Systematic Review of Activity-Based Interventions to Improve Neurological Outcomes after SCI January 1998 – March 2009
Conducted by the Shepherd Center with support from the National Institute on Disability and Rehabilitation Research.
Table of Contents
Plain language summary
Methods and procedures
Results and conclusions
Acknowledgements and statement concerning conflict of interest
The Activity-Based Interventions to Improve Neurological Outcomes systematic review and narrative synthesis involved collecting and grading research information for rigor and meaning. The Shepherd Center Systematic Review Group utilized the system for grading research in terms of rigor and meaning developed by the BU Supported Housing Study Group (Farkas, Rogers and Anthony, 2008). The Shepherd Center Group used the same instruments, the first being: “Standards for Rating Program Evaluation, Policy or Survey Research, Pre-Post and Correlational Human Subjects” (Rogers, Farkas, Anthony & Kash, 2008) and the second: “Standards for Rating the Meaning of Disability Research” (Farkas & Anthony, 2008). The lead author of the Activity-Based Interventions to Improve Neurological Outcomes is Deborah Backus. A ful listing of contributors is contained in the section “Contributors”.
“Activity-based interventions”, or “activity-based therapy”, has become a common term in the spinal cord injury rehabilitation field. The suggested promise of recovery of neural function, and therefore the return of valued functional abilities, such as walking or using ones’ hands, has led health care providers to establish new activity-based programs, without evidence supporting the use of the interventions employed in these programs. Hope for recovery of neural function and abilities has led consumers to seek out such programs, without critically analyzing the putative benefits of these programs given their specific type of injury.
Activity-based interventions or therapies include any therapy activity, or intervention, that is focused on improving muscle function and sensory perception below the level of injury, and not simply accommodation or compensation for the paralysis and sensory loss due to the spinal cord injury (SCI), in order to improve overall function after SCI. This systematic review examined the literature to see if there is evidence to suggest that activity-based interventions can lead to neural recovery after SCI.
Results of this systematic review show that there are few randomized controlled trials in SCI, and even fewer related to assessing the effectiveness of activity-based interventions for individuals after SCI. There is, however, evidence reported from studies using other approaches, such as quasi-experimental and descriptive research methods, that support the use of ABint to improve function in some persons with incomplete SCI.
Many questions remain after this systematic review. How much ABint is needed to gain improvements, and for how long? This question remains unanswered because the studies that do exist have used different dosages and frequencies of treatment. Who are the best persons to receive or participate in ABint? What is the level of injury or degree of completeness (AIS a, B or C)? For instance, there is little to no research in people with complete SCI research was performed in individuals with complete (AIS A) or sensory incomplete (AIS B) SCI. Future study is needed to more fully explore how the field can help these individuals reach their maximal potential after SCI.
Consumers related to SCI rehabilitation and research should continue to monitor the evidence regarding the efficacy of ABint for SCI to determine if any given program or approach is the right one for any given person with SCI. This will lead to more realistic expectations on the part of the patient with SCI and their loved ones and caregivers, more creativity on the part of clinicians to incorporate the ABint appropriate to any given individual with SCI, and perhaps more agreement and reimbursement of such programs by the payers.
Deborah Backus, PhD, PT, Associate Director SCI Research
2020 Peachtree Rd, NW
Atlanta, GA. 30075
Shepherd Center Systematic Review Group:
David Apple, MD
Lesley Hudson, MS
Jennith Bernstein, PT
Amanda Gillot, PT
Jennifer Huggins, OT
Ashley Kim, OT
Rationale for the Review.
Within the past ten years, the tremendous advances in neuroscience research, as well as the development of new technology geared toward spinal cord injury (SCI) have led to the expectation of “cure” and “lifetime recovery” after SCI. One simply needs to enter the term “spinal cord injury” on the web to pull up myriad sites dedicated to the pursuit of the cure or neurorecovery, or promises that, given the right treatment or participation in an intense “activity-based” program, an individual with paralysis and sensory loss can achieve increased function during post-acute recovery from SCI. The information that emerges includes explanations of experimental protocols in the United States and beyond. Although many individuals with SCI believe that they can achieve greater neural recovery and functionality given the opportunity to pursue an organized, intense, and often costly, treatment program, unless there are identifiable functional goals, third party payers will generally not reimburse for these services. Although it may be natural for some individuals with SCI to believe that treatment after acute care and beyond the insurance company’s predetermined, tangible functional goals will continue to yield results, much of this belief is based on anecdotal reports of success with intense activity based programs, and little empirical evidence.
Recovery can occur after the acute phase of SCI (Kirshblum et al 2004), however, the extent of recovery possible, in what individual (i.e. type, level and extent of injury), and what needs to be done to maximize neural and/or functional recovery, remains in question. Furthermore, whether neural recovery is necessary or sufficient for maximal functional gains, or where this neural recovery needs to occur for functional improvements in humans, remains unknown.
There is substantial evidence in animal models suggesting that the use of intense and repeated sensory stimulation, and intense motor practice, or exercise, can elicit plastic changes throughout the neural axis. Animals who exercise repeatedly and intensely demonstrate changes ranging from increases in growth factors, such as Brain-Derived Neural Factor (BDNF) in the CNS, that are related to both sensory (Hutchinson et al 2004) and motor (Ying et al., 2008) changes with exercise, to arborization of motor neuron dendrites in both intact animals who participate in intense exercise and those with SCI (Gazula et al. 2004). Others report increases in axonal sprouting and synapse recovery, as well as motor recovery, after treadmill training in spinal injured rats (Goldschmidt et al. 2008), and reactivation of neural circuits in the spinal cord with forced-use training (McDonald et al, 2002; Perez et al 2004). Various lines of evidence suggest that there is neural plasticity along the neural axis in humans in response to intense activity, including both sensory stimulation and repeated movement. What remains unclear, however, is whether and where neural plasticity can be facilitated after SCI in humans, and if, in turn, this neural plasticity is necessary and sufficient for substantial functional improvements after SCI.
Encouraging evidence from other patient populations, such as stroke, suggests that intense, focused, repeated active movement of impaired limbs, especially when combined with sensory augmentation, appears to be beneficial for improving function, and inducing neural changes in the cerebral cortex, after stroke. Constraint-induced therapy (CIT), whereby one is “forced” to use the affected arm and hand repeatedly, has been shown to significantly improve upper extremity function in individuals with chronic deficits due to CVA (Butefisch et al 1995; Van der lee et al 1999; Volpe et al 2000). This approach is only now starting to be evaluated in individuals with tetraplegia (Beekhuizen & Field-Fote 2007, Hoffman & Field-Fote 2007).
Locomotor training approaches in SCI, however, are similar to this model of constraint-induced therapy in that the lower extremities are “forced” (facilitated or assisted), either manually or with robots, to move in a stepping pattern that can be used for walking. In fact, locomotor training is widely accepted as a treatment approach for individuals with incomplete SCI. Behrman and colleagues (Behrman et al. 2006) recently reviewed the existing literature on locomotor training approaches after SCI, as well as their own data, and suggested a paradigm shift, whereby clinicians consider the importance and relevance of using approaches that facilitate the weaker muscles after injury and not just adopt compensatory strategies after SCI. Whether changes in locomotor function correlate with neural plasticity, and where in the neural axis these correlations may be found, is still relatively unclear.
The assumption from these different lines of evidence from animal models of SCI, clinical trials in other patient populations such as stroke, and related to locomotor training in SCI, is that intense activity, in the form of repeated active movement, often combined with augmented sensory stimulation, can lead to neural or functional improvements, or both, in humans with any level of injury, with any degree of completeness of injury. In fact, this has led to the development of activity-based programs around the globe, inviting individuals with tetraplegia or paraplegia, complete or incomplete injuries, at any age to participate in order to achieve their maximal potential, and perhaps even full recovery of walking. Yet, a careful review of the literature is imperative in order to examine the evidence related to this issue. Several questions remain unanswered: Do the findings in animal models of SCI translate to humans with SCI? Which individuals with SCI (i.e. age, gender, level of injury, completeness of injury) actually improve in function? Is there neural recovery that can explain improvements in function in humans? Is neural recovery necessary, or sufficient, for substantial and meaningful functional recovery after SCI? Consumers related to SCI rehabilitation and research, whether the individual with SCI, the caregivers, clinicians, or payers for services, require answers to these questions, and can all benefit from a careful review of the literature related to the use of intense activity in order to assist them in making educated decisions about whether or not to pursue, recommend, or fund a given treatment or program.
Objectives of the Review
The main objective of this review was to evaluate all literature in the last 10 years (1998-2008) related to the efficacy for improving neural activity and function with the use of intense therapies, often referred to as “activity-based interventions”, in individuals with paralysis and sensory loss due to spinal cord injury (SCI). “Activity-based interventions” include any therapy activity, or intervention, that is focused on improving muscle function and sensory perception below the level of injury, and not simply accommodation or compensation for the paralysis and sensory loss due to the SCI, in order to improve overall function after SCI. The term “activity-based intervention” can be interchanged with “activity-based therapy”, however, in this review, there are interventions that are provided for research purposes only, and not by a skilled professional or with therapeutic intent, thus the term “activity-based intervention” or ABint, will be used throughout. Since there are many interventions that fall into the category of an ABint, each intervention will be fully defined for the reader, and evaluated rigorously for how well defined the intervention is within the publication. The assumption for this systematic review was that there is important and significant literature that has been published in SCI research, and specifically related to ABints. There are few randomized controlled trials in SCI in general, and even fewer randomized and controlled studies related to the efficacy of ABint for neural recovery in SCI, however, this study group felt that the methods employed by the Boston University Center for Psychiatric Rehabilitation, Innovative Knowledge Dissemination & Utilization Project for Disability & Professional Stakeholder Organizations/ NIDRR Grant # (H133A050006) (E. Sally Rogers, Marianne Farkas, co-PIs) would be useful for examining the issue of ABint for neural and/or functional recovery in SCI. Using this methodology, the group hoped to find the “best” evidence available to explore this issue. The evidence has been evaluated for both meaning and rigor, and the strength of the evidence evaluated and taken into consideration when drawing conclusions from the current literature. The secondary objective of this review is to develop and disseminate products that will inform consumers related to SCI (patients, caregivers, clinicians, educators, administrators of programs) about the efficacy of various activity-based interventions for any given individual with SCI.
Overall, the methods employed for this systematic review were defined by the Boston University Center for Psychiatric Rehabilitation, Innovative Knowledge Dissemination & Utilization Project for Disability & Professional Stakeholder Organizations/ NIDRR Grant # (H133A050006) (E. Sally Rogers, Marianne Farkas, co-PIs), and modified only slightly to accommodate the needs of the SCI field and the literature available in the SCI field.
The ABint study group included any study describing the effects of activity-based interventions on neural and/or functional recovery after spinal cord injury (SCI). The study group defined “activity-based interventions” as: any intervention that specifically uses tools and interventions to improve muscle activation below the level of injury in the spinal cord, and does not rely on compensatory mechanisms for improving function after SCI. Such interventions include interventions that combine intense active movement with one or more of the following: facilitation techniques to activate muscles below the level of injury (such as the use of tactile or vibratory stimulation); electrical stimulation (surface or indwelling); locomotor training (manual or robotic); upper extremity robotics; and intense strength training. ABints do NOT include the use of electrical stimulation or robotics as neuroprosthetics, or tools to replace the lost function below the level of injury.
“Neural recovery” or changes in neural function is defined as measurable changes in neural circuitry or neuronal activity at any level of the neural axis in response to injury or learning. One issue related to determining the efficacy of ABint for inducing neural changes after SCI is that there is no consensus on how to measure neural or neurological recovery in SCI. The major difficulty is that there are no scales that are sensitive to small sensory or motor changes that occur after SCI. Most often neural or neurological change is reported as an improvement on the American Spinal Injury Association (ASIA) impairment scale. This is inadequate for a variety of reasons. Since the current ASIA Impairment Scale (AIS) only assigns neural level in the trunk using the sensory scores, the scale is not sensitive enough to detect motor changes between spinal levels T2 and T10. The reader is encouraged to see Steeves and colleagues (2006) for review. In addition, although the ASIA motor scores are often used as an indicator of neural change, this is often misleading or an incomplete picture of what is really happening in the neuromuscular system after injury or intervention. There is a difference between a motor score that increases because a new muscle is activated (goes from “0” to >/= “1”), and a motor score that increases because a muscle that was already activated increased by a grade (e.g. Increasing from “3” to “4). Given that AIS and the ASIA motor scores are often used as markers for neural recovery at this time, studies that include AIS classification, ASIA sensory or motor scores, or muscle strength changes, will be included in the classification of neural outcomes for the purpose of this review. Other measures of neurological changes include measurement of activity in a neural circuit, such as via a reflex, increases in neural factors, such as BDNF, and demonstrations of supraspinal activity with imaging tools or stimulation, such as functional magnetic resonance or transcranial magnetic stimulation, respectively.
Functional change, or improvement, is often thought to be a marker of neural changes, or recovery, after SCI. Changes in function, however, may not be due to neural changes or recovery at all, but instead, may be due to changes in muscle function or compensatory strategies. “Functional ability” includes any skill that leads to improved mobility (locomotion, bed mobility, transfers) or activities of daily living. Examples of outcome measures for function include, but are not limited to, the Functional Independence Measure (FIM), the Spinal Cord Independence Measure, the Walking Index for Spinal Cord Injury (WISCI), the Jebsen Hand Function Test (JHFT), and the Action Research Arm Test. There are several others, and the reader is encouraged to see Steeves et al. (2006) for further review.
Whether neural changes during recovery after SCI are necessary or sufficient for functional changes, or if such neural adaptations or regeneration, need to be above or below the level of injury, or both, in order to be meaningful (i.e. improve function) is also not known. Therefore, the review of the literature includes both neural and functional outcomes related to participation in ABint and is formatted to group studies into one of three groups: 1) neural and functional outcome measures were used; 2) only neural outcomes were measured, and 3) only functional outcomes were measured. When a study attempted to relate neural to functional changes, we elaborated upon it.
Search Terms used:
- Activity-based (activity based) therapy, rehabilitation, exercise
- Electrical stimulation, neuromuscular stimulation, Bioness
- Resistance training, exercise
- Locomotor training, gait training
- Upper extremity training, facilitation
- Sensory stimulation
- Neural recovery
All terms were paired with spinal cord injury or SCI, paraplegia, tetraplegia and quadriplegia.
The following websites/tools were explored with the search terms identified above:
- Search engines:
- Pubmed (Medline)
- Google Scholar
- Cochrane Reviews
- The citations contained in each article for additional potential articles and reports
The study group excluded the following types of studies/publications/documents from the primary review:
- Studies conducted in animal models (included instead as background)
- Policy statements
- Needs assessments
- Instruments related to outcome measures
- Therapy satisfaction or quality of life studies
- Therapy satisfaction or quality of life studies
- Program models
- Conceptual models
- Process evaluations
- Reviews (included instead in the background)
The rationale for this exclusion was that such documents and articles, while important for the field, could not be subjected to ratings for their rigor and their meaning. The study group did not include conference proceedings, dissertations or government proceedings. There are many studies/articles on health-related outcomes as the result of activity-based interventions, and therefore this is the topic of a second systematic review, and those articles were not reviewed for this summary. Articles that discuss activity-based interventions in patient populations other than SCI are not discussed here (Dromerick et al. 2006). Finally, articles that included discussion of other strategies, such as stem cells and surgical interventions (Belegu et al. 2007; Lim 2007; McDonald 2004), or combinations of ABint and pharmacological approaches (Ying Z 2005), were not included in this review, even if they included a discussion of activity-based interventions. Commonly the objective of these articles was not to rigorously evaluate the efficacy of the ABint for individuals with SCI, but rather to explore the potential options available to these individuals.
Any study published in the 10 years prior to the date of the systematic review (1998-2008), and up to the point of writing this paper (i.e., March 2009) was considered for this review. Any evidence prior to these dates was deemed to be unacceptable, given the extent of changes in the field of SCI rehabilitation and research in the past 10 years.
The lead reviewer queried the databases and located articles, and scanned the titles and abstracts of articles for relevance to SCI, activity-based interventions, neural recovery, and functional recovery. If the title appeared relevant, the abstract was reviewed and if it was deemed likely to meet inclusion criteria, the article was obtained. A checklist of inclusion/exclusion criteria was completed for each article to facilitate tracking of articles screened. The article had to report on the use of at least one activity-based intervention and had to include at least one outcome measure that assessed either neural changes or recovery, functional ability, or both. Studies were then grouped according to whether or not they included neural outcome measures (measuring changes in neural activity anywhere along the neural axis, from the neuromuscular junction to supraspinal centers), functional outcome measures (measuring any skill that leads to improved mobility), or both.
Once a complete list of articles for review was compiled, that list was sent to 11 experts (Michelle Basso, Jan Black, Andrea Behrman, Edelle Field-Fote, Susie Harkema, Joe Hidler, George T. Hornby, Sarah Morrison, Keith Tansey, Candy Tefertiller, Leslie VanHiel) in SCI rehabilitation and research. We asked those experts to review the list to insure that no relevant article or report was omitted. This step yielded 10 new citations that were appropriate for review.
The following study designs were accepted for this review:
- Experimental: Employed methods including a random assignment and a control group or a reasonably constructed comparison group
- Quasi-experimental: No random assignment, but either with a control group or a reasonably constructed comparison group
- Descriptive: Neither a control group, nor randomization, is used. These included case studies and reports, and pre-post designs.
After some discussion amongst the reviewers, case studies were included in this review. The decision was based on the fact that case studies offer one level of evidence that can be used not only to guide treatment, but also to further formulate more appropriate research questions. In a recent review of single subject experimental designs, Schlosser makes the case that case studies are a form of single subject experimental designs. Single subject experimental designs are n=1 designs that use repeated observations or measurement in a time-series design (Schlosser 2009). The unit “n” can be one subject, or can be a group of subjects, organized in a case series. Schlosser (2009) further reviews the phases of research that include Phases 1, in which the focus is to on selecting the therapeutic effect and estimate its’ magnitude, and Phase 2, in which the goal is to discover the extent of the therapeutic effect, including defining the target population, exploring subject characteristics, refining the treatment protocol, and preparing for a clinical trial. Therefore, case studies were included, and subject to the same rigor standards as the other research designs included in this review. Interestingly, the rigor standards used in this review are remarkably similar to those presented in the review of single subject experimental designs discussed by Schlosser (2009).
Some studies were poorly described, or poorly defined, planned and executed, making it difficult to determine the study design. When this difficulty was encountered, the lead reviewer sought the opinion of two other reviewers to determine the study design.
Training of Reviewers
A total of 7 raters were trained to assist in this systematic review. All were individuals affiliated with Shepherd Center, including the co-investigators of the SCI Model Systems grant, the Associate Director of SCI Research, and four clinicians. However, 5 raters rated the majority of the articles.
Training of the reviewers focused on the goals of this review, how to perform a systematic review, the kind of evidence indicating rigor in research, and the criteria that would be used to determine if an acceptable indicator that rigor and meaning had been achieved at each of the points on the respective rating scales. All individuals were trained in the use of the rigor and meaning rating scales by reviewing each item in the scale and discussing the meaning of the item and the evidence that could be considered for each indicator. Research articles were used as training devices by having each rater independently review articles and then discuss their ratings until agreement was achieved. Inter-rater reliability was established through three separate joint ratings of articles over the course of the review period. Each reviewer independently reviewed the same article for both meaning and rigor, then the group met as a whole to review these ratings. When there was a discrepancy between the ratings from different reviewers, these were discussed. The majority of time the ratings differed only by one point. In cases when the discrepancy was more than one point, the lead reviewer would review the definitions for that criterion and the different reviewers would discuss their rationale for their score. Consensus was reached by the group in 100% of the cases where there was originally a discrepancy, and the lead reviewer did not need to intervene or request further review by an additional, outside reviewer. One issue that was resolved after the first reliability session related to the case studies. All of the reviewers, except for one, were consistent in their ratings. One reviewer rated all case studies with significantly lower scores due to a bias that case studies could not be rigorous. Once discussed with relation to issues raised by Schlosser (2009) related to the usefulness of case studies, and when considered with the rigor rating scale, this issue was resolved.
The rigor and meaning scales ranged from a score of “1”, which corresponded to no rigor or meaning, respectively, to a score of “5”, which corresponded to excellent rigor or meaning, respectively. After all articles were scored for both rigor and meaning by the trained reviewers, the lead reviewer tallied the overall rigor and meaning scores (one score for each scale) for each article.
Overall Meaning was determined by tallying the scores on the three sections of the scale. If all three sections had at least two indicators checked “yes”, the overall rating was Level 5; if all three sections have at least one indicator checked “yes”, the overall meaning was Level 4; if two out of three sections had at least one indicator checked “yes, the overall meaning was Level 3; if one out of three sections had at least one indicator checked “yes”, the overall meaning was Level 2; finally, if no sections had a “yes” checked, the overall meaning was Level 1. Articles with a total score of 3 or more were included for consideration of rigor.
Overall Rigor was determined first by the rating on the item that rated the overall methodology for the study: “Study/research uses rigorous or sound research methods that allow the questions of interest to be addressed”. Then the overall average score was compared to the rating on this item. When there was a discrepancy, the lead reviewer examined the cause.
The articles that were deemed to be rigorous and meaningful were then summarized for this review.
Summary of Articles Reviewed
The lead reviewer considered 40 articles for inclusion. After various inclusion and exclusion criteria were considered, specifically, whether neural or functional outcomes were included, 27 articles were rated for rigor and for meaning. Of these, 21 met the criteria for both meaning and rigor, and are discussed further in this report.
A total of 6 studies were deemed as meaningful (scores 3 or greater out of 5), but were excluded from the narrative synthesis after rating for rigor because methodology scores were low enough that the conclusions could not be considered robust or valid (rigor scores less than 3). Issues such as very poor research designs with major threats to internal validity, retrospective measurement, very large attrition of study subjects were reasons for exclusion.
The majority of the 21 articles reviewed were classified as Descriptive (n=16), with experimental (n=3) and quasi-experimental (n=2) designs following.
Upper vs. Lower Extremity
The majority of articles were related to evaluation of the effectiveness of ABint for lower extremity function (n = 17), and more specifically related to the effectiveness of locomotor training approaches (n = 15) in SCI. The other two articles in this category employed strengthening or other interventions. Three articles reported on ABint for upper extremity function.
Adult vs. Pediatric or Adolescent Populations
One article reported on the findings in a pediatric client, and one included findings in adolescents. The vast majority of these articles reviewed were related to studies in adults (n = 19).
Acute vs. Chronic
The majority of articles reviewed were conducted in individuals with acute injuries (<1 year post-SCI) (n = 15), while three were conducted in individuals with chronic injury and three were conducted in individuals with injuries across the entire spectrum of recovery.
Neural vs. Functional Changes
The majority of studies reviewed utilized both neural and functional outcome measures (n=13) to assess the efficacy of the ABint utilized, and the majority of these used the ASIA motor score as the primary marker of neural changes.
Results of Systematic Review
The review is divided into two sections. The first relates to all studies that used lower extremity or gait-related interventions, and the second to upper extremity interventions. Studies are then further subdivided in each of these two sections based on the experimental design. Both groups were classified into one of the following: 1) experimental, 2) quasi-experimental and 3) descriptive.
Lower extremity and Gait-Related Interventions
Experimental Design Studies
Neural and Functional Outcome Measures
There were 2 studies that used an experimental approach in the lower extremity studies, and which met both rigor and meaning criteria, and employed randomization and a well-defined control group (Dobkin et al. 2006, Field-Fote et al. 2005). In a single-blinded, randomized, multi-center clinical trial, Dobkin and colleagues (Dobkin et al. 2006) compared the efficacy of manual locomotor training with over-ground gait training to over-ground training in adults with acute spinal cord injury (SCI). Their 146 subjects were comprised of individuals from six regional centers who were receiving inpatient rehabilitation for the first time, and who were classified with an incomplete injury (American Spinal Injury Association Impairment Scale (AIS) score of B, C, or D), between spinal level C5and L3, and a Functional Independence Measure (FIM) locomotion score less than 4. The level of injury was further classified such that those with injuries between C5 and T10/11 were classified as upper motor neuron injuries, and those with injuries T12 to L3 were classified as lower motor neuron injuries. Those with lower motor neuron injury on one side and upper motor neuron injury on the other were classified with upper motor neuron injury. All of these individuals were within 8 weeks of their SCI when enrolled, and were between 16 and 69 years old. Randomization was based on upper versus lower motor neuron injury and the AIS rating (B, C, D) across all six centers.
Subjects in both groups (locomotor or control) received the standard rehabilitation interventions provided in inpatient and outpatient therapy for mobility and activities of daily living training at their respective centers. Each group also received some form of locomotor training. The locomotor (LM) group received manual locomotor training on the treadmill, followed by over-ground walking training, while those in the control (CONT) group received over-ground walking training only. Neither group received any other form of specified walking training other than these specific interventions.
For the LM group, subjects were first stretched for up to 10 minutes, which was followed by walking on the treadmill with harness support and manual assistance for 20 to 30 minutes (3 to 10 minute increments). The amount of weight support and speed were adjusted so that the individual was able to walk at least 0.72 m/s, with the goal of walking at 1.07 m/s. Treadmill walking was followed by 10 to 20 minutes of over-ground walking with assistance. The focus of all walking, whether on the treadmill with partial body weight support provided by the harness, or over-ground, was on trunk and lower extremity kinematics and limb loading.
The CONT group spent their walking training time either standing or stepping, depending on their individual fatigue, and followed essentially the same order as those in the LM group: stretching (10 minutes), followed by standing or walking for 30 to 45 minutes. Those who could walk practiced in the parallel bars or over-ground with therapist assistance, assistive devices and orthotics as necessary. The subjects in the CONT were not allowed to use the treadmill or harness at all during this 12-week training period.
Both groups trained for 12 weeks, and although there was a specific order of activities followed for all training, the amount of time spent on each activity during a given session was tailored to the individual’s needs. Both the LM and CONT training was conducted for one hour each day, at a location remote from where subjects received their other therapies. Subjects in both groups were permitted to walk at other times during therapy, and to perform trunk and lower extremity strengthening exercises. The number of training sessions for individuals in both groups varied between 45 and 60 sessions, depending on how quickly they reached the highest functional walking speed (0.98 m/s).
The primary outcome measures were obtained pre-intervention, then every 2 weeks for 12 weeks, at the end of the training intervention, and at 6 months and 12 months after enrollment in the study, while the secondary outcomes measures were collected at 3, 6, and 12 months post-enrollment. There were no neural-related outcome measures included in the primary outcome measures for this study; however, secondary outcome measures of neural changes due to the interventions included the ASIA lower extremity motor score (LEMS) and the Ashworth scale. Other outcome measures included: the FIM lower extremity score and over-ground walking speed (primary outcome measures), as well as the distance walked 6-minute walk test, the Berg Balance Scale, the Walking Index for Spinal Cord injury (WISCI), and the SF-54 (secondary outcome measures). Data analysis was completed on all subjects who finished at least 6 weeks of the interventions in both groups, and who met all criteria for the study.
There are several meaningful outcomes for this study, the main being that there were no significant differences between the LM and CONT groups in terms of most of the outcome measures, and specifically in terms of the neural outcome measures. Neither group experienced any significant change in Ashworth scores, or in frequency of spasms. No differences were found in adverse events, and neither group reported excess muscle strain, or joint pain.
Although there were no significant differences between groups, there are other points to note. The majority of subjects with AIS C classification, in both the LM and CONT groups, achieved independent walking. In addition, the majority of those classified as AIS C had a FIM lower extremity score >/= 6 at 6 months post-enrollment. Individuals in both groups (LM and CONT), with AIS C and D classification demonstrated a significant increase in walking velocity, which was consistent with functional community ambulation. In addition, speed continued to increase between 3 and 6 months, in both groups. Finally, for individuals classified as AIS B, neither intervention led to improvements in over-ground walking. Only those that were enrolled as AIS B, who converted to AIS C during the intervention phase, actually improved in walking speed.
Although the authors report that earlier studies demonstrated improvements in walking independence and speed post-manual LM training, they also acknowledge that these earlier studies were conducted in individuals with chronic injuries, while this study was conducted in individuals with acute SCI. Several other issues may have confounded their finding that there was no significant difference between the LM and CONT groups. One factor of significance is that there may not have been enough of a contrast between the LM plus over-ground training (LM) and the over-ground interventions (CONT). Both interventions were more intense and more task-specific than that which may be generally provided in traditional therapy, which may have lead to the lack of significant differences in outcomes between the two groups.
Functional Outcome Measures Only
Field-Fote and colleagues (2005) compared different locomotor training approaches in a randomized controlled trial. They enrolled 27 adults with chronic (> 1 year post-SCI) motor-incomplete SCI, that were randomly assigned, based on their pre-training LEMS, to one of four groups: 1) treadmill training with manual assistance (similar to the locomotor training intervention in the trial presented above, Dobkin et al) (TM), 2) treadmill training with stimulation (TS), 3) over-ground walking with stimulation (OG), and 4) treadmill training with robotic assistance (LR). Prior to this study, these four training approaches had not been compared in adults with motor-incomplete, chronic, SCI.
All subjects, regardless of the intervention, were trained with body-weight support, which was modified based on the amount of knee flexion during the stance phase or toe dragging during swing phase, but was always < 30%. All subjects participated in training for 60-minute periods, and trained for 5 days/week for 12 weeks, and were allowed to rest during each session as needed.
Subjects in the treadmill training groups (TM, TS) were allowed to use the handrails, but not to bear full weight through their arms. In addition, all subjects in these groups were encouraged to walk at their maximum walking speeds, as long as walking quality was maintained. The techniques employed were those recommended by Behrman and Harkema (2000), which essentially involved providing enough assistance to execute a step with normal kinematics. Subjects in the treadmill training with robotic assistance (LR) were progressed in walking speed based on a pre-determined algorithm until they could reach the maximum speed of 3.2 km/hr, or 2 miles/hr by 5 weeks. Subjects in the over-ground group (OS) were instructed to walk as fast as possible around the 80-foot track, and were allowed to use both the upper extremity assistive device and lower extremity orthotic with which they were most comfortable. There was no attempt to advance either device with training.
Subjects who performed training with electrical stimulation (TS, OS), all had the same relative placement of electrodes, positioned to get the most robust flexion withdrawal response. Stimulation parameters were 300 to 600 ms train, 50 Hz, 5 to 20 mA. These parameters were adjusted throughout the training sessions to prevent habituation.
The outcome measures for this report assessed function only, and included a 6-meter walk and a 2-minute walking test. For all walking tests, the subjects were allowed to walk at self-selected walking speeds, and were videotaped for evaluation of walking performance. Subjects were allowed to use whichever assistive and/or orthotic devices to which they were accustomed. These results reported here are part of a larger scale study that has not yet been reported at this time.
There were 7 subjects in each group, except for the LR group, which had 6 subjects. The number of training sessions over the 12-week training session ranged from 27 to 54. Subjects in all four groups improved in walking performance, and there was no significant difference between groups. Subjects in each of the four groups demonstrated improvements in walking speed, and those with the most impairment in walking function showed the greatest improvements. Power analysis suggested that more subjects would be required in each group in order to detect a significant difference; however, there was a trend for greater improvement in walking in the electrical stimulation groups (TS, OS). The authors acknowledge, however, that this is the treatment approach with which their laboratory is most experienced. In addition, although subjects did improve, none were able to discard their wheelchairs and walk independently or in the community.
Quasi-Experimental Design Studies
Reviewed in this category were studies that did not employ random assignment, but that used an activity-based intervention, and a reasonably constructed comparison group. Only one study fulfilled this criterion (Grasso et al. 2004). Grasso and colleagues (2004) evaluated the neural and functional outcomes in 22 adults (17 to 60 years old), half with acute SCI (1 to 6 months post-SCI) and half age-matched controls with no neurological impairment. The injury levels ranged from C7 to L2. Five subjects were classified clinically as AIS A, four as AIS C, and the remainder as AIS B. All subjects performed daily manual assisted locomotor training sessions, and worked on increasing speed and decreasing the amount of body-weight support and manual assistance required. Training time ranged from one to three months.
The neural outcome measures included the Modified Ashworth Scale (MAS) (Bohannon and Smith 1987), as well as kinematic and electromyography (EMG) data collected during stepping attempts on the first day of training, and every 15 days thereafter. The kinematic data was analyzed for end-point path (the mean area derived from the foot trajectory), end-point (foot-trajectory) variability, velocity curvature, and inter-segmental coordination. EMG data was analyzed for spatiotemporal patterns of motor neuron pool activities in the spinal cord. This is passed on the rostrocaudal distribution in the spinal cord of motor neuron pools innervating the different muscles in the human spinal cord based on a reference population (Kendall et al. 1993). The Rivermeade Mobility Index (RMI), the Walking Index of Spinal Cord Injury (WISCI), and the Garrett Scale (Garrett et al. 1987) were utilized.
The results regarding walking were similar to what had already been reported: those who could not walk before training (n=5) could walk within the treadmill environment post-training, but not overground. Three subjects were able to walk independently in the community post-training, and the remainder maintained some degree of disability in walking overground. Of great interest are the neural outcomes, suggesting neural plasticity in a distributed fashion in the neural axis.
In general, the trajectory of the foot during stepping in subjects with SCI who were able to step in the first session, stepping started very irregularly, but progressed to the shape typically found in able-bodied individuals. The majority of these subjects (n = 8) also demonstrated a significantly greater end-point path, indicating a longer step length and greater foot clearance during stepping. The inter-segmental kinematic coordination also changed in part after training, in that the angular oscillations, which are stereotypical in able-bodied subjects, increased in amplitude and decreased in variability. The phase-relationship between limb segments remained abnormal, most likely due to different muscle activation patterns from what is available in able-bodies individuals. Thus, although the foot position in space closely approximated that of the able-bodied subjects during stepping on the treadmill, the EMG activity deviated from that of the able-bodied subjects, and continued to increase with training.
The averaged and normalized EMG waveforms were mapped on the published charts of the segmental localization. In subjects classified with AIS C SCI, the activity in the upper lumbar (L2 to 4) segments of the cord started later and lasted longer than in able-bodied subjects, while that in the lower lumbar cord (L5 to S2) lasted for a shorter duration than in the able-bodied subjects. This activation in L5 to S2 corresponds to weight acceptance and activation of the hip extensors and ankle plantarflexors. Then large regions of the cord, more widespread than for able bodied, became active at the transition between stance and swing. This pattern in the subjects with SCI suggests that the control of stepping has changed its distribution after training, and is different from that in the able-bodied population.
Neural and Functional Outcome Measures
Winchester and colleagues evaluated supraspinal activation patterns to evaluate neural changes using functional magnetic resonance imaging (fMRI) in four adult men (20 to 49 years old) with motor incomplete (AIS C or D) tetraplegia after 12-weeks of robotic locomotor training (Winchester et al. 2005). They also measured functional outcomes using the Walking Index for Spinal Cord Injury II (WISCI II) and over-ground gait speed. Time post-SCI ranged from 14 weeks to 48 months, and thus included adults with both acute and chronic injuries.
The subjects were supported with a harness and provided enough support to allow for optimal kinematics in the lower extremities during stepping. Since this was robot-assisted locomotor training, the subjects were also fitted with the driven gait orthosis to assist with lower extremity stepping. Subjects were scheduled for 60-minute sessions 3 times a week for 12 weeks. The 60 minutes included set up and take down. Subjects initially walked at 2.0 kmph, and the speed was increased, as the subject was able to step with adequate toe clearance. At the end of the training program, the speeds varied form 2.3 to 3.2 kmph across the group of subjects. Subjects also were trained over-ground once they could support 80% of their body weight on the treadmill. Treadmill training was continued, however, for a minimum of 20 minutes per session even after over-ground training was initiated. The assistive and orthotic devices required for safe walking with the most appropriate kinematics were allowed when walking over-ground.
Outcome measures (fMRI, WISCI II and over-ground gait speed) were collected pre- (1 to 2 days prior to initiating training) and post- (immediately, first day after completion of training) the training. The motor task during data acquisition with the fMRI was ankle plantar flexion and toe flexion (20s on, 20 s off), and was repeated unilaterally for a total of 10 repetitions.
All but one subject was able to walk over-ground after the intervention period, but the speed of ambulation varied, as did the assistive and orthotic devices utilized. All four subjects, however, demonstrated significant changes in supraspinal activation patterns with the fMRI post-training. The functional ability in walking was related to the amount of signal and location of signal in the fMRI. The two subjects who achieved independent over-ground ambulation post-training not only demonstrated increases in activation of the sensorimotor cortex related to toe and foot movements, but also demonstrated the most marked increase in cerebellar activation. The one subject who was not able to ambulate at all demonstrated very little activation in the cerebellar cortex post-training. The subjects who received the robotic locomotor training the soonest after injury demonstrated the greatest improvement in functional and motor scores.
Neural Outcome Measures Only
Another way to assess neural function at the level of the spinal cord is to employ evaluation of the H-reflex pre- and post-training. This is the method used by Trimble, Kukulka and Behrman (1998), in one individual with sensory (AIS B) incomplete paraplegia at S1/L2. The subject, as well as 12 other able bodied subjects, trained on a treadmill at speeds equivalent to his over-ground fast walking speed. All subjects trained at the fast walking speed 30 minutes per day, every other day for 10 days, then continued three times a week for 4 months. This study did not employ either body-weight support or manual or robotic assistance. Prior to and at the completion of training, the soleus H-reflex was recorded. There was a significant increase in the threshold of the right soleus H-reflex, which approached that of able-bodied subjects. This neural change corresponded with functional changes seen during walking (i.e., increases in self-selected and fast walking speeds).
Still others have attempted to improve lower extremity neural control and function using other activities, such as functional electrical stimulation cycling (Griffin et al. 2008), or intense strengthening to enhance walking ability (Gregory et al. 2007). Griffin and colleagues (Griffin et al. 2008) evaluated the efficacy of functional electrical stimulation (FES) cycling on neural ASIA sensory and motor scores, amongst other health-related variables, in 18 adults with chronic paraplegia or tetraplegia. The majority of subjects were classified with incomplete injuries (n=13), and the remainder were classified as complete, although the method for determining these classifications was not elaborated upon.
Outcome measures were assessed before and after a 10-week intervention period that consisted of training 2 to 3 times a week for 10 weeks on an FES cycle in an inpatient hospital. The stimulation frequency for the FES cycle was constant at 50 Hz, and the maximal stimulation intensity was 140 mA, and adjusted to maintain a cadence of 49 rpm. When the cadence dropped below 35 rpm, the subjects were allowed to go into a 2-minute cool down period. If they had not cycled for a total of 30 minutes, they were allowed to rest for 5 minutes before attempting to cycle again. The resistance was only increased by 1 kp after the subject was able to cycle for three consecutive sessions for 30 minutes without interruption.
Neural improvements were demonstrated by significant improvements in ASIA LEMS scores, as well as the sensory scores, following the 10 weeks of training. These improvements coincided with increases in cycling power over the duration of the intervention period, and suggest that the FES cycle might be a viable alternative for improving motor function in the lower extremities for individuals with incomplete SCI.
Functional Outcome Measures Only
Hicks and colleagues evaluated the long-term efficacy and carryover of long-term manual body-weight supported locomotor training in 14 adults with chronic (1.2 to 24 years post-SCI), incomplete (AIS B (n=2)), or AIS C (n= 12)) SCI. The majority had cervical level injuries (n = 11), and the remainder had thoracic level injuries. All subjects trained 3 times a week until 144 sessions were completed, which required approximately 12 months. All subjects were offered the opportunity to continue training 1 time per week or fitness training 2 times per week for 8 months following completion of the locomotor training program. Functional walking was measured by speed of walking, distance walked, amount of body-weight support required and the Modified Wernig Scale. Assessment were performed every 36 sessions, and also 8 months after completion of the locomotor training program, if subjects returned for that long-term follow-up (n= 12).
All subjects demonstrated significant improvements in walking on the treadmill, as has been reported by other investigators. Only three of the subjects were able to walk overground (but were wheelchair dependent for mobility) prior to training, however, following training, 6 other subjects were able to walk overground, with varying degrees of assistance required. The authors measured other outcomes, but based simply on functional outcomes, their findings suggest that individuals with incomplete, chronic SCI, may not experience significant functional improvements with locomotor training, Since the investigators did not measure neural outcomes, it is difficult to know if this lack of improvements is due to a limited capacity of the nervous system to recover this long after injury or if there are other physiological limitations, such as limits in muscle plasticity this far post-SCI.
One adaptation to treadmill training has been to augment the sensory experience by either loading the lower extremities while the subject is walking, or by applying resistance with the robotic device, while walking in the robotic device with a velocity-dependent resistance. Lam et al. (2008), evaluated the functional gains in 9 adults (ages 46 to 73 years) with either acute or chronic, incomplete (AIS D) SCI who walked with one or the other of these forms of sensory augmentation while surface EMG was recorded from the bilateral rectus femoris, biceps femoris, tibialis anterior and medial gastrocnemius in 6 of the 9 subjects (in three subjects, EMG was recorded from one side only). Two subjects only walked with the weights, 3 only walked with the robot-generated resistance, and four walked under both conditions. In both conditions, subjects walked on a treadmill, with body-weight support for safety only (i.e. there was no body-weight support provided). Data collection began when subjects appeared, and reported, comfort in walking on the treadmill. At least 20 steps were recorded at baseline. Then, in the case of the weighted condition, the treadmill was stopped and the weights applied. The subjects in the weighted group then walked for 1 minute each with 1, 2, and 3 kg weights around the lower leg (at midshank). In the case of the robotic training device, after one minute of comfortable walking, during which baseline data was collected, the device applied a velocity-dependent moment against the hip and knee, which represented a viscous resistance throughout the step cycle. The resistance trial for the robotic-generated resistance lasted between 17 and 85 steps, depending on the number of rest breaks.
The six subjects who walked with the weights around their leg demonstrated no consistent change in their pattern of walking while the weights were on their legs, regardless of the amount of weight. There was, however, an increase in biceps femoris (knee flexion) activity during the weighted conditions in all subjects, and in some, there was an increase in tibialis anterior and rectus femoris, during swing phase. The activation of the biceps femoris increased as the amount of weight increased, but the peak knee flexion decreased during swing compared to baseline. After the weight was removed, muscle activities returned to the baseline levels. In the seven subjects who walked with the robotic-generated resistance, there was much more variability between subjects in terms of where the increased flexion was noted, but the over all there was also a decreased knee flexion during swing phase.
The investigators also recorded the first several steps in 5 subjects who walked under the weighted condition, and in 4 who walked in the robotic-resistance condition. In the weighted condition, all subjects demonstrated a significant increase in peak knee flexion when compared to baseline, whereas this change was not significant in the robotic-resistance condition. The findings of this study suggest that proprioceptive feedback can be provided during treadmill training using either weighted resistance, or resistance applied via the robotic training device itself. Furthermore, the increased proprioceptive input may facilitate greater flexor activity during the swing phase of gait.
Fourteen of the studies using the descriptive study design evaluated the efficacy of activity-based intervention for lower extremity.
Case Studies and Reports
There are a variety of activity-based interventions that affect lower extremity function that have been evaluated using the case study or report approach. These studies meet the criteria set forth by Schlosser (2009), and were evaluated using the same rigor scale as the other studies in this review.
Neural and Functional Outcome Measures
Some groups have employed the manual locomotor training approach similar to that outlined in the Dobkin et al. study (2006), whereby subjects walk with body-weight support over a treadmill with manual assistance to the lower extremities to facilitate the gait kinematics that closely approximate able-bodied gait kinematics (Behrman & Harkema 2000, Behrman et al. 2005, Behrman et al. 2008, Protas et al. 2001). In an earlier publication, Behrman and Harkema (2000) reported that three adults with motor incomplete SCI (AIS C or D), and one subject with complete (AIS A) SCI, improved walking over the treadmill. All but one subject were paraplegic (injury level below T1), and one was tetraplegic (C6). The three subjects with motor incomplete SCI all had been injured less than one year, whereas the one subject with complete SCI had been injured for approximately one year when the training was initiated.
Training procedures were similar to those described previously in the Dobkin et al. study, in that subjects walked with body-weight support on a treadmill and manual assistance from trainers to provide the appropriate sensory cues required for optimal gait kinematics. Over-ground training in this study was initiated when an individual could maintain independent standing while supporting at least 80% of their body weight, and could initiate stepping with appropriate kinematics in at least one leg. The three subjects with AIS C or D classification trained 3 times per week, while the one with AIS A trained 5 times per week.
The only neural-related outcome measure for this study was the ASIA lower extremity motor score (LEMS). The subject classified with an AIS A paraplegia did not improve in ASIA LEMS, although she did improve in stepping on the treadmill. She did not receive over-ground training and was not evaluated for over-ground walking.
Of note, not only did the time post-injury and the AIS vary, but the subjects that were motor incomplete were also all males, whereas the one with complete SCI was a female, and the ages varied between 20 years old for two subjects, 43 and 45 for the other two. Nevertheless, both subjects with motor incomplete, AIS C, SCI, improved in ASIA LEMS, as well as stepping on the treadmill, while the subjects with AIS D classification did not show improvements in lower extremity motor scores that corresponded with their improvements in walking, not only on the treadmill but over-ground as well. Only the two subjects with AIS D classification could be evaluated pre- and post-training in walking over-ground, and both of these subjects demonstrated improvements in walking speed and distance. The one subject with AIS C classification was not able to walk over-ground prior to the intervention, was not evaluated formally post-intervention, but attained the ability to walk over-ground.
Behrman and colleagues report similar findings in a subsequent case report (Behrman et al. 2005, Behrman et al. 2008). In 2005, they reported that a 55-year-old man with an acute SCI leading to tetraplegia at spinal levels C6/7, and with AIS D classification, did not demonstrate significant changes in neural function (i.e., improvements in LEMS), although he did demonstrate improvements in walking function. After 45 sessions of manual locomotor training, 5 times a week for 9 weeks, he demonstrated improvements in walking speed, distance, and kinematics over-ground.
The second case study evaluated both neural and functional outcomes following locomotor training in a pediatric subject (Behrman et al. 2008). The study subject demonstrated similar outcomes to those reported in adult subjects, suggesting that the methods most appropriate for improving neural function after SCI in children may be similar to those utilized in adults. This subject was a 4.5-year-old child with AIS C tetraplegia, at spinal levels C6/7, who participated in manual locomotor training for 16 months. He began this program approximately one year after injury. Evaluations were performed during the five days before initiation of training, and upon completion of training. The only neural-related outcome measures were the ASIA LEMS and somatosensory scores. The intervention included both locomotor training over the treadmill, with manual assistance, and over-ground walking each session. At the completion of 76 sessions, this child demonstrated no significant increases in LEMS or sensory scores on the ASIA exam. He did, however, demonstrate improvement in walking independence, to the point that he could ambulate in the community with a rolling walker and with a self-selected gait speed of 0.29 m/s and maximum speed of 0.48 m/s.
Prosser (2007) reported similar findings to those of Behrman and her laboratory, in a pediatric client. The subject was a 5-year old girl with SCI at C4, AIS C, in a Brown-Sequard pattern, with a mild head injury as well. Locomotor training was added to her inpatient physical therapy program, approximately one month after her injury. The primary outcome measures were the AIS, the Functional Independence Measure for Children (WeeFIM II) mobility score and the WISCI II. The subject was trained on a body-weight support system that was slightly different from the system used by Behrman and colleagues in that the body-weight support was measured with the subject standing over a scale and bearing down with her weight. The weight on the scale was subtracted from her body-weight to determine the percent of support she was receiving. Overground training was initiated 10 weeks after her injury when she was able to independently step with her right leg on the treadmill. Initially, the subject used a rolling walker and an articulating ankle-foot orthosis (AFO) on her left leg, and walked with assistance of two people overground. The focus, throughout both treadmill and overground training was on kinematics of gait. Locomotor training was performed 3 to 4 times per week, for a total of 6 months, and ranged form 10 minutes for the first three sessions, to 20 minutes for the remainder of the sessions.
The subject was able to decrease her body-weight support over the duration of the locomotor intervention, from 80% support to 10% support during the 6th month of training. Her speed also improved, increasing from 0.27 m/s to 0.98 to 1.12 m/s. At the completion of training she had progressed from the rolling walker, AFO, and assistance of two people, to walking with bilateral Loftstrand crutches, the AFO, and supervision only overground. Her LEMS increased from 4/50 to 29/50. The WeeFIM II scores improved from 5/35 to 21/35 in mobility, and the WISCI II from 0 to 12. Of course, since this is a case study, during the acute phase of recovery, and since the subject also received concurrent inpatient therapy, it is not possible to determine causality between locomotor training and improvements in walking function. However, this report does demonstrate that manual locomotor training can be used in at least some clients in the pediatric population with no harm to the subject.
In a multicenter case series, Wirz et al. (2005) evaluated the functional (primary outcome measures) and neural (secondary outcome measures) outcomes pre- and post-training with robotic locomotor training in 20 adults (16 to 64 years old) with motor incomplete (AIS C or D) tetraplegia (n = 11) and paraplegia (n = 9). Levels of injury ranged from C5 to L1, thus including subjects with lower motor neuron injury, unlike the locomotor training studies already reported. All subjects had chronic injuries (2 to 17 years post-SCI). Similar to the manual locomotor training that was described previously, while in the robotic system, subjects were provided with body-weight support while walking on a treadmill. The major difference between the manual training and the robotic training employed in this study is that for this study, robotic assistance was utilized. For robotic assistance, the legs were positioned in the orthoses of a “Driven Gait Orthosis” that was powered by a computer while subjects walked on the treadmill, which assisted them in their stepping motion. Subjects trained in approximately 45-minute sessions, 3 to 5 times per week for 8 weeks.
Functional outcome measures included the WISCI II, the 10-Meter Walk Test (10MWT), 6-Minute Walk Test (6MWT), and Timed Up & Go (TUG). Neural outcome measures included the ASIA LEMS, the Ashworth Scale and the Spinal Cord Assessment Tools for Spasticity (SCATS). All of the subjects who were unable to walk prior to the intervention were able to walk over-ground after the intervention. Only two of those who could ambulate prior to the intervention demonstrated functional improvements on the WSICI II. Subjects who were able to walk over-ground demonstrated significant increases in gait speed and distance walked, and there was no difference in rate of improvement between the first 4 weeks and the final 4 weeks of training. All but two the subjects tested in the TUG demonstrated improvements in balance, as measured by a decrease in the time to perform the test, with the greatest change during the first four weeks of training. Of note, there was less improvement in individuals with injuries above T11, and in individuals who were not taking anti-spasticity medications. In addition, there was also a significant correlation between the pre-training performance and the magnitude of improvements for the 10MWT and the 6MWT. The slower walkers had the greatest improvements in gait speed and distance.
Only 10 subjects were tested in the neural outcomes, and only at one of the centers. The only significant increase in LEMS was between the 4 and 8-week assessments. In the majority of the subjects (90%), the changes in LEMS did not correlate with the changes in performance on the walking function tests (10MWT, 6MWT, TUG). There was also a significant decrease in spasticity, but only in the extensor spasm score as measured with the SCATS.
Protas and colleagues (Protas et al. 2001) expanded the number of neural outcomes used in their pilot study by including not only the commonly used ASIA classification and the Ashworth assessment of spasticity, but also the Brain Motor Control Assessment (Sherwood, McKay & Dimitrijevic 1996, Sherwood, Priebe & Graves 1997). The Brain Motor Control Assessment (BMCA) employs surface electromyographic (EMG) recordings of muscle activity during a standardized protocol to assess changes in motor control. This protocol involves voluntary movements, reinforcement maneuvers, and reflex stimulation. In this study, three subjects with chronic, motor incomplete, thoracic SCI were enrolled to train on a treadmill 1 hour per day, 5 days per week, for 12 weeks. The protocol for the locomotor training was similar to those studies reported earlier, in that subjects walked with body-weight support on the treadmill, with trainers providing manual assistance and verbal cueing to facilitate optimal kinematics during stepping. Subjects walked until they reported fatigue, and then were allowed to rest before continuing. Over-ground walking in this study, however, was initiated in all subjects after 3 weeks of training, regardless of the amount of body-weight support, walking speed, or stepping kinematics. Neural changes reported included a shift in EMG activity toward that which is seen in able bodied subjects in two of the subjects, however, there were no other consistent findings related to neural function across all three subjects. There were no significant changes in motor function, and no significant changes in Ashworth scores, although one subject reported a reduction in his clonus as training progressed. Of interest, however, is the report that all 3 subjects tripled their gait speed and endurance s a result of training, and that these changes are independent of the neural outcomes.
Hornby, Zemon and Campbell evaluated the efficacy of robotic-assisted body-weight supported treadmill training (progressed to treadmill training with minimal manual assistance) on neural and functional improvements in three adults with motor incomplete SCI. Outcome measures included ASIA LEMS as a marker of neural recovery, as well as the Functional Independence Measure (FIM) locomotor subscale, the Walking Index for SCI II (WISCI II), 10-Meter walk test, 6-Minute Walk test, Timed “up & Go” test, and the Functional Reach test and the modified sitting Functional Reach test. All three subjects were classified as AIS C, with high-level lesions (two at C6, and one at T2). One patient was 13 years old, six weeks post-SCI, another was 40 years old, 5 weeks post-SCI, and the third was 43 years old, 18 months post-SCI.
Similar to the manual locomotor training described previously, in the robotic system, subjects were provided with body-weight support while walking on a treadmill. The major difference between manual training and the robotic training employed in this study is that, initially, robotic assistance was utilized based on subject performance. Each subject transitioned to body-weight supported treadmill training with minimal to no assistance. For robotic assistance, the legs were positioned in the orthoses of a “Driven Gait Orthosis” that was powered by a computer while subjects walked on the treadmill, which assisted them in their stepping motion. Subjects were transitioned to walking on a treadmill with body-weight support, with manual assistance of one person, in order to encourage voluntary stepping, when able to walk over-ground with minimal physical assistance and assistive and orthotic devices as required. Manual assisted locomotor training was discontinued when the subject had reached a plateau (lack of increase in weekly obtained FIM locomotor scores, WISCI II scores and less than 10% increase in gait speed or distance) for 4 weeks. Subjects trained between 16 and 20 weeks.
Two of the three subjects experienced substantial increases in their lower extremity motor scores, as well as walking ability. One subject, however, showed fluctuations in the lower extremity motor scores, but also improved in walking ability, as measured in all cases by increased speed and distance ambulated post-intervention.
Some researchers and clinicians have postulated that the “best” activity-based interventions are those that are “task-dependent”, meaning that the increased activity is provided by the task, or within the task, desired to improve. For example, if interested in improving walking, the “best” way to do so is to find task-dependent ways to train, i.e. on the treadmill, either with manual or robotic assistance. Gregory et al. (2008), however, provide evidence of functional improvements in three adults (22, 61, 58 years old) with chronic (17, 27, 24 months post-SCI, respectively) who trained with an intensive lower extremity resistance program 2 to 3 times a week, for 12 weeks. The strengthening program included plyometric training as well as traditional strength training techniques. All subjects were able to walk independently upon enrollment, without an assistive device, to allow the investigators to evaluate benefits of this program for improving gait speed.
Both neural and functional outcome measures were used. Strength measurements were performed of the plantarflexor and knee extensor muscle groups using the Biodex isokinetic dynamometry. Voluntary activation deficits, or the inability to activate the muscle voluntarily, were determined using a single-biphasic, supramaximal pulse at rest and during a maximal voluntary contraction. Temporal and spatial characteristics of gait were measured during a 10m walk over the GaitRite, as well as over a split-belt treadmill with instrumentation to measure 3-d ground reaction forces.
All three subjects demonstrated increases in peak torque production in the plantarflexors and knee extensors, and a decrease in time to peak tension and rate of torque production in these muscles. There was also an increase in maximum gait speed post-training and in self-selected gait speed. Finally, there were also concurrent increases in step length in both limbs, increases in both anterior/posterior and vertical ground reaction forces on the GaitRite, as well as improved symmetry in propulsion. The findings from this study suggest that adults with chronic weakness and gait deficits due to SCI can benefit from intense strengthening, even if not task-specific, and that some of these benefits may translate to functional activities.
Functional Outcome Measures
The efficacy of functional electrical stimulation (FES) has been reported previously (and will be discussed later in this review), but this was in conjunction with some other functional task, such as massed practice of upper extremity activities (Hoffman & Field-Fote 2007, Beekhuizen & Field-Fote 2007), or walking on a treadmill (Field-Fote et al. 2005). Thrasher and colleagues (2005), reported evidence that adults (n=5) with chronic, incomplete SCI, who trained with FES as a neuroprosthesis for overground walking demonstrated significant improvements in walking without the neuroprosthesis, overground, including increases in walking speed, stride length and step frequency. Two of the subjects walked slightly slower at the time of the post-evaluation (10 weeks), but all remained significantly faster than at the beginning of the treatment. One subject required a different assistive device, to provide more support, although his walking speed and parameters improved post-training.
The findings from experimental, quasi-experimental and descriptive design studies in the lower extremity or in gait-related descriptive studies suggest that, while there is a significant lack of evidence from randomized controlled studies supporting ABint to improve neural outcomes in individuals with SCI, there is some evidence using other methods, in a rigorous manner, that have meaning to the consumers related to SCI research:
- Functional changes are possible after ABint in individuals with chronic, motor incomplete (AIS C or D) SCI:
- Adults with motor incomplete SCI (AIS C or D), regardless of the level of injury, may experience improved walking (function) following ABint that are task specific to lower extremity function or gait;
- Strengthening may also be a necessary or sufficient component for improving some aspects of gait in individuals with incomplete SCI;
- Children with similar types, levels and degree of completeness of SCI may have the capacity to improve neural function with intense, task specific ABint, however, only one case study supports this conclusion at this time.
- Neural changes are possible following ABint targeted to both lower extremity;
- Evidence from experimental and quasi-experimental studies support the findings in descriptive studies, that neural improvements are possible in individuals with AIS C SCI, regardless of the level of injury, in response to ABint that are task specific.
- In individuals with AIS D SCI, however, there are no significant changes in the motor scores, despite the fact that there are functional improvements in walking. This may be due to the fact that there is a ceiling effect to ASIA motor scores, and not only may other muscles be involved in walking, which are not measured with the ASIA motor scale, but there may also be problems other than just strength affecting walking, such as intersegmental coordination and control. This hypothesis remains to be fully tested.
- ABint appears to improve ASIA sensory scores in adults with AIS C SCI, however it remains unclear how this relates to function, and
- Whether somatosensory augmentation is required or sufficient to lead to neural and functional changes remains unclear.
- Locomotor ABint in adults with motor incomplete SCI may lead to supraspinal plasticity (i.e. neural changes) post-training, may be related to functional outcomes in walking, and warrants further investigation.
- Neural recovery may not be necessary, or sufficient, to lead to functional changes in individuals with AIS C or D SCI.
- Whether individuals with acute injury and chronic injury respond similarly to the same interventions is not yet clear.
- A consensus on the primary and secondary outcomes of activity-based interventions would be beneficial for advancement of research and treatment in SCI.
- Furthermore, a consensus on the meaning of “neural improvements” and study of the relationship between neural outcomes and functional changes is imperative to further the understanding of mechanisms of neural plasticity, and the necessity or usefulness in facilitating greater function after SCI.
- Further studies are required using similar interventions and methodology, and specifically similar outcome measures to determine if there is actually neural benefit from these interventions.
- Studying neural benefits due to activity-based interventions with a randomized design is feasible, and evidence from quasi-experimental studies suggests that experimental designs are possible and would be useful in order to gain a greater understanding of the efficacy of activity-based interventions for neural and functional outcomes in individuals with SCI.
- The outcomes reported in these descriptive studies should be confirmed or rejected with larger scale studies in order to determine the neural effects of ABint in individuals with either acute or chronic injury.
Upper Extremity Interventions
Experimental Design Studies with an Activity-Based Intervention.
There was one study that used an experimental approach, and which met both rigor and meaning criteria, and employed randomization and a well-defined control group (Beekhuizen & Field-Fote 2005). Beekhuizen and Field-Fote conducted a study that employed an experimental design in order to evaluate neural and functional outcomes after providing an activity-based intervention in adults (age 16 to 70 years, average 38 +/- 15 years) with tetraplegia due to chronic, motor incomplete (AIS C or D) SCI, at or rostral to spinal level C7. In this study, 24 subjects were randomized to one of four groups: 1) massed practice training combined with somatosensory peripheral nerve stimulation (MP+SS), 2) somatosensory peripheral nerve stimulation only (SS,), 3) massed practice training only (MP), and 4) no intervention (CONT). All interventions were delivered for 2-hour sessions, 5 days a week for 3 weeks.
Subjects in both MP groups (MP, MP+SS) practiced a variety of upper extremity tasks, in a program designed by the authors. Subjects performed repetitive practice of tasks in each of 5 categories that included: gross upper extremity movement, grip, grip with rotation, pinch, and pinch with rotation. Each of these categories in turn had 14 tasks that emphasized that motion. The tasks in each category were performed for 25 minutes before moving on to the next category. A one-minute rest period was allowed at the completion of each test category.
The somatosensory peripheral nerve stimulation groups (MP+SS, SS) were treated with median nerve stimulation at a frequency of 10 Hz, on/off duty cycle of 500/500ms, and pulse duration of 1 ms, delivered at 1 Hz. The somatosensory peripheral nerve stimulation was applied in conjunction with the MP in the MP+SS group. Subjects in the control group did not participate in any organized therapy during their three-week period, and were only tested at the beginning of a three-week period and at the end of this period.
Outcome measures were chosen to evaluate both neural and functional outcomes. Those focusing on neural outcomes were maximal pinch grip force, Semmes-Weinstein monofilament sensory testing, and the intensity of cortical stimulation required to evoke a motor threshold response in the thenar muscles of the treated limb. Functional outcomes were assessed with the Jebsen-Taylor Hand Function Test (JTHFT) and the Wolf Motor Function Test (WMFT).
All intervention groups demonstrated significant improvements on the JTHFT when compared to the CONT group, with the greatest improvements reported for the MP+SS groups. The MP+SS and SS groups demonstrated significantly greater improvements on the WMFTas well, but again, the greatest improvements were found in the subjects in the MP+SS groups. Somatosensory scores, as tested by the monofilaments, were significantly greater in the MP+SS group than in the CONT group. The MP+SS, as well as the MP, group differed significantly from the CONT group. The results reported by Beekhuizen and Field-Fote indicate that MP+SS, MP and SS interventions may improve functional hand use when compared to no intervention, as in the CONT group, in adults with motor incomplete tetraplegia. Furthermore, those in the MP+SS and SS groups also experienced improvements in upper extremity function, as well as pinch grip force compared to the CONT group. However, only the MP+SS group experienced improved sensory scores. Of significance is the issue that the authors were not able to demonstrate that those in the MP+SS group would achieve greater neural and functional gains. Both the MP+SS and SS groups demonstrated significantly greater improvements, suggesting that the sensory input was a critical factor. The fact that the SS group showed similar improvements to the MP+SS group suggests that SS may be useful in individuals who are not able to participate in the intense physical activity, such as that required in activity-based interventions. However, the SS group did not show significant changes in cortical excitability when compared to the MP and MP+SS groups, this suggests that massed practice is of great importance for maximal benefits to be achieved. The authors acknowledge that more subjects or a longer training period may have yielded more significant differences between the different intervention groups and the control groups.
Quasi-Experimental Design Studies
Only one study fulfilled the criteria for quasi-experimental design and it was related to upper extremity function (Beekhuizen & Field-Fote 2005). Again, Beekhuizen and Field-Fote used both neural and functional outcome measures to determine the efficacy of massed practice training and massed practice training combined with somatosensory peripheral nerve stimulation in a randomized trial with two experimental groups in individuals with motor incomplete (AIS C and D) tetraplegia (at or rostral to C7). There were two groups: one that received massed practice training and one that received a combination of massed practice training (MP) and somatosensory peripheral nerve stimulation (MP+SS). Ten subjects were randomly assigned to one of these two groups. Subjects were between 16 and 70 years of age, and were at least one-year post-SCI. Subjects in both groups received the assigned intervention 5 days per week, for 3 weeks. Each session lasted 2 hours.
Outcome measures included both neural and functional assessments. Neural outcome measures included the stimulus intensity to elicit a motor threshold response in the thenar muscle of the treated hand, and the motor-evoked potential amplitude at 20% over the motor threshold intensity, and measure of the maximal pinch grip force. Functional assessments included the WMFT and the JHFT.
The tasks for this study were the same as those reported in “Experimental Design Studies” (Beekhuizen & Field-Fote 2005). Subjects in the massed practice groups (MP, MP+SS) performed repetitive practice of the 14 tasks in each of the 5 categories that included: gross upper extremity movement, grip, grip with rotation, pinch, and pinch with rotation.
The somatosensory peripheral nerve stimulation group (MP+SS) consisted of simultaneous median nerve stimulation while the subjects performed the massed practice tasks. The protocol was similar to that used for stroke patients (Conforto et al. 2002). The electrical stimulation parameters were frequency of 10 Hz, on/off duty cycle of 500/500ms, and pulse duration of 1 ms, delivered at 1 Hz. First the intensity was increased until there was a visible contraction of the thumb muscle, and then decreased until there was no visible contraction observed, in order to activate large fiber somatosensory afferents.
There was no significant difference in the motor threshold measures between the two groups or in the motor evoked potential, although both groups demonstrated a decrease in the motor threshold for activation of the thenar muscles via cortical stimulation, suggesting an improved ability to activate the thenar muscle. Despite this lack of difference between groups in terms of neural changes, there was a difference in functional ability post-intervention. The MP+SS group experienced significant improvements between pre- and post-assessments of pinch grip force and also demonstrated significantly greater increase in strength than the MP group. Similarly, the MP+SS group also experienced significantly different pre- and post-test scores on the WMFT, whereas both the MP+SS and the MP groups demonstrated improvements post-test on the JHFT. The MP+SS group, however, demonstrated a greater improvement than the MP group on the JHFT. These results suggest that the augmentation with somatosensory input via peripheral stimulation of the median nerve during the task is a significant variable to be considered.
Descriptive Design Studies
While fourteen of the studies that employed a descriptive design were evaluating the efficacy of activity-based intervention for lower extremity, only one case reported the effects of ABint effects on neural changes and functional outcomes after upper extremity training (Hoffman & Field-Fote 2007). Hoffman and Field-Fote evaluated neural improvements after bimanual massed practice training in an adult male with C6 tetraplegia. Neural outcomes included evaluation of somatosensory function, strength, and changes in the cortical map and motor threshold with transmagnetic stimulation (TMS) following training. Somatosensory function was assessed using the ASIA sensory scale for the upper extremity and the Semmes-Weinstein monofilament test to the region of the median nerve. The ASIA motor test was used to assess strength in the upper extremity. Hand function was also assessed, using the JTHFT, and bimanual skills were measured using the Chedoke Arm and Hand Activity Inventory.
The subject participated in a bimanual massed practice task in conjunction with somatosensory peripheral nerve stimulation of the median nerve, for 2 hours per day, 5 days a week, for 3 weeks. The protocol described previously in “Quasi-Experimental Designs” (Beekhuizen & Field-Fote) were employed for this study as well, at a stimulus intensity just below that which would elicit a movement in muscles innervated by the median nerve. Both arms were used throughout the intervention, using modifications for bimanual activity of the same tasks also described previously (Beekhuizen & Fote). The movement categories were again grasp, grasp with rotation, pinch, pinch with rotation, and finger isolation, with 14 tasks in each category. While both limbs were used to perform the tasks, only the right median nerve was stimulated during performance of these tasks. This allowed outcomes to be compared between the two sides at the completion of the intervention.
There were sensory changes, as demonstrated in the ASIA motor exam as well as the monofilament testing, in the right upper extremity and not the left post-training. Specifically, the subject demonstrated improvements from absent or impaired to normal in 4 dermatomes on the right side, for light touch and pin prick, but only in one dermatome on the left side. Responses to the monofilament testing improved by 2 monofilament diameters on the right side, and not at all on the left side. Although the motor threshold for the thenar muscle did not change with TMS, the area of active sites for the right biceps muscle increased by 9 cm2 , and the normalized map volume increased by 7.5 cm3, and the center of the map also shifted in the brain. Thus, this intervention of bimanual training in conjunction with somatosensory peripheral nerve stimulation led to neural changes. Similarly, the subject also demonstrated statistically significant improvements on the hand function tests. Only the triceps muscle increased in strength post-intervention, but there was no difference between the two sides. These changes suggest that ABint in individuals with incomplete SCI can improve neural function at both spinal and supraspinal levels of the neural axis.
The findings from experimental, quasi-experimental and descriptive design studies in the upper extremity suggest that while there are few studies exploring the neural and/or functional benefits of activity-based interventions to improve lower extremity and locomotor function, there are even fewer evaluating the efficacy of ABint for improving neural or functional outcomes in the upper extremities post-SCI. The findings from these studies suggest that:
- Adults with motor incomplete SCI (AIS C or D), regardless of the level of injury, may experience improved neural and upper extremity functional outcomes following ABint that are task specific (i.e. reaching activities for upper extremity function);
- The common characteristics for upper extremity interventions that improved neural outcomes in adults are high intensity and task specificity:
- The role of afferent input appears critical, but it is not yet clear in what manner, and to what degree, for improving neural function related to upper extremity activity, after SCI;
- Randomized controlled trials have not yet been completed in children with SCI related to upper extremity function;
- Upper extremity ABint appears to improve ASIA sensory scores in adults with AIS C SCI, however it remains unclear how this relates to function.
- The outcomes reported in these descriptive studies should be confirmed or rejected with larger scale studies in order to determine the neural effects of upper extremity ABint individuals with either acute or chronic injury,
Summary and Conclusions Related to the Efficacy of Activity-Based Interventions for Improving Neural Outcomes after SCI
Despite the fact that there are few randomized controlled trials in SCI, and specifically related to the efficacy of activity-based interventions (ABint) for individuals after spinal cord injury (SCI), there is evidence reported from studies using rigorous approaches with quasi-experimental and descriptive research methods, substantiating the use of ABint to improve function after SCI in some individuals with incomplete SCI. Questions remain as to who, specifically (level of injury, completeness, chronicity, etc.), can improve function after incomplete SCI, and what the appropriate intervention is for any given individual with SCI. Of further note is the fact that none of this research was performed in individuals with complete (AIS A) SCI, and only a small number report findings in individuals with motor incomplete (AIS B) SCI. Most likely, this is in part due to the fact that, in animal models of SCI and humans, there has been essentially no demonstration of neural recovery or substantial functional recovery after complete SCI. This does not mean that the nervous system does not have the capacity to change in individuals with complete SCI, or that improvements in function can not happen given the adequate program or treatment. Future study is warranted to more fully explore mechanisms for accessing neural plasticity in individuals with AIS A (complete) SCI, and to determine if there is a way to promote improved walking or hand function using ABint after SCI.
The common characteristics of both upper and lower extremity ABint that improved either neural or functional outcomes, or both, are that the ABint were delivered with high intensity and were, for the most part, task specific. However, a variety of dosages, frequency and duration of treatment interventions have been employed, in individuals with different levels, degrees, completeness, and chronicity of SCI, making it difficult to compare across studies. Based on these studies, further research is warranted. Future studies should focus on examining the effects of different dosages, frequencies and duration of ABint in a variety of individuals with different levels, degrees (AIS classification), and chronicity of SCI in both the upper and lower extremities. Consumers related to SCI rehabilitation and research should continue to monitor the evidence regarding the efficacy of ABint for SCI to determine if any given program or approach is warranted for any given individual with SCI. This will lead to more realistic expectations on the part of the patient with SCI and their loved ones and caregivers, more creativity on the part of clinicians to incorporate the ABint appropriate to any given individual with SCI, and perhaps more agreement and reimbursement of such programs by the payers.
The activity-based interventions study group would like to thank the staff of NIDRR for their support of this undertaking including our project officer, Pimjai Sudsawad. In addition, we would like to thank the principal investigators, E. Sally Rogers and Marianne Farkas for their guidance and support, as well as Megan Kash for development and assistance with the database and analysis tools.
Jennith Bernstein, PT
Amanda Gillot, PT
Jennifer Huggins, OT
Ashley Kim, OT
We also appreciate the time and effort of the experts who reviewed our list of references to determine if all relevant articles were included in this review:
Michelle Basso, PT, EdD
Jan Black, MSPT
Andrea Behrman, PhD, PT
Edelle Field-Fote, PhD, PT
Susie Harkema, PhD
Joe Hidler, PhD
George T. Hornby, PhD, PT
Sarah Morrison, PT
Keith Tansey, MD, PhD
Candy Tefertiller, MPT, ATP, NCS
Leslie VanHiel, MSPT
Statement Concerning Conflict of Interest
No study group member has a conflict of interest in this area of SCI research. Shepherd Center has one grant also funded by NIDRR related to evaluating the efficacy of a post-acute activity-based program for individuals with incomplete tetraplegia. However, no member of the study group has a fiduciary interest in the delivery of activity-based interventions for individuals with SCI.
Andreasen DS, Allen SK, Guthrie KB, Jennings BR, Sprigle SH (2004). “Exoskeleton for Forearm Pronation and Supination Rehabilitation”, IEEE Engineering in Medicine and Biology Society Conference.
Beekhuizen KS, Field-Fote EC (2005). Massed practice versus massed practice with stimulation: Effects on upper extremity function and cortical plasticity in individuals with incomplete spinal cord injury. Neurorehabilitation Neural Repair 19: 1.
Beekhuizen KS, Field-Fote EC (2008). Sensory stimulation augments the effects of massed practice training in persons with tetraplegia. Arch Physical Medicine Rehabil, 89: 4.
Behrman A. L., Lawless-Dixon A. R. L., Davis S. B., et al. (2005). Locomotor training and outcomes after incomplete spinal cord injury. Physical therapy, 85: 12.
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 non-ambulatory child with chronic, severe, incomplete cervical spinal cord injury. Physical Therapy, 88: 6.
Behrman AL, Harkema SJ (2000). Locomotor training after human spinal cord injury: A series of case studies. Physical therapy, 80:7.
Butefisch C, Hummelsheim H, Denzler P, Mauritz KH. (1995). “Repetitive training of isolated movements improves the outcome of motor rehabilitation of the centrally paretic hand”. J Neurol Sci 130: 59 – 68.
Castro MJ, Apple DF Jr, Hillegass EA and Dudley GA. (1999). Influence of complete spinal cord injury on SM morphology within six months of injury. Eur. J. Appl. Physiol. 80:373-378.
Castro MJ, Apple DF Jr, Rogers S and Dudley GA. (2000). Influence of complete spinal cord injury on SM mechanics within six months of injury. Eur. J. Appl. Physiol. 81:128-131.
Conforto AB, Kaelin-Lang A, Cohen LG. (2002). Increase in hand muscle strength of stroke
patients after somatosensory stimulation. Ann Neurol. 51:122–125.
Cozens JA (1999) “Robotic assistance of an active upper limb exercise in neurologically impaired patients”, IEEE Transactions on Rehabilitation Engineering. Evaluation of an Innovative Activity-Based Program for Individuals with SCI.
Dobkin, B., Apple, D., Bareau, H., Basso, M., Behrman, A., Saulino, M., Scott, M., & the Spinal Cord Injury Locomotor Trial (SCILT) Group, (2006). Weight-supported treadmill vs. over-ground training for walking after acute incomplete SCI. Neurology, 66: 6.
Edgerton VR, and Roy RR. (2002). Paralysis recovery in humans and model systems. Curr. Opinion Neurobiol. 12:658-667.
Edgerton VR, de Leon RD, Harkema SJ, et al. (2001). Retraining the injured spinal cord. J Physiol. 15;533(Pt 1):15-22.
Edgerton VR, Tillakaratne NJT, Bigbee AJ, de Leon RD and Roy RR. (2004). Plasticity of the spinal circuitry after injury. Ann. Rev. Neurosci. 27:145-167.
Field-Fote EC, Lindley SD, & Sherman AL (2005). Locomotor training approaches for individuals with spinal cord injury: A preliminary report of walking-related outcomes. Journal of Neurological Physical Therapy, 29: 6.
Gardner MB, Holden MK, Leikauskas JM, Richard RI (1998). Partial body weight support with treadmill locomotion to improve gait after incomplete spinal cord injury: a single-subject design. Phys Ther, 78:4.
Grasso R, Ivanenko YP, Zago M. et al. (2004). Distributed plasticity of locomotor pattern generators in spinal cord injured patients. Brain 127.
Gregory CM, Bowden MG, Jayaraman A, Shah P, et al. (2008). Resistance training and locomotor recovery after incomplete spinal cord injury: A case series. Spinal Cord, 45: 6.
Griffin L, Decker J, Hwang JY, Wang B, Kitchen K, Ding Z, & Ivy JL. (2008). Functional electrical stimulation cycling improves body composition, metabolic and neural factors in persons with spinal cord injury. Journal of Electromyogr Kinesiol.
Harkema SJ, Dobkin BH, Edgerton VR. (2000)Pattern generators in locomotion: Implications for recovery of walking after spinal cord injury. Topics in Spinal Cord Injury Rehabilitation, 2000; 82-96.
Harkema SJ. (2001) Neural plasticity after human spinal cord injury: Application of locomotor training to the rehabilitation of walking. Neuroscientist, 7(5):455-68.
Hicks AL, Martin KA, Ditor DS, Latimer AE, Craven C, Bugaresti J, and McCartney N. (2003). Long-term exercise training in persons with spinal cord injury: effects on strength, arm ergometry performance and psychological well-being. Spinal Cord, 41: 34-43.
Hicks AL, Adams MM, Martin K, et al. (2005). Long-term body-weight-supported treadmill training and subsequent follow-up in persons with chronic SCI: Effects on functional walking ability and measures of subjective well-being. Spinal Cord, 43: 6.
Hoffman LR, Field-Fote EC. (2007). Cortical reorganization following bimanual training and somatosensory stimulation in cervical spinal cord injury: A case report. Phys Ther, 87: 6.
Hornby TG, Zemon DH, & Campbell D. (2005). Robotic-assisted, body-weight-supported treadmill training in individuals following motor incomplete spinal cord injury. Phys Ther, 85: 6.
Kirshblum S, Millis S, McKinley W, Tulsky D. (2004). Late neurologic recovery after traumatic spinal cord injury. Arch Phys Med Rehabil 85(11): 1811-1817.
Krebs HI, Hogan N, Aisen ML, & Volpe BT. (1998) “Robot-Aided Neurorehabilitation”. IEEE Transactions on Rehabilitation Engineering.
Krebs HI, Volpe BT, Aisen ML, Hogan N. (2000) Increasing productivity and quality of care: Robot-aided neuro-rehabilitation. J Rehabil Res and Dev.
Lam T, Wirz M, Lunenburger L, & Dietz W. (2008). Nsci-82 Swing phase resistance enhance flexor muscle activity during treadmill locomotion in incomplete spinal cord injury. Neurorehabil Neural Repair, 25:6.
Maegele M, Mueller S, Wernig A, Edgerton VR, and Harkema SJ. (2002). Recruitment of spinal motor pools during voluntary movements versus stepping after human spinal cord injury. J Neurotrauma, 19(10):1217-29.
Mangold S, Keller T, Curt A, & Dietz V. (2005). Transcutaneous functional electrical stimulation for grasping in subjects with cervical spinal cord injury. Spinal Cord, 43: 6.
McDonald JW, Becker D, Sadowsky CL, Jane JA Sr, Conturo TE, Schultz LM. (2002). Late recovery following spinal cord injury: Case report and review of the literature. J Neurosurg: Spine, Sept:252-265.
McDonald JW, Sadowsky CL. (2002). Spinal Cord Injury. Lancet 359:417.
Modlesky CM, Bickel CS, Slade JM, Meyer RA, Cureton KJ and Dudley GA. (2004). Assessment of skeletal muscle mass in men with spinal cord injury using dual-energy X-ray absorptiometry and magnetic resonance imaging. J Appl Physiol. 96(2):561-5.
Perez MA, Floeter MK, Field-Fote EC. (2004). Repetitive sensory input increases reciprocal Ia inhibition in individuals with incomplete spinal cord injury. J Neurol Phys Ther, 28(3): 114 – 121.
Prosser LA. (2007). Locomotor training within an impatient rehabilitation program after pediatric incomplete spinal cord injury. Phys Ther, 87: 6.
Prostas EJ, Holmes A, Qureshy H, Johnson A, et. Al. (2001). Supported treadmill ambulation training after spinal cord injury: A pilot study. Arch Phys Med Rehabil, 82.
Steeves JD, Lammertse D, Curt D, Fawcett JW, Tuszynski MH, Ditunno JF, Ellaway PH, Fehlings MG, Guest JD, Kleitman N, Bartlett PF, Blight AR, Dietz V, Dobkin BH, Grossman BR, Short D, Nakamura M, Coleman WP, Gaviria M, Privat A. (2006). Guidelines for the conduct of clinical trials for spinal cord injury (SCI) as developed by the ICCP panel: clinical trial outcome measures. Spinal Cord 45: 206–221.
Thrasher TA, Flett HM, & Popovic MR. (2005). Gait training regimen for incomplete spinal cord injury using functional electrical stimulation. Spinal Cord ,4: 6.
Trimble MH, Kukulka CG, Behrman AL. (1998). The effect of treadmill gait training on low-frequency depression of the soleus H-reflex: comparison of a spinal cord injured man to normal subjects. Neurosci Letter 246: 3.
Trimble MH, Behrman AL, Flynn SM. (2001). Acute effects of locomotor training on overground walking speed and H-reflex modulation in individuals with incomplete spinal cord injury. J Spinal Cord Med. 24: 2.
Van der Lee JH, Wagenaar RC, Lankhorst GJ, Vogelaar TW, Deville WL, Bouter LM. (1999). Forced use of the upper extremity in chronic stroke patients: results from a single-blind randomized clinical trial, Stroke, 30:2369-75.
Volpe BT, Krebs HI, Hogan N, Edelstein OL, Diels C, Aisen M.(2000). A novel approach to stroke rehabilitation; robot-aided sensorimotor stimulation. Neurology 54: 1938-44.
Winchester, P., McColl, R., Querry, R., Foreman, N., Mosby, J., Tansey, K., & Williamson, J. (2005). Changes in supraspinal activation patterns following robotic locomotor therapy in motor-incomplete. Neurohabil Neural Repair, 19: 6.
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. Arch Phys Med Rehabil, 86: 4.