Introduction

Stroke remains a leading cause of long-term disability in the United States at a cost of $38 billion per year. About 650,000 persons survive a new stroke yearly and 7 million Americans live with the complications of stroke [1]. Despite evidence that participation in formal rehabilitative therapies lessens disability after stroke [2], less than a third receive inpatient or outpatient therapies [3]. Of those who do access therapies, the frequency of use varies by geographic location and socioeconomic status. For these patients, the amount of rehabilitation available has progressively fallen as subacute stroke inpatient stays have dropped to an average of less than 16 days and as Medicare has capped the number of outpatient therapy sessions to 15/year [4•]. In effect, these declines in service may limit rehabilitation gains and place greater burdens on caregivers. In contrast to these fiscally driven realities, the science underlying stroke rehabilitation offers new directions to improve outcomes.

Scientific advances based on animal models have sharpened our understanding of the genetic, molecular, physiologic, cellular, and behavioral adaptations that drive and may limit the recovery of function [5]. Novel types of therapies based on manipulating mechanisms of learning and memory, neurogenesis and axonal regeneration, and neurotransmitters and growth factors can facilitate the recovery process in models. In patients, non-invasive modalities including functional and structural magnetic resonance imaging (MRI) and neuronal excitatory and inhibitory stimulation tools such as transcranial magnetic stimulation (TMS) are characterizing changes in connectivity between brain regions after stroke [6]. Therapeutic strategies for patients are also being drawn from engineering and computer science. Wireless health and communication technologies have produced wearable sensors to remotely record the quality and quantity of walking practice, smartphone apps to cue practice, and tele-rehabilitation programs to enable treatment in the home or community [7].

Evidence from adequately powered, randomized control trials demonstrating the efficacy of new interventions when compared to existing therapies has been far outpaced by the number of novel strategies being developed. Trials can be confounded by the patho-anatomic and functional heterogeneity of patients, the complexity and cost of delivering an intervention, and uncertainties regarding optimal therapy timing, dose, and duration [8]. Additionally, the outcome measures used in trials are often relative surrogates for patient performance rather than direct measures of the types, quantity, and quality of physical functioning [9]. When the goal is to assess the use of the upper extremity, walking, exercise, and participation in home and community activities [10], existing measures may not fully capture clinically important changes in physical or cognitive impairments, disability, or health-related quality of life and participation [11]. Despite these confounds, recent trials do provide useful evidence about behavioral, pharmacologic, and neurostimulation treatments for stroke, as well as near future hope for biological interventions for the most highly impaired patients.

Overview of Care in Rehabilitation

Patients are admitted for inpatient stroke rehabilitation usually because they are unable to walk without considerable human assistance and are dependent in other self-care tasks, yet have adequate memory, attention, and home support to be able to be discharged without the need for skilled nursing placement [12]. In the U.S., Medicare requires that patients can tolerate at least three hours of therapist-directed treatment a day, usually begun within 5-10 days after onset of stroke. Internationally, the time from stroke onset to rehab admission is 1-6 weeks and the duration of inpatient care is 3-8 weeks, but longer in Japan where a more comprehensive post stroke care system is available [13].

The goals of inpatient therapy can include increased independence for self-care activities (e.g., feeding, grooming, bowel and bladder care); the ability to perform safe toilet and wheelchair transfers; walking with or without assistive devices such as canes and orthoses that can brace the ankle and help control the knee; improved receptive and expressive language skills; and better executive, visual-perceptual, working memory, and other cognitive skills. In the outpatient setting, patients work with therapists to refine and build upon these skills to increase their functional independence in the home and community [14•].

During rehabilitation, physical, occupational, and speech therapists enable the practice of tasks of importance to patients, set and update realistic goals within the limitations of residual reflexive and voluntary neural control, and instill a regimen of daily skills practice of progressive intensity and difficulty. Therapists may utilize neuromuscular facilitation techniques to begin to guide the re-acquisition of motor skills, before building from simple to more complex actions that comprise goal-directed behaviors [15].

Principles Underlying Rehabilitative Therapies

Two basic principles influence approaches to patient treatment. The first is that the adult central nervous system is adaptive, or plastic, and has some capacity to re-organize itself to recover degraded cognitive and motor functions. Animal studies are identifying genetic and biochemical pathways involved in the establishment of new anatomic connections and functional network reorganization (e.g., axonal sprouting, dendrite proliferation, neurogenesis) [16•]. In patients, changing patterns of brain activation appreciated by MRI and other non-invasive imaging techniques reflect regional plasticity of the neuronal ensembles that represent actions and thoughts. Such changes are time-dependent and associated with learning and practice, as well as behavioral compensation for the loss of pre-stroke neural control. Thus, the brain of stroke patients, like healthy persons, constantly undergoes anatomic and physiologic changes induced by motor learning.

The second principle is that progressive, skilled motor practice is essential for continued gains at any time after stroke onset. Training must engage the attention, motivation, and learning networks of the brain to be effective. Better gains also depend on greater sparing of the neural networks that represent the components of a behavior. Although observational studies suggest that maximal functional gains are made by 3 months after onset, these studies do not account for other changes that can occur with regular practice, such as improved walking speed and distance or greater coordination in the use of an affected hand [17]. Large, randomized controlled trials in neurologic rehabilitation have reported long-lasting functional improvements after 2-12 weeks of skilled motor practice in patients who were weeks to years past onset of hemiparesis [18, 19••, 20••]. Thus, starting at the time of initial rehabilitation, physicians ought to instill in their patients a regimen of daily repetitive skills practice that can be carried over into the outpatient setting and into daily activities.

Interventions for Mobility

Fitness and Muscle Strength

Clinicians should emphasize ways for persons with stroke to augment their general conditioning and muscle strength in both the affected and unaffected limbs. Pre-morbid deconditioning due to sedentary behavior exacerbates the fall-off in activity resulting from new neurologic disability [21]. Indeed, patients disabled by stroke take half as many steps, use their affected arm much less, and have longer daily sedentary periods compared to healthy age-matched persons [22•, 23, 24]. It becomes very difficult for the hemiparetic person to achieve an aerobic effect from exercise, due to a combination of central weakness, inactivity, and muscle atrophy [25]. This is of concern because secondary stroke prevention recommendations include at least a half-hour of daily exercise rigorous enough to have at least a mild aerobic effect [26]. Just as important, higher levels of physical activity are associated with greater neurogenesis, better performance on cognitive tasks, less age-related hippocampal atrophy, and a reduced risk for vascular dementia.[27•, 28].

Standard rehabilitative therapies include selective muscle strengthening by isometric and isokinetic exercises to improve the power and endurance of affected and unaffected muscle groups. Sets of moderate resistance exercise with weights or elastic bands are feasible for most patients. Simply standing up and sitting down 5-10 times during commercials on television can improve proximal leg strength. Aerobic exercise training, whether by treadmill, over ground walking, or recumbent cycling, can produce a conditioning effect and increase walking speed and endurance [29]. The most impressive results for aerobic exercise training have been reported in chronic stroke patients who have recovered sufficient motor control to participate in moderate-to-vigorous physical activity [30••]. Questions remain about how best to provide and reinforce aerobic exercise, such as through a support group [31], and how to maintain compliance with exercise [32]. Physicians can encourage more frequent daily walks over longer distances and at faster speeds in addition to more formal exercise.

Over-ground Walking and Balance Training

Over-ground gait training is an integral component of standard physical therapies to improve dynamic balance and ensure safe ambulation in the home. Patients first practice trunk and head control, sit-to-stand balance, and then stepping in the controlled environment of the parallel bars. Over-ground training emphasizes clearance of the paretic foot to initiate leg swing, knee stability in stance, and stepping with a more rhythmic, safe gait pattern, using an assistive device or orthotic as needed. A Cochrane review found positive correlations between the amount of over-ground training and small improvements in gait speed with no significant increase in the number of adverse events such as falls [33]. Falls are a common outcome for patients recovering from stroke, with an incidence of over 40 percent for more than one fall in the first year [34]. The addition of a series of balance and truncal exercises, either as a supplement to inpatient therapies [35] or as part of an outpatient tele-rehabilitation intervention, [36] may prove to be a cost-effective means by which to prevent further disability.

Body Weight-supported Treadmill Training

Body weight-supported treadmill training (BWSTT) enables supervised, repetitive, task-related practice of walking. Patients with limited motor control wear a chest harness connected to an overhead lift to reduce the need to fully load a paretic leg. The treadmill induces rhythmic stepping, although the paretic leg and trunk often require physical assistance by therapists. The expectation, based on animal studies, was that BWSTT would increase the amount of practice while enabling more normalized sensory inputs to better drive motor output for stepping. The Locomotor Experience Applied Post Stroke (LEAPS) trial, however, failed to identify an additional clinical benefit of BWSTT as compared to a home exercise program of a similar intensity and duration [20••]. Although initially a highly regarded potential intervention for poor walkers, BWSTT may not reflect the task-related environment of over-ground training for motor learning [37•]. The cost in equipment and personnel with the expertise to deliver BWSTT make it an intervention to be tried only for patients who have at least modest motor control, but are not making progress with intensive over-ground training.

Robotic Gait Assist Devices

Electromechanical-assistive devices, including robotic steppers and exoskeletons, provide patients with either full or partial guidance of the lower limbs during the phases of the gait cycle [38]. As compared to BWSTT, for example, these devices can provide automated gait training on a treadmill or elliptical-like device and require no hands-on supervision by therapists. To date, the devices have generally not led to greater overall gains in over-ground walking parameters than the same intensity of more conventional physical therapy [39]. Robotic devices are being introduced that may better enable motor learning by letting patients make kinematic errors during practice. Very recently, wearable, lightweight, motorized exoskeletons have become available that assist with hip or knee flexion and weight bearing while stepping over ground. Although rather expensive, they may enable slow ambulation when otherwise not feasible; controlled studies will be needed to determine if their use can augment standard rehabilitation practice.

Functional Electrical Stimulation

FES is a technique that takes advantage of peripheral nerves and muscles left unaffected by damage to the central nervous system. Electrical stimulation is applied to trigger contraction and relaxation of select muscle groups. In the case of walking, excitation of the common peroneal nerve by an externally placed stimulator results in dorsiflexion at the ankle to aid paretic foot clearance. Small, randomized studies of external [40] and implanted [41] electrodes have reported improvements in gait lasting at least six months after the intervention. Though several commercial devices are available in the United States, efforts have only recently begun to demonstrate their potential cost effectiveness [42].

Interventions for the Upper Extremity

Constraint-induced Movement Therapy and Bimanual Practice

Therapy for the hemiparetic arm might begin with single-joint attempts at movement before proceeding gradually to more complex, multi-joint actions, then task-specific practice such as reaching to grasp a coffee cup, a process known as shaping. Facilitation of skilled motor practice for the upper extremity can take several forms, including shaping plus constraint-induced movement therapy (CIMT). This technique includes 6 hours a day of progressive task-related practice with restraint of the unaffected limb all day for 2 weeks. Increased use and faster skilled movements of the affected limb may result and persist for up to two years [43•]. However, the intervention has shown efficacy only in patients who can partially extend the wrist and fingers, meaning they have fair motor control and at least modest corticospinal tract sparing. Extensive restraint may not be as critical to gains as the high intensity of practice with a therapist; gains have been seen with just 2 hours of daily practice and without restraining the unaffected hand all day [44•, 45]. When the hand is chronically very weak, commercially available forearm-hand orthotic devices with embedded FES electrodes can enable a hand grasp or finger pinch to assist functional use.

Bimanual practice with simultaneous arm movements aims to activate the bilateral motor cortices and enhance input to the affected upper extremity, thereby leading to increased functional use of the paretic arm and hand. In small trials, bimanual practice has resulted in a similar degree of functional recovery as CIMT [46•].

Mechanical Devices to Assist Arm Movements

Mechanical devices range from spring-loaded orthotics to assist a specific movement, such as wrist extension, to fully automated, robotic limb prostheses for patient-triggered assistance of shoulder, elbow and wrist movements. Patients practice a series of specific joint movements by guiding an object on a computer screen through a maze. As with the electromechanical-assistive devices designed for gait training, robotic arm devices may enable more practice with more normalized limb kinematics. Used as a supplement to standard care, such devices may provide a benefit, [47, 48] but a similar degree of function can usually be achieved using standard therapies at the same intensity [19••]. Most are too expensive for home use. These devices may prove more efficacious in combination with other rehabilitative therapies such as non-invasive brain stimulation, [49] but further research is needed.

Pharmacologic Therapies to Limit Spasticity

It is often not medically necessary to treat increased muscle tone, unless spasms or flexor postures of the upper extremity cause pain, skin breakdown or interfere with hygiene. Baclofen and tizanidine are frequently used as first-line agents and dantolene’s effects on calcium action may also reduce hypertonicity. Botulinum toxin injected into selected muscle groups will reduce flexor or extensor postures around a joint for about 3 months, but usually does not improve functional use of a highly paretic hand [50, 51]. Shoulder pain is common after hemiplegic stroke, associated with subluxation and joint stresses [52]. Rapid management of pain with light exercise, range of motion, and anti-inflammatory medications can help prevent pain-induced spasticity in the arm and hand. Inversion and plantar flexion of the foot can also be lessened by medications and botulinum toxin to try to improve stepping. When a muscle is partially paralyzed by the toxin, daily stretching and ranging of the affected joint are necessary to maintain the improvement.

Interventions for Aphasia

Melodic Intonation and Constraint-induced Therapies

A range of individual speech and language therapy techniques have been developed to address the wide variety of aphasic syndromes that occur after stroke [53]. Most patients need a multi-modal approach to build on their strengths and to limit frustration in word finding and fluency. Melodic intonation therapy was developed for patients who have poor expression but good comprehension. This technique uses simple melodies and rhythmic tapping to engage networks that subserve prosody of language [54]. In a nod to the massed-practice paradigm of CIMT, constraint-induced aphasia therapy was developed as a means to improve verbal output [55]. Where comprehension is poor and output is perseverative, therapies have little effect. Regardless of the treatment modality employed, regular home-based practice with the family is imperative for the development of social communication.

Digital Technologies

Advances in digital communication technology have led to treatments for aphasia that can be personalized and delivered in the home setting [56]. For example, a recent study of speech entrainment that delivered an audiovisual intervention on an iPod screen reported a significant increase in verbal output for chronic stroke patients with Broca’s aphasia [57]. Several helpful computer programs for home practice are also available. Treatment of speech and cognitive disorders is likely to be a growing application of smartphone, tele-rehab, and other Internet-enabled practice and cueing paradigms.

Interventions for Visual Field Deficits and Inattention

Visual field loss and visual hemi-inattention degrade long-term functional outcomes in patients with stroke [58]. While it is unlikely that rehabilitation will result in recovery from a hemianopia, computer-based compensatory therapy may assist in directing visual search and attention into the area of loss [59]. The direct dopamine agonist rotigotine modestly lessened hemispatial neglect in sub-acute stroke patients [60•]. For spatial hemi-neglect, a prism in eyeglasses will shift the center of vision toward the abnormal field to improve reading and some self-care tasks [61].

Interventions for Locked-In Syndrome

Brain-machine interfaces utilize direct communication between the nervous system and devices outside of the body to enable communication or the performance of goal-directed movements. The devices are most needed for people with locked-in syndrome from brainstem stroke who are without voluntary control of their limbs. Alterations in the amplitude of the mu rhythm by thoughts about an action, are recorded with electroencephalography electrodes and interpreted by a computer algorithm, allowing patients to select letters or words on a computer screen for communication or to search the Web [62]. More advanced systems can record directly from implanted microelectrodes over a variety of cortical regions. An interface then controls directional movements of a prosthetic limb [63•]. While some of these technologies are coming into routine use, many challenges about cost and reliability remain.

Non-invasive Brain Stimulation

In addition to being employed to study brain physiology and neuroplasticity [64], techniques including TMS and transcranial direct current stimulation (tDCS) have been used to modulate cerebral plasticity in combination with physical training. Most trials focus on the recovery of arm function, [65] though the techniques are being exploited to increase verbal output in aphasia [66], improve swallowing [67], and increase walking speed [68] to give but a few examples. Generally modest gains in aspects of motor control have been reported when TMS is combined with other rehabilitative therapies [49, 69•, 70]. Similar equivocal results have been reported for tDCS protocols [71•]. A lack of consensus persists regarding appropriate patient selection, stimulation protocol, location, and duration [72]. The Food and Drug Administration has not approved their use outside of research, except for some types of depression. These techniques seem to work best in patients who have some residual voluntary movement.

Modulation of sensorimotor cortex excitability can also be achieved through the stimulation of peripheral nerves, either in isolation [73] or in conjunction with cortical stimulation [74] [75]. Definitive evidence that peripheral nervous system activation leads to improved functional outcomes is not yet available [76].

Mirror and Virtual Reality Therapies

The connections between parietal cortex and pre-motor and primary motor regions can be modulated by action observation and mirror therapy [77]. These techniques involve patients watching the movements of healthy individuals or, via a mirror, the unaffected limb. The subject attempts to mimic the observed movements. In contrast to other rehabilitative techniques such as CIMT, action observation and mirror therapy can be performed on patients with more severe limb paresis [78]. Clinical benefit has been reported in meta-analysis of small trials, but the magnitude of benefit depends upon the comparator therapy provided [79].

Virtual reality (VR) therapies use technology to combine action observation with repetitive skills practice. The hope is that this strategy will be especially engaging and reinforce practice paradigms. As simple as a commercially available video game that can be played at home or as complex as a system that measures joint angles in the arm and provides visual corrective feedback, VR has generated much excitement in the rehabilitation community as a means to promote and monitor skills practice [80]. Individual trials have reported benefits [81], but given the diversity of interventions and outcomes used, efficacy for a particular type or degree of impairment has not yet been demonstrated [82•].

Pharmacologic Interventions

Attempts to augment stroke recovery by modulating the neurotransmitter pathways of the central nervous system can also involve medications. Amphetamine showed promise in highly selected patients for motor gains, but no adequately powered trial has been completed after twenty years of small studies [83]. Efforts to boost cerebral dopaminergic action through the use of ropinirole proved ineffective in patients with chronic stroke [84]. The NMDA receptor antagonist memantine was reported to improve spontaneous speech and naming skills in chronic stroke patients with aphasia [85]. While individuals seem to occasionally improve in response to neurotransmitter-related drugs, no specific recommendations can be made.

The FLAME trial [86] tested fluoxetine in combination with standard rehabilitative therapies and reported better Fugl-Meyer motor scores, which tests voluntary movements against gravity, for those patients who received the drug. This work needs to be replicated [87]. The drug may also provide benefits as an anti-depressant, as depression affects at least 30 % of patients within one year after stroke.

Cell-based and Biologic Therapies

Embryonic and mesenchymal stem cells, cultivated precursors of neurons and oligodendrocytes, and other autologous and commercialized cells are actively undergoing investigation as potential treatments for stroke. Cells could replace lost neurons or glial cells, remyelinate damaged axons, or produce substances such as growth factors that could help drive network function and plasticity [88]. Several reported trials have not reported clinical efficacy, but others are being planned [89, 90]. To be of value, cellular and biologic interventions will have to be combined with applicable rehabilitation strategies to optimize their incorporation and action in neural networks.

Off-shore stem cell clinics are all too easy to find on the Internet. These high-priced cellular interventions can have a powerful placebo effect for patients with neurologic disease. Organizations that study stem cell research policies recommend that no patient should participate in or pay to receive a cellular intervention outside of a registered trial with a formal safety monitoring committee, in order to enable scientifically valuable information to be derived from the trial [91, 92].

Miscellaneous Approaches

Acupuncture is frequently offered in Asian countries for stroke rehabilitation. While individual patients may report a benefit, controlled trials have generally found little or no added value for improving specific impairments and disabilities [93, 94]. A recent trial reported that hyperbaric oxygen therapy may improve functional outcomes after stroke [95]. The study design was less than optimal, however. The cost of this treatment modality is high, there are risks accompanying its use, and the science underlying its utilization in chronic stroke is difficult to support. Etanercept, approved by the FDA for use in psoriatic and rheumatoid arthritis, has been proffered as a treatment for chronic stroke in various clinics. The manufacturer, Amgen, specifically points to a lack of evidence for its use in stroke and the few published reports by one dermatologist are highly biased and lack proper scientific theory, design, and interpretation of results [96].

Conclusions

Most survivors of a stroke are left with chronic disability. Rehabilitation efforts during the initial three to six months after stroke should aim to maximize patients’ physical, communicative, and cognitive functioning. Continued improvement in the chronic phase of stroke can occur with regular, progressive skills practice of goal-directed tasks in the home [12]. Many new rehabilitation strategies, built upon attempts to leverage technological developments to augment the effects of practice, are opening innovative avenues to amplify gains in performance at any time after stroke. The future of stroke rehabilitation remains one of promise and challenge in treating residual disabilities, especially for testing biological interventions for neural repair in the most profoundly affected individuals.