellar patients. Brain 120 ( Pt 2): 327337 Timmann D, Horak FB (2001) Perturbed step initiation in cerebellar subjects: 2. Modification of anticipatory postural adju stments. Exp Brain Res 141: 110120 Tinetti ME (1986) Performance oriented assessment of mobility problems in elderly patients. J Am Geriatr Soc 34: 119126 Tinetti ME, Williams TF, Mayewski R (1986) Fall risk index for elderly patients based on number of chronic disabilities. American Journal of Medicine 80: 4294344 Ting LH, Macpherson JM (2005) A limited set of muscle synergies for force control during a postural task. J Neurophysiol 93: 609613 Torres Oviedo G, Macpherson JM, Ting LH (2006) Muscle syner gy organization is robust across a variety of postural perturbations. J Neurophysiol 96: 15301546 van de Port IG, Kwakkel G, Lindeman E (2008) Community ambulation in patients with chronic stroke: how is it related to gait speed? J Rehabil Med 40: 2327
180 v an Emmerik RE, van Wegen EE (2002) On the functional aspects of variability in postural control. Exerc Sport Sci Rev 30: 177183 van Emmerik RE, Wagenaar RC (1996) Effects of walking velocity on relative phase dynamics in the trunk in human walking. J Biomech 29: 11751184 van Hedel HJ, Wirth B, Dietz V (2005) Limits of locomotor ability in subjects with a spinal cord injury. Spinal Cord 43: 593603 Vaynman S, Gomez Pinilla F (2005) License to run: exercise impacts functional plasticity in the intact an d injured central nervous system by using neurotrophins. Neurorehabil Neural Repair 19: 283295 Visintin M, Barbeau H (1994) The effects of parallel bars, body weight support and speed on the modulation of the locomotor pattern of spastic paretic gait. A p reliminary communication. Paraplegia 32: 540553 Wagenaar RC, van Emmerik RE (2000) Resonant frequencies of arms and legs identify different walking patterns. J Biomech 33: 853861 Wallen P, Williams TL (1984) Fictive locomotion in the lamprey spinal cord in vitro compared with swimming in the intact and spinal animal. J Physiol 347: 225239 Waters RL, Adkins R, Yakura J, Vigil D (1994) Prediction of ambulatory performance based on motor scores derived from standards of the American Spinal Injury Associatio n. Arch Phys Med Rehabil 75: 756760 Waters RL, Mulroy S (1999) The energy expenditure of normal and pathologic gait. Gait Posture 9: 207231 Webster KE, Wittwer JE, Feller JA (2005) Validity of the GAITRite walkway system for the measurement of averaged a nd individual step parameters of gait. Gait Posture 22: 317321 Wernig A, Nanassy A, Muller S (1999) Laufband (treadmill) therapy in incomplete paraplegia and tetraplegia. J Neurotrauma 16: 719726 Winter DA, Patla AE, Frank JS (1990) Assessment of balance control in humans. Med Prog Technol 16: 3151 Winter DA (1995) Human balance and posture control during standing and walking. Gait Posture 3: 193214 Wirz M, van Hedel HJ, Rupp R, Curt A, Dietz V (2006) Muscle force and gait performance: relationships aft er spinal cord injury. Arch Phys Med Rehabil 87: 12181222
181 Wirz M, Zemon DH, Rupp R, Scheel A, Colombo G, Dietz V, Hornby TG (2005) Effectiveness of automated locomotor training in patients with chronic incomplete spinal cord injury: a multicenter trial. A rch Phys Med Rehabil 86: 672680 Wolpaw JR, Tennissen AM (2001) Activity dependent spinal cord plasticity in health and disease. Annu Rev Neurosci 24: 807843 Woollacott MH, Tang PF (1997) Balance control during walking in the older adult: research and its implications. Phys Ther 77: 646660 Wrisley DM, Walker ML, Echternach JL, Strasnick B (2003) Reliability of the dynamic gait index in people with vestibular disorders. Arch Phys Med Rehabil 84: 15281533 Wydenkeller S, Liechti M, Muller R, Curt A (2006) I mpaired scaling of responses to vestibular stimulation in incomplete SCI. Exp Brain Res 175: 191195 Yamaguchi T (1992) Activity of cervical neurons during forelimb fictive locomotion in decerebrate cats. Jpn J Physiol 42: 501514 Yang JF, Lamont EV, Pang MY (2005) Split belt treadmill stepping in infants suggests autonomous pattern generators for the left and right leg in humans. J Neurosci 25: 68696876 Zajac FE, Neptune RR, Kautz SA (2003) Biomechanics and muscle coordination of human walking. Part II: Lessons from dynamical simulations and clinical impl ications. Gait Posture 17: 117 Zehr EP, Stein RB, Komiyama T (1998) Function of sural nerve reflexes during human walking. J Physiol 507 ( Pt 1): 305314
182 BIOGRAPHICAL SKETCH Kristin Alayne Vamvas Day, a native of Cincinnati, Ohio, graduated from Ohio University in 2000 with a bachelor s degree in biological s ciences. A year prior to earning her undergraduate degree, she was admitted in absentia to Ohio University School of Phy sical Therapy and ultimately graduated with her master s degree in physical t herapy in 2002. Upon passing her national licensure examination, she began practicing as a staff physical therapist in Greenville, South Carolina, at Greenville Memorial Hospital, a level one trauma center. At this facility, she gained an immense passion for the care of patients with neurologically traumatic injuries, mentored by a team of highly experienced interdisci plinary therapists. Kristin returned to Cincinnati in 2004 and c ontinued to practice in acute care and inpatient rehabilitation at Good Samaritan Hospital. A self proclaimed aggressive and progressive physical therapist, Kristin recognized what had evolved as the routine nature of physical therapy practice and furth er realized that the lack of evidence to guide decision making may be largely responsible. Therefore, in January 2006, she returned to graduate school with the goal of advancing the science behind neurological rehabilitation. She enrolled in the Rehabilita tion Science Doctoral program at the University of Florida under the mentorship of Dr. Andrea Behrman, an expert in the field of spinal cord injury re habilitation and recovery. During her four and a half years at Florida, Kristin investigated locomotor int erventions targeted at walking recovery in individuals with spinal cord injury. Furthermore, her dissertation investigated an unexplored area in this patient population, measurement of walking balance recovery. Throughout her doctoral education, Kristin received full financial support through the Neuromuscular Plasticity
183 Training p rogram, funded through a National Institutes of Health T32 training grant, as well as through Department of Physical Therapy teaching assistantships. She additionally received a o ne year Early Career Rehabilitation Research Award from the Florida Department of Health Brain and Spinal Cord Injury Program. Kristin graduated with a Doctor of Philosophy in May 2010, aspiring to translate her experiences in spinal cord injury mobility r esearch as well as her clinical background in traumatic brain injury to explore physical treatments for persons with disorders of consciousness.