Approximately 1,275,000 individuals in the United States and 41,000 individuals in Canada are living with the consequences of SCI. The incidence of SCI is estimated to be over 12,000 new cases annually in the United States and 1,400 new cases in Canada (Gibson et al., 2009). The average age at injury today is 48 years; it was 40 years in 2005 and 29 years between 1973 and 1979. Enhanced emphasis on emergency medical care at the site of the accident and other interventions has resulted in a change in the extent of SCI: previously most injuries were complete, but now the majority are incomplete -injuries. Statistics have shown that 52.6% of injuries are incomplete (tetraplegia, 34%; paraplegia, 19%), while only 47% are complete injuries (tetraplegia, 18%; paraplegia, 23%). Methylprednisolone was introduced in the 1990s as a drug to minimize the secondary effects of injury, and although its efficacy is still being debated, it is considered to have contributed to a shift towards improved recovery after SCI. This shift has increased the potential for walking recovery with sparing of sensorimotor function below the level (p. 6 ) of the lesion (Stover et al., 1999); however, physical rehabilitation has remained focused on compensation.

Functional Deficits After Spinal Cord Injury

There are many functional deficits in the locomotion of individuals following SCI. For clinically complete and severely injured incomplete individuals, locomotion has been unattainable. Individuals with motor complete SCI exhibit paralysis of the musculature below the level of the lesion. They are unable to voluntarily activate trunk and leg muscles below the spinal cord lesion. Deficits or complete loss of sensation and proprioception disrupt sensory feedback concerning posture and limb position, pressure, or tactile information. Restoration of pre-injury ambulation ability and standing is not expected and is not a goal of rehabilitation. Passive standing and mobility are achieved by alternative strategies and devices.

Though individuals with motor incomplete SCI have greater potential to achieve walking, few individuals achieve a full return to normal ambulatory function. Those who do achieve ambulation walk with a slow, asymmetrical gait pattern, have balance difficulties, depend on assistive devices for support, and use braces to compensate for weakness or paralysis. The first striking difference observed in the gait of ambulatory individuals with incomplete SCI is reduced walking speed. In a retrospective review, speed data were pooled from 162 individuals with incomplete SCI across 20 studies (Barbeau et al., 1998), showing a wide spectrum of walking speed abilities, ranging from total incapacity to near-normal speed. Next, the kinematic pattern at the trunk, hip, knee, and ankle joints reveals a different profile between individuals with SCI and those without injury when both groups are compared at their natural walking speeds (Pepin et al., 2003a, 2003b). Those with incomplete SCI walk with more ankle dorsiflexion during the initial double-stance period, with less plantarflexion during the push-off component. Flexion of the knee at initial contact is accentuated, although it is reduced during swing (Barbeau et al., 1999). Besides being of lesser amplitude, the peak ankle plantarflexion during push-off and the peak knee flexion during swing are reached later. Finally, in the SCI population, the maximal hip extension tends to be less.

Joint stiffness, including the relative contribution of the stretch reflex and the intrinsic properties of the ankle extensors in normal and spastic subjects, has been studied by several groups (Dietz et al., 1981; Dietz & Berger, 1983; Sinkjaer & Magnussen, 1994; Thilmann et al., 1991). An increase in the reflex gain and a decrease in inhibition during the swing phase have been reported in individuals with incomplete SCI (Fung & Barbeau, 1994; Sinkjaer et al., 1996; Yang et al., 1991). These findings suggest that increases in both the reflex gain and the non-reflex torque could contribute to the increased stiffness of the ankle joint seen in individuals with incomplete SCI during walking. Hence, both alterations in central mechanisms and changes in intrinsic properties of the muscle fibers could be responsible for the increased stiffness and for the decreased walking speed (Mirbagheri et al., 2001).

In addition, a comparison of the kinetics (moment and mechanical power) of individuals with incomplete SCI and of non-injured individuals walking at natural speeds presents significant differences at the ankle, knee, and hip joints. The individuals with SCI have smaller plantarflexor moments generated at the ankle joint during terminal (p. 7 ) stance, and the use of short leg braces accentuates this weakness. At the knee joint, individuals with incomplete SCI have a higher flexion moment at initial contact and during terminal stance. On the other hand, they present a lower extension moment during mid-stance. Also, individuals with incomplete SCI exhibit an increase in the flexor moment at the hip during mid-stance and a decrease in hip extensor moment during terminal stance. This results in an important difference (decrease) in energy generation. At the knee joint, lower energy absorption during loading response is also observed (Nadeau et al., 1999).

The electromyogram (EMG) of lower limb muscles during walking reveals alterations in both timing and amplitude in individuals with SCI compared to uninjured individuals (Fung & Barbeau, 1994). Co-activation of muscle activity at proximal and distal joints is often reported in individuals with incomplete SCI (Fung & Barbeau, 1994). Further abnormal activation of the soleus muscle, including a broadened and flattened EMG profile with an early activation and clonus during the stance phase, is commonly seen in early stance.

In summary, individuals with incomplete SCI exhibit dysfunction in most of the critical events of the gait cycle. Weakness of the knee extensors modifies the gait cycle during the loading phase and weakness of the ankle plantarflexors corresponds to changes in terminal stance and pre-swing phases (Bajd et al., 1997). The initial and mid-swing phases are affected by changes in the passive and reflex stiffness of the ankle plantarflexors as well as weakness of the ankle dorsiflexors (Dietz et al., 1981). Finally, terminal swing is affected by a decrease in angular velocity generated by the weakness of the hip flexors. This resultant behavior is a product of intersegmental dynamics that are dependent on the velocity of the movement (Hoy & Zernicke, 1986; Wisleder et al., 1990).

In humans, uphill walking is a demanding task that requires specific modifications in the trunk and the pelvis and in lower limb movements (Lange et al., 1996; Wall et al., 1981) and muscle activation patterns (Brandell, 1977; Lange et al., 1996; Simonsen et al., 1995). As the walking grade increases, more propulsion has to be generated from the lower limbs (Brandell, 1977) and postural adjustments must be performed to maintain equilibrium. Following SCI, the basic locomotor pattern is altered (Barbeau et al., 1998), and the ability to adapt to changes in the environment could be affected. For example, increasing the treadmill slope from 0 to 15 degrees induces a gradual increase in hip and knee flexion and in ankle dorsiflexion from late swing through the stance phase, and the vertical displacement of the greater trochanter is stable across grades in all uninjured subjects. In individuals with incomplete SCI, however, lower limb motions do not show consistent patterns of adaptation. Individuals who are unable to achieve the steepest slope (15%) show only increased hip flexion in early stance combined with elevation of the greater trochanter (hip hiking) during the swing phase in all walking conditions. This hip hiking is used, in part, to compensate for the decrease or the absence of the tibialis anterior activity during the swing phase (Leroux et al., 1999).

Researchers also investigated the contribution of the trunk and pelvis during inclined walking in uninjured subjects and those with incomplete SCI (Leroux et al., 2002, 2006). Briefly, their results show the importance of trunk and pelvic segments in the postural adaptation to inclined walking. Trunk and pelvic movements do not seem to participate directly in the generation or absorption of energy required for walking up and down slopes. Total angular excursions are consistent across walking grades in sagittal and transverse planes and vary to a small degree in the frontal plane. However, modifications in (p. 8 ) trunk and pelvic vertical alignment allowed lower limbs to perform the most efficient patterns of movements during uphill and downhill walking. Thus, the researchers propose that trunk and pelvic postural modifications likely assist lower limbs when adapting to inclined walking (Ladouceur et al., 2003; Leroux et al., 2002, 2006). In SCI individuals the inability to control the trunk and pelvis contributes to the difficulty in successfully negotiating an incline.

Stepping over an obstacle is another locomotor task with greater demands and challenges to postural control when compared to steady-state walking on level ground. The anticipatory locomotor adjustments exhibited when stepping over obstacles shows a spectrum of adaptation capacities similar to walking speed and slope. The greater trochanter during quiet standing was used as a reference to measure the relative vertical displacements. Uninjured subjects increased both hip and knee flexion during early swing when encountering 5- and 30-mm obstacles. At the opposite end of the spectrum of adaptations, individuals with incomplete SCI are typically incapable of clearing small obstacles of only 5 mm. They do not adapt the trajectory of the fifth metatarsal during the obstructed condition (30 mm), showing that the unobstructed foot trajectory is insufficient to clear such a low-height obstacle. However, there are changes in the locomotor pattern with an elevation of the greater trochanter and a decrease in knee flexion at the initiation of the swing phase (Ladouceur et al., 2003). Therefore, different compensation strategies can be adopted by individuals with incomplete SCI, such as elevating the greater trochanter to adapt to increased speed, or increasing hip circumduction and/or knee flexion to clear obstacles.

Examples of other functional changes in the musculoskeletal system following neurologic insult are increased stiffness of the passive components of the ankle joint (Mirbagheri et al., 2001, 2002), increased fatigability and modification of the biochemical properties of motor units (Cope et al., 1986; Stein et al., 1991; Veltink et al., 2000; Yang et al., 1990), and a higher incidence of osteoporosis (Demirel et al., 1998; Garland et al., 1992; Wilmet et al., 1995). Furthermore, depending upon the level of the injury, there is a modification of the autonomic regulation system, including adaptations of the circulatory system and bladder, digestive, and sexual function (Hooker et al., 1993; Raymond et al., 1997; Yamamoto et al., 1999). SCI is also associated with changes within the central nervous system, such as weakness, hyperactive spinal reflexes (Dietz et al., 1981; Sinkjaer et al., 1996), muscle co-activation, and loss of sensory function (Fung & Barbeua, 1994; Sinkjaer et al., 1996; Yang et al., 1991). For individuals with clinically complete injury these complications tend to be more severe and debilitating, and after incomplete injuries these modifications contribute to deficits in walking and to postural problems related to bearing weight, maintaining balance, and developing propulsion.

Physical Rehabilitation After Spinal Cord Injury

The key factors that are considered before initiation of standard gait rehabilitation for individuals with SCI are voluntary motor control, range of motion, muscle tone, sensation, functional abilities, posture, skin integrity, and autonomic function (Behrman et al., 2009). The assessed voluntary function of the 10 key lower limb muscles (e.g., left and right hip flexors, knee extensors, ankle dorsiflexors, long toe extensors, and ankle plantarflexors) is used as a key predictor for potential of recovery of locomotion. In addition, (p. 9 ) upper limb strength is assessed for the availability of controlling balance assist and support devices. This assessment and the subsequent interpretation by the rehabilitation specialist usually determines the approach to reaching ambulatory functional goals and the level of effort that will be given toward neuromuscular recovery versus compensation approaches.

Historically, the predominant approach for rehabilitation for persons with SCI has been to compensate for the neuromuscular deficits caused by the injury (Halstead & Grimby, 1995; Sadowsky & McDonald, 2009; Whiteneck et al., 2009). Upper limbs can be used to compensate for the lack of trunk control and leg paralysis, allowing individuals with SCI to move their own body and to perform activities of daily living. Therapy targets strengthening the muscles above the level of the lesion. Leg braces are used to provide support when muscles are weak or paralyzed. For persons with complete or severe incomplete SCI, bracing is used to fully support the leg and sometimes assist at the hip joint and pelvis (e.g., long leg braces, LSU reciprocating gait orthoses). Strategies to ambulate include vaulting over supported limbs while weight-bearing through the arms with crutches using a swing-through gait pattern.

Assistive devices have been traditionally used to accomplish mobility (Sisto et al., 2008). A consequence of the many altered features of gait in individuals with incomplete SCI is an increase in the energy requirement necessary for ambulation, an issue for consideration when deciding on an assistive device. Melis and colleagues (1999) showed that individuals using walkers tend to walk slower than crutch users, who tend to walk slower than cane users. The opposite relationship exists for the maximal amount of force exerted on a walking aid during gait: walker users tend to place the most amount of force on their aid, cane users the least. These results suggest that walking speed might be related to maximal axial force, or that the limiting factor in the speed of an individual’s gait might be, in fact, the type of walking aid itself. More recently, functional electrical stimulation (FES) has developed as an alternative orthosis, compensating for lack of voluntary activation of leg muscles for standing and walking (Bogataj et al., 1995; Ditunno & Scivoletto, 2009; Kirshblum, 2004; Stein et al., 1993). The FES orthosis can be used to provide dorsiflexion during the swing phase of the gait cycle or can provide gross flexion and extension when used as bilateral leg orthoses in lieu of long leg braces (e.g., ParaStep System). When these devices are removed, there is no therapeutic effect on the user’s ability to move his or her legs or walk more easily. Thus, these devices achieve a mobility goal but do not restore neuromuscular function.

The wheelchair, manual or power, remains an optional and alternative means of daily mobility. Where wheelchair propulsion approximates the energy requirement of normal walking (Blessey, 1978), the energy cost of walking by individuals with incomplete SCI is higher than that for speed-matched walking by able-bodied participants (Stein et al., 1993), making upright mobility aids much more demanding. Movement strategies that take advantage of principles of physics (such as levers and momentum) and substitution via alternative muscle use (Behrman et al., 2006; Sisto et al., 2008; Somers, 2001) provide the basis for accomplishing everyday tasks (e.g., dressing, transferring in and out of bed, pressure relief, rolling over in bed). Individuals do not recover the ability to perform everyday tasks as performed prior to injury but are trained to accomplish tasks using a new behavioral strategy such as using an alternative body segment, an external aid, or physical assistance (Barbeau et al., 2006; Kleim, 2006). The reader is referred to the Spinal Cord Injury Rehabilitation Evidence, volume 2, for an up-to-date review of (p. 10 ) interventions (Chapter 7: Lower Limb Rehabilitation Following Spinal Cord Injury; http://www.scireproject.com/home.php).

Stroke is defined as a focal cerebrovascular event in which sudden loss of brain function caused by interruption of the flow of blood to the brain or by rupture of blood vessels in the brain persists beyond 24 hours. In the United States alone, over 795,000 individuals suffer a first-time or recurrent stroke annually (Lloyd-Jones et al., 2010). Stroke ranks as the seventh highest cause of burden of disease worldwide in terms of disability-adjusted life years and as the single most important cause of severe disability in people living at home (Lopez & Mathers, 2006). Of those who survive a stroke annually, 73% will incur subsequent disability. In 75% of the post-stroke population, walking dysfunction is a significant contributor to post-stroke disability and is associated with sensorimotor deficits and hemiparesis (Duncan, 2007).

Functional Deficits After Stroke

An impaired ability to walk is a significant contributor to long-term disability and burden of care after stroke. Approximately one third of people surviving acute stroke are unable to walk 3 months after admission to a general hospital (Wade et al., 1987). Those who do achieve independent walking are still limited in community ambulation, with motor impairments contributing to balance dysfunction post-stroke (Jorgensen et al., 1995). Individuals post-stroke with mild to moderate motor impairments who achieve ambulatory status are at significantly greater risk for falls (Forster & Young, 1995; Keenan et al., 1984). Of those who are ambulatory, 40% will have severe impairments including balance deficits, limiting ambulation to household status. With community ambulation, the risk of falls is 73% within 6 months after the stroke. The occurrence of a fall within this period compounds the risk for further fall incidence fourfold.

The gait pattern of individuals who have sustained a stroke was thoroughly described in a review by Olney and colleagues (1998). The two immediate impairments of most significance to gait performance are diminished strength, or the inability to generate voluntary muscle contractions of normal magnitude in any muscle groups, and inappropriately timed or inappropriately graded muscle activity. Reduced walking speed and longer stance phases have been observed for both the affected and unaffected lower limbs. Typically, the stance phase is longer in duration and occupies a greater proportion of the gait cycle on the unaffected side compared to the affected side. The third difference is a greater proportion of the gait cycle spent in double support in individuals post-stroke than that of able-bodied individuals walking at normal speeds. In addition, with hemi-paresis, variations in joint excursions include several deviations at initial contact and reduced excursions during swing (Olney et al., 1998). Examples include decreased hip flexion at initial contact, increased hip flexion at toe-off, and decreased hip flexion during mid-swing; more knee flexion at initial contact and less knee flexion at toe-off and mid-swing; and more ankle plantarflexion at initial contact and mid-swing and less ankle plantarflexion at toe-off.

(p. 11 ) Studies have classified the kinematic patterns of individuals post-stroke into subgroups (De Quervain et al., 1996; Sullivan et al., 2008) in combination with spatial temporal characteristics (Mulroy et al., 2003). Individuals with severe hemiparesis who walk slowly have been described by two kinematic patterns: extensor thrust or extended pattern and the buckling knee or flexed pattern. The extended pattern is characterized by increased ankle plantarflexion and knee hyperextension throughout the stance phase, with this pattern also dominating the flexor phase. Thus, insufficient ankle flexion and knee flexion are exhibited during swing. In contrast, the flexed pattern is characterized by increased ankle and knee flexion during stance with deficient hip extension prior to swing initiation. The walking kinematics of individuals with mild to moderate impairments and moderate to fast walking speeds are more comparable to those of healthy individuals. Moderate-speed walkers demonstrated a greater flexor pattern during stance (knee flexion) with decreased hip extension and plantarflexion at pre-swing compared to the faster walkers. Interestingly, the kinematics of the non-paretic leg also change after stroke. Such changes likely occur to compensate for the motor impairments of the paretic leg and may also reflect speed-dependent effects. Increased hip flexor moments were observed in late stance and were positively correlated to walking speed. Hip muscle activation may thus compensate for weak plantarflexors in some faster walkers post-stroke (Nadeau et al., 1999).

In comparison, when speeds are matched between post-stroke individuals and healthy controls, joint moment and power profiles are similar in pattern, yet differ in the amplitude (comparing hemiparetic to paretic legs and hemiparetic to healthy controls). Ankle joint plantarflexor power and moments are reduced or absent in the hemiparetic leg, in contrast to being equal or greater in the nonparetic legs. More recently, the -anterior–posterior ground reaction force has been used to quantify the contribution of the paretic leg to forward propulsion during walking (Bowden et al., 2008). The percentage of paretic leg propulsion differs across stroke severity: severe (16%), moderate (36%), and mild hemiparesis (49% of normal). This measure may serve to discriminate between recovery or restitution of limb function and compensation after a therapeutic intervention (Sullivan et al., 2008).

Little is known about the contribution of the lower limbs and upper portions of the body to the adaptation of the gait pattern in post-stroke individuals. During normal gait, the main functions of the pelvis and trunk are to maintain body equilibrium and achieve smooth locomotion (Saunders et al., 2004; Thorstensson et al., 1984; Thurston & Harris, 1983; Wall et al., 1981). With gait pathologies such as stroke, movements of the pelvis and trunk may be used to compensate for lower limb deficits and show excessive range of motion.

Post-stroke individuals exhibit difficulties adapting their locomotor pattern to change speed and to uphill walking (Leroux et al., 1999). Unlike healthy subjects, who show a clear pattern of adaptation at the hip, knee, and ankle joints, post-stroke subjects mainly use the hip when adapting to uphill inclines. While the activation of plantarflexor muscles increases in control subjects during uphill walking, it does not in post-stroke subjects, a modification likely related to a weak push-off. Thus, a compensatory mechanism from axial and proximal muscles may be needed when adapting to inclined walking.

Furthermore, these compensatory mechanisms may depend on the severity of sensory or motor impairments. More impaired post-stroke individuals may use pelvis and trunk movements to a greater extent to compensate for lower limb deficits and adapt to (p. 12 ) different inclines. For instance, subjects with unilateral hip pain due to osteoarthritis (and avascular necrosis) use larger pelvic and trunk movements to compensate for limited hip motion (Thurston, 1985). These subjects show an abnormal lateral elevation of the pelvis during the swing phase to compensate for limited hip flexion and adopt a toe-floor clearance similar to normal walking. Excessive rotations from the pelvis and trunk have also been reported in subjects with hip pain (Thurston, 1985) and in those with hemiplegia due to stroke (Wagenaar & Beek, 1992). These studies show that abnormal movements from the trunk and pelvis can be used to overcome lower limb deficits in pathologic gait.

Physical Rehabilitation After Stroke

There are a number of different approaches to physical therapy following stroke. Prior to the 1940s, these primarily consisted of corrective exercises based on orthopaedic principles related to the contraction and relaxation of muscles, with emphasis placed on regaining function by compensating with the unaffected limbs (Langhorne et al., 1996). In the 1950s and 1960s, techniques based on available knowledge were developed, including the methods of Bobath (Davies, 1985; Lennon & Ashburn, 2000; Wagenaar & Beek, 1992) and Rood (Goff, 1969) and the proprioceptive neuromuscular facilitation approach (Knott, 1968). These approaches targeted activating weak muscles and reducing spasticity after stroke to normalize movements. In the 1980s, the potential importance of neuropsychology and motor learning (Schmidt & Lee, 2005) was understood and incorporated into post-stroke rehabilitation (Sullivan et al., 2008; Turnbull, 1982; Winstein et al., 2003, 2007). Recent evidence is mounting for effective rehabilitation of walking post-stroke that includes task-specific training, resisted strengthening for the upper and lower extremities, and aerobic training (Sullivan et al., 2008).

Alternatively, impairment-based strategies have also demonstrated benefits. Resisted strength training has predominantly been viewed as inappropriate in post-stroke persons due to concern for increased spasticity and abnormal movement patterns. However, evidence now indicates that strengthening can improve function without increased spasticity or greater dyscoordination (Foley et al., 2009). Similarly, aerobic training has addressed the effects of inactivity and resultant deconditioning post-stroke, particularly in the chronic condition. Because post-stroke individuals have a greater risk for a second stroke, exercise is highly recommended to decrease their risk. Specific guidelines and monitoring of vital signs allow aerobic exercise to be safely integrated into a rehabilitation program for post-stroke individuals, with benefits for cardiovascular health and walking. Various modes of training have been used, such as stationary cycling, treadmill walking, water-based exercise, overground walking, and stair climbing (Foley et al., 2009). The reader is referred to the Canadian Stroke Network for ongoing systematic review of the evidence for lower extremity training and mobility post-stroke.

Physical rehabilitation is embarking on a new era requiring a paradigm shift in our thinking and clinical decision making. Neurorehabilitation, rather than relying primarily on (p. 13 ) traditional approaches based on the observation, skills, and assumptions from master clinicians (Adler et al., 1999; Bobath, 1979; Kleim, 2006; Knott, 1968), now emphasizes the important role of evidence-based medicine. Evidence-based practice is defined as a “conscientious, explicit and judicious use of best evidence in making decisions about individual patients” (Sackett, 1996). The practice of evidence-based medicine means integrating individual clinical expertise with the best available external clinical evidence from systematic research. Research evidence is one component of the clinical decision-making process and is combined with clinical expertise; client preferences, needs, and priorities; and available resources to result in the best practice and highest achievable level of recovery of function for each individual.

Implementing evidence-based practice requires accessing research findings, acquiring new knowledge, adopting new concepts, implementing new interventions, and performing objective evaluation. Academic programs in rehabilitation sciences led by multidisciplinary teams of researchers provide opportunities for the advancement of -evidence-based neurorehabilitation practices for improving recovery for individuals after neurologic injury and disease.

The translation of evidence into practice is a new and challenging approach for physical therapists (Jette, 2006). The application of evidence-based therapy to today’s academic programs, however, remains relatively new to many practitioners. In reviewing the evidence for a designated therapeutic intervention, it is important to distinguish the essential components that are different from other approaches. The evidence for much of neurorehabilitation practice is limited but has been disseminated in the clinical literature as summarized reviews of evidence and in the outcomes of such historical conferences such as NU-STEP (Northwestern University Special Therapeutic Exercise Project), II STEP, and III STEP (Callahan et al., 2006) to inform clinical decision making and -practice.

Evidence-Based, Restorative Strategies for Rehabilitation After Neurologic Insult: Locomotor Training

New knowledge of the neurobiological control of walking and the plasticity of the nervous system in response to repetitive activity is reshaping the direction of physical rehabilitation after neurologic injury (Callahan et al., 2006). Partnerships of basic scientists, clinicians, and applied scientists, as exemplified in the authorship of this text, have resulted in translation of such knowledge from the laboratory to the human condition and formed the basis for activity-based therapies (Behrman & Harkema, 2008; Dromerick et al., 2006) such as Locomotor Training.

Several reviews of Locomotor Training provide detailed information and interpretations of the literature relative to clinical decision making and are recommended to the reader (Behrman et al., 2006; Lam et al., 2008; Mehrholz et al., 2008). In reviewing the literature, consideration should be given to the specific patient population, severity of injury, and time since injury. Not all neurologic patient populations will be represented in the literature, and some may actually never be tested within a randomized clinical trial or cohort studies because of their low prevalence in society. The good clinical judgment and expertise of the clinician, in combination with the literature, will allow for decision making to benefit subpopulations of clients with injury or disease. Publishing such case (p. 14 ) studies or case series may be very informative when gold standard randomized clinical trials examining large populations are not feasible and also may provide a valuable resource for clinicians implementing evidence-based practice.

In addition, when reviewing the evidence, the meaningfulness and relevance of outcome assessments for recovery of walking function should be considered carefully (Bowden et al., 2008; Lam et al., 2008; Mehrholz et al., 2008). For instance, while gait speed is an important outcome with functional relevance (Perry et al., 1995), change to a less restrictive assistive device (e.g., from a rolling walker to bilateral forearm crutches) may be accompanied by an initial decrease in gait speed or even an outcome with a slower gait speed, but it still indicates recovery. Even improvements in walking speed reported as large-percentage changes may not reflect meaningful changes in speed (e.g., a speed change from 0.03 m/s to 0.06 m/s is a 100% change) or result in functional walking recovery (Behrman & Harkema, 2007). Last, improvements in gait speed may be accomplished by a persistent and even enhanced compensated gait pattern (e.g., greater and faster hip hike and elevation to clear a foot during advancement of a limb as “swing”) as opposed to neuromuscular recovery (i.e., a more upright trunk with increased hip extension and less upper extremity weight-bearing).

When reading the literature, clinicians will need to carefully review the methodology to understand the critical elements of the intervention and the variances among those presented. A case in point may best be seen by reviewing an article (Vidoni et al., 2008) reporting an intervention using a body weight support on the treadmill (BWST) that is accompanied by an online video of the intervention (http://www.jnptextra.org/pvideos.cfm#2; June 2008, video 1, BWST training, and September 2006, Video Limb Kinematics during Treadmill Walking). A quick review of this video noting “BWST” in the descriptors of the therapy will discern two very different approaches to a rehabilitation protocol with the aim of improving walking after stroke. While use of a BWST is a common denominator and the described interventions exhibit some similarity, the video representations of the training differ significantly from one another. Thus, the interventions, although noted as “BWST training,” are not the same intervention. While the published visual aid for these two articles certainly provided clear differences, the reader of the literature more often is required to detect such differences from the written, published article alone. This remains a challenge but is a necessary step in conducting evidence-based practice. That said, the question remains as to what the active and critical ingredients in a therapy are as described in the literature. Such active ingredients may include the emphasis on specific and integrated afferent input (body weight load, treadmill speed, arm swing), manual- or robotic-assisted training, inclusion of overground training, duration and intensity of training, translation/integration beyond the clinic, and progression strategies.

In addition, while task-specific retraining is considered for stroke rehabilitation, clinical practice guidelines have not been specifically developed. Clearly identifying the active ingredients and exercise intensity of a therapeutic intervention will enhance the ability to assess and compare therapeutic interventions. While a recent Cochrane review (Moseley et al., 2005) concluded that there is inconclusive evidence that task-specific treadmill or body weight–supported treadmill training is effective after stroke, the heterogeneity of the population and variable training paradigms reported in the literature likely make comparisons difficult to interpret. Clinical efficacy and ongoing clinical trials (p. 15 ) (Duncan, 2007, Pohl et al., 2002; Sullivan et al., 2002, 2007) should continue to pinpoint who will benefit relative to severity, when an intervention is beneficial relative to stroke onset, and training intensity (frequency and duration).

A further example of implementing evidence-based practice in the clinic is the Christopher and Dana Reeve Foundation NeuroRecovery Network (Harkema et al., 2011a b). The network, a set of seven clinical sites in the United States, has adopted Locomotor Training as a standardized activity-based therapy for rehabilitation of individuals with motor incomplete SCI. A standardized protocol is implemented and standardized outcomes are evaluated periodically throughout the intervention. This process allows for reassessment of the therapeutic program, its intensity and duration, and its effectiveness with specific populations and ultimately affords further knowledge that can be applied to clinical decision making. While the merit and benefit of activity-based therapies continue to be assessed in specific patient populations, the NeuroRecovery Network also addresses the challenges clinicians face to meet the therapeutic demands necessary to promote learning and behavioral change after neurologic injury. Evaluation of the outcomes from this unique programmatic venture into rehabilitation and healthcare for providing activity-based therapies will guide future clinical decision making for responders and non-responders and thus patient selection, selection of outcomes, and development of outcome measures to facilitate dissemination of new discoveries into everyday clinical practice.