Rehabilitation of arm function after stroke. Literature review




Abstract


Introduction


In the recent literature we can find many articles dealing with upper extremity rehabilitation in stroke patients. New techniques, still under evaluation, are becoming the practical applications for the concept of post-stroke brain plasticity.


Methods


This literature review focuses on controlled randomized studies, reviews and meta-analyses published in the English language from 2004 to 2008. The research was conducted in MEDLINE with the following keywords: “upper limb”, “stroke”, “rehabilitation”.


Results


We reviewed 66 studies. The main therapeutic strategies are: activation of the ipsilesional motor cortex, inhibition of the contralesional motor cortex and modulation of the sensory afferents. Keeping a cortical representation of the upper limb distal extremity could prevent the learned non-use phenomenon. The modulation of sensory afferents is then proposed: distal cutaneous electrostimulation, anesthesia of the healthy limb, mirror therapy, virtual reality. Intensifying the rehabilitation care means increasing the total hours of rehabilitation dedicated to the paretic limb (proprioceptive stimulation and repetitive movements). This specific rehabilitation is facilitated by robot-aided therapy in the active-assisted mode, neuromuscular electrostimulation and bilateral task training. Intensifying the rehabilitation training program significantly improves the arm function outcome when performed during subacute stroke rehabilitation (< six months). Ipsilesional neurostimulation as well as mental practice optimize the effect of repetitive gestures for slight motor impairments. Contralesional neurostimulation or anesthesia of the healthy hand both improve the paretic hand’s dexterity via a decrease of the transcallosal inhibition. This pathophysiological mechanism could also explain the positive impact of constraint-induced movement therapy (CI therapy) in an environmental setting for chronic stroke patients.


Conclusion


To ensure a positive functional outcome, stroke rehabilitation programs are based on task-oriented repetitive training. This literature review shows that exercising the hemiparetic hand and wrist is essential in all stages of a stroke rehabilitation program. New data stemming from neurosciences suggest that ipsilesional corticospinal excitability should be a priority.


Résumé


Introduction


La rééducation du membre supérieur hémiplégique après accident vasculaire cérébral fait l’objet d’une large littérature actuelle. Les nouvelles techniques en cours d’évaluation sont l’application pratique du concept de plasticité cérébrale post-lésionnelle.


Méthodes


Cette revue de la littérature porte sur les essais contrôlés randomisés, revues et méta-analyses publiés en langue anglaise de 2004 à 2008. La recherche a été effectuée dans MEDLINE avec les mots-clés suivants : « upper limb », « stroke », « rehabilitation » .


Résultats


Soixante-six études ont été revues. Les principes thérapeutiques essentiels sont : l’activation du cortex moteur ipsilésionnel, l’inhibition du cortex moteur contralésionnel et la modulation des afférences sensorielles. Conserver une représentation corticale de l’extrémité distale du membre supérieur préviendrait le syndrome de non-utilisation acquise. La modulation des afférences sensorielles est proposée dans ce sens : électrostimulation cutanée distale, anesthésie de la partie proximale du membre parétique, thérapie par miroir, réalité virtuelle. L’intensification de la prise en charge consiste à augmenter le volume horaire total de rééducation dédiée au membre supérieur parétique (stimulations proprioceptives et répétition du mouvement). Elle est facilitée par l’usage du robot en mode actif aidé, l’usage de l’électrostimulation neuromusculaire et par la réalisation de tâches bimanuelles. L’intensification améliore significativement le pronostic fonctionnel de préhension lorsqu’elle est proposée en phase précoce (< six mois). La neurostimulation ipsilésionnelle ou encore l’imagerie mentale ont été évaluées en cas de déficit moteur léger : elles potentialisent l’effet de la répétition du geste. La neurostimulation contralésionnelle ou l’anesthésie de la main saine améliorent la dextérité de la main parétique via une diminution de l’inhibition transcalleuse. Ce mécanisme physiopathologique pourrait aussi expliquer l’effet régulièrement positif de la thérapie de contrainte en situation écologique chez l’hémiparétique chronique.


Conclusion


La répétition intensive du geste dans un objectif fonctionnel est la base de la rééducation du membre supérieur hémiplégique. Cette revue montre que la mobilisation du poignet et de la main parétique est indispensable à toutes les étapes du programme de rééducation. Les nouvelles données issues des neurosciences suggèrent en effet de prioriser l’excitabilité corticospinale ipsilésionnelle.



English version



Introduction


In the literature we can find many recent articles dealing with rehabilitation of upper-limb impairment in stroke patients. New techniques, still under evaluation, are the practical application for the concept of post-lesion brain plasticity .



Methods


For this literature review focusing on rehabilitation management of upper-extremity (UE) motor impairment following stroke, we analyzed the publications from these past 5 years (MEDLINE 2004 to 2008; keywords “stroke”, “upper limb”, “rehabilitation”). We only selected randomized controlled studies (RCS), literature reviews and meta-analysis published in the English language for which the main judgment criteria included the evaluation of motor impairment and/or arm function capacity. Thus amongst 103 studies initially identified, 66 were selected (56 RCS, three meta-analysis, seven literature reviews). The results are presented separately for subacute stroke (< 6 months) and chronic stroke (> 6 months). We differentiated the severely impaired subjects from the moderately impaired ones according to the scales available for each study (Brunstrom motor stages classification, Fugl-Meyer score – upper limb, Action Research Arm Test score, etc) or if these were not available we based our differentiation on the presence or not of voluntary distal motricity (active extension of the wrist and fingers greater or less than 10°).


In the first part we underline the theoretical framework. In the second part, we describe the rehabilitation techniques that are under experimental evaluation (sensory stimulation, activation of the ipsilesional motor cortex, inhibition of the healthy hemisphere). The third part focuses on rehabilitation techniques evaluated under clinical conditions (hours spent on classic rehabilitation training, neuromuscular electrostimulation, CI therapy, virtual reality, robot-aided therapy).


In conclusion, we propose a therapeutic strategy based on the stage and severity of the disease.



New data stemming from neurosciences



Learned non-use phenomenon


The paralysis induced by the brain lesions leads to a neuromuscular dysfunction spontaneously self-fostered and self-aggravated. This is the learned non-use phenomenon (or functional motor amnesia of the hemiplegic patient according to Meige ). This clinical phenomenon is directly linked to the post-lesion cortical somatotopic reorganization within the primary somatomotor areas of the lesioned hemisphere. This post-lesion plasticity sets in quite quickly a few hours after stroke .



Concept of maladaptative plasticity


The rehabilitation strategies of functional compensation by the healthy limb can perpetuate the learned non-use phenomenon . Inhibiting the healthy hemisphere becomes a therapeutic goal (constraint or anesthesia of the healthy upper limb, low-frequency transcranial magnetic stimulation of the healthy motor cortex).



Concept limits


The healthy hemisphere does not play a univocal role after a stroke. The spontaneous UE neurological recovery after a stroke can occur through contralesional motor cortex recruitment . This bihemispheric reorganization allows the healthy hemisphere to efficiently contribute to a unimanual motor task by the paretic arm.


The motor facilitation observed in the bilateral mode could be explained by the fact that the healthy hemisphere removes its inhibition upon the lesioned hemisphere . Furthermore, the corticospinal fibers stemming from the healthy hemisphere, nondecussated and aimed for proximal motricity, could be recruited .



The rehabilitation model proposed by the neurosciences



Criticism of the neurodevelopmental approach


The neurodevelopmental approach according to the Bobath theory is widely accepted without any validated evidence of its superiority . The Bobath theory, aiming to restore the postural control as a prerequisite to repetitive task training in order to be as physiological as possible, is opposed to theories inspired by Carr and Shepherd on motor skill training where the training program should focus on performing a functional task, regardless of the motor strategies used . A physiological training (i.e. with various functional tasks directly linked to daily life activities) promotes the acquired gesture ability ( Table 1 ) . The sensory environment must be “enriched” with multimodal stimuli (e.g. proprioceptive, visual, etc.) .



Table 1

The conceptual evolution of stroke rehabilitation promotes repetitive movements for performing functional tasks in an environment with different stimuli. New rehabilitation techniques make this training possible even in case of severe paralysis (robot-aided therapy, mental practice…).












































Conceptual framework Sensorimotor Functional Environmental
Neurodevelopmental therapy Impaired-oriented training Task-oriented repetitive training
Rehabilitation techniques Bobath Brunstromm Arm BASIS training Robot-aided therapy (active, resistive) Neuromuscular electrostimulation integrated in a functional strategy Robot-aided therapy (active aided) Constraint-induced movement therapy (CI therapy) Virtual reality Mental practice
Therapeutic strategies Normalize the voluntary movement Reduce motor impairment Reduce the arm function impairment for trained tasks
Facilitation and muscle inhibition Muscle reinforcement Task-oriented repetitive training (mass practice, shaping) a
Achieve a retention and generalization of the training
Diverse tasks presented in random order

a The word “shaping” means a positive conditioning obtained when progressively increasing the difficulty of the trained task. Situations of failure are thus avoided, the physiotherapist offers positive feedback to the patient for each successful completed training step. “Mass practice” means the task is repeated in an intensive manner for each new difficulty level.




Study of the ipsilesional corticospinal excitability


Clinically, the best prognosis factor for evaluating the probability of regaining hand dexterity is the UE residual voluntary motor ability at 1-month post stroke . The localization of the brain lesion, its type and size are also recognized prognosis factors to predict the final outcome of the patient’s functional recovery . It is possible to evaluate the corticospinal excitability in the first week poststroke. Within this timeframe, the presence of motor evoked potential (MEP) on the hand’s intrinsic muscles prefigures a better outcome than the lack thereof. However, the negative predictive value for lack of MEP is weak, the onset of a cortical evoked response can take several months and the link between distal hand functions recovery and upper limb distal MEP is complex .


The study of fractional anisotropy (FA) of the internal capsule by diffusion tensor imaging can detect and quantify corticospinal tract degeneration . In the chronic stroke patient, the decrease of the FA ratio (lesioned/healthy hemisphere) is correlated to the degree of distal voluntary motricity . Stinear et al. showed that the potential response to chronic stroke motor rehabilitation training can be predicted by evaluating the corticospinal tract excitability and the FA value . The lingering of a motor evoked response on the hand permits noteworthy functional gains up to 3-years poststroke. When faced with the absence of a motor evoked response, the lack of more than 75% of FA predicts very limited benefits gained with this training.


These extraclinical indicators are not part of our daily medical practice. The studies listed in this article underline their relevance as tools for establishing therapeutic strategies (restoring or compensating an impaired motor capacity) based on the knowledge of the corticospinal tract excitability.



Evaluation under experimental conditions



Sensory stimulations


During motor training, the subject’s sensory environment can be modulated by decreasing or increasing the kinesthetic, exteroceptive, visual and attention information. The “increased” feedback is probably quite useful for arm function rehabilitation .



Attentional information


The feedback can be verbal while the subject performs a motor task. In case of moderate motor impairment, to give some information to the patient on the quality of his interjoint coordination would be more efficient than regular feedback on his or her task performance . This result is coherent to those obtained by trunk restraint for compensating movements in case of severe motor impairment .


Partial deafferentation-deefferentation of the upper limb can be done by a selective regional anesthesia of the upper nerve roots of the brachial plexus. This technique, coupled with distal motor retraining, could lead to significant functional improvements for the hemiparetic hand (study on seven chronic cases) . The idea would be to reorganize the sensory and motor maps in favor of the distal extremity under-represented after a stroke. In this study, the exploration by MEP seems to corroborate this hypothesis. We can see the relevance of this technique in subacute stroke when there is a minimal distal motricity rendered inoperative by proximal syncinesia pattern.



Physiotherapy


Two hours of transcutaneous sensitive neurostimulation are sufficient to improve the paretic hand function of chronic stroke patients. The somatosensory stimulation of the median, ulnar and radial nerves is delivered at the level of the paretic wrist . Given prior to the rehabilitation training session, this stimulation would boost the motor training impact, probably by triggering an ipsilesional intracortical disinhibition .


Using thermal stimulation (TS), where patients are encouraged to take their paretic arm away when they feel an uncomfortable sensation, could promote the recovery of proximal motricity in early stroke rehabilitation .


The impact of acupuncture on upper limb motor recovery is not conclusive . On the other hand, electroacupuncture (EA), technique closer to the transcutaneous sensitive neurostimulation, increases the functional improvements of classic early stroke rehabilitation .



Mirror therapy


Mirror therapy consists in creating the illusion of perfect bilateral synchronization. Initially used for treating phantom limb pains of amputees, it has recently been the object of some publications for poststroke cases and recently a RCS . Patients are instructed to perform bimanual flexion-extension movements of the wrists and fingers (30 min/day, for a duration of 4 weeks). Their paretic limb is not visible: either hidden behind a simple panel (control group) or hidden by a mirror that reflects the movements of the healthy limb (experimental group). Furthermore both groups are given a neurodevelopmental treatment. Mirror therapy gives better motor capacity and autonomy scores for tasks involving the upper limb and the acquired results last for 6 months. This study included 40 stroke patients, all had a lesion on their dominant hemisphere, a characteristic that would enhance a positive response to this bimanual therapy . Results interpretation is quite limited due to two facts: non consistent poststroke delay (3 months to 1 year) and lack of control over the time spent in classic rehabilitation training (from 2 to 5 hours per day).



Activation of the ipsilesional motor cortex



Constraint-induced movement therapy (CI therapy)


The clinical experiments on hemiparetic monkeys that had undergone deafferentation underlined the reversibility of the learned non-use phenomenon acquired by the mandatory use of the paretic limb . Thus, CI therapy as described by Taub et al. is the most complete application of the functional task paradigm. The rehabilitation training for the paretic arm is quite intensive (60 hours, 6 hours/day, over a 10-day period). The exercises are applied according to the “intensive mass practice approach”, i.e. breaking down an arm function task into simple tasks performed separately and repeated several times, as the participants improve in performance, the complexity and difficulty of the tasks were increased in an attempt to continue to challenge them. The subjects wore a restraining device on their healthy limb during the day in order to force them to use their paretic limb to perform their daily tasks. The Liepert et al. study reports the use of cortical mapping by transcranial magnetic stimulation and shows a lingering modification of the cortical activity after CI therapy . Other observations suggest a correlation between the cortical activity changes and the CI therapy response. These series of cases unveil a great inter-individual variability regarding the activation patterns triggered by task training . This horizontal plasticity, mainly the widening of the paretic hand representation beyond the primary somatomotor areas is non-specific since it is observed after bimanual training .



Mental imagery


Mental imagery can be defined as the conscious representation of an action and is based on a subliminal activation of the motor neuron system. The latter is not only involved in performing a movement but also in imagining actions, recognizing various tools, comprehending an other person’s behavior and observational learning . Mental imagery training is used by athletes to acquire new motor skills. We should differentiate the mental imagery exercises performed at the first and third person. In the first case the subject is the actor performing an imaginary movement (“kinesthetic imagery”), in the second case he or she is the spectator of a mental representation of his or her own body in movement (“visual imagery”).


For stroke patients, many studies argue in favor of the first technique (“kinesthetic imagery”) but the recent review by Braun et al. counterbalances these results: RCS are rare, the protocols applied vary from one study to the next even though they usually consist in subjects mentally repeating a daily task they have previously seen performed by a third party . The positive impact of mental imagery training can be translated into increased arm function capacities (Action Research Arm Test) but not in terms of motor impairment (pinch force) . This positive impact is restricted to the tasks practiced both mentally and physically, in that case mental imagery should only be considered as a secondary technique . A recent RCS versus placebo suggests that mental imagery, performed after physical rehabilitation exercises, considerably improves motor impairment and arm function capacities . The patients included in this study are chronic stroke patients with moderate motor impairment and free from severe cognitive disorders. However, a more specific cognitive assessment would be useful to differentiate the subjects capable of correctly performing mental imagery exercises from those who are unable to perform them correctly (“chaotic motor imagery”) or even not able to perform it at all . Such tools are available for assessing this aptitude and stick to treatment compliance (e.g.: Motor Imagery Questionnaire). Finally the efficacy in subacute stroke and/or in case of severe motor impairment has rarely been studied even though it is under these conditions that mental imagery could compensate for the lack of therapeutic options.



Bilateral movement training


The aptitude to coordinate both upper limbs during a bilateral in-phase (BIP) task is partially retained in the stroke patient . Bilateral movement training validated by the meta-analysis of Stewart et al., does not appear clearly superior or even as efficient as the unilateral mode in some other studies . Many factors could explain these divergent results: poststroke delay, degree of motor impairment, type of bilateral movement training proposed (proximal or distal, functional or sensorimotor) and amount of movement repetitions.



High-frequency transcranial magnetic stimulation (TMS)


Repetitive transcranial magnetic stimulation (rTMS) modulates motor cortex excitability. Its inhibiting or facilitating effect is directly linked to the chosen stimulation frequency as demonstrated by the “paired-pulse stimulation” technique described by Kujirai et al. . In paired-pulse cortical stimulation experiments, conditioning subthreshold stimuli suppress the electromyographic (EMG) responses of relaxed muscles to suprathreshold magnetic test stimuli at short interstimulus intervals (ISIs) (1–5 ms) due to the effect of GABAergic inhibition within the motor cortex. Long ISIs (6–15 ms) facilitate the first stimulation (glutamatergic interneurons). Furthermore, during repeated stimulation, the stimulation frequency modulates the cortical excitability: a frequency lower than 1 Hz reinforces the intracortical inhibition, conversely, a frequency above 5 Hz facilitates the cortical excitability. The effects of rTMS can transitory linger after stimulation. This lingering effect is based on the induction of a phenomenon of long-term depression (LTD) and synaptic long-term potentiation (LTP) . In neuropsychiatry rTMS is the focus of several clinical studies: in drug-resistant depression where it could offer a real alternative to electroconvulsive therapy, in treating tinnitus, abnormal movements and more recently poststroke patients . In the latter case, clinical studies versus placebo are possible with the control group receiving magnetic stimulation below 10% of the resting motor threshold. The patient perceives the noise and vibrations triggered by the magnetic stimulation and feels the weak electrical current induced at scalp level.



Single ipsilesional rTMS session


This crossover study rTMS-placebo included 15 chronic stroke patients with slight motor impairment. This single ipsilesional rTMS session (10 Hz, 80% of the motor threshold) included eight trains of impulses, and each train was immediately followed by the repetition of a complex motor task with the paretic fingers. After rTMS, the precision and execution speed of the motor task are immediately improved, this result is correlated to the improvement of the ipsilesional cortical excitability (amplitude of the MEP). No adverse events were reported but no follow-up was conducted for this study. This study suggests that rTMS facilitates motor retraining after a stroke. However the choice of a complex motor tasks leads to a confusion factor: the improvement could be linked to attentional parameters.



Transcranial electrical stimulation


Transcranial electrical stimulation of the ipsilesional motor cortex is proposed to improve arm functions of the chronic stroke patient. Hummel et al. directed a double-blind, sham-controlled, crossover study to test the hypothesis that non-invasive stimulation of the motor cortex could improve motor function in the paretic hand of patients with chronic stroke. This was possible because the current delivered to the placebo group was of such low intensity (1 mA) that the subject could not differentiate, after a few seconds, if the electrostimulation stopped or lingered on . The six subjects tested were at least 2 years poststroke and only one case had a lesion of the cerebral cortex. In all cases, the primary motor cortex was spared by the lesion. The UE motor impairment is very slight (96% of the Fugl-Meyer score). The stimulation is applied during 20 minutes to the primary motor area of the paretic hand. Hand function was measured using the Jebsen-Taylor Hand Function Test (JTT), JTT measured in the paretic hand improved significantly with non-invasive transcranial direct current stimulation (tDCS) only. In parallel, the ipsilesional corticospinal excitability increased as well. The clinical effect lingered on for 25 minutes after the session but disappeared after 10 days. These functional results are comparable to those obtained by rTMS.


The electrical stimulation of the primary motor cortex using implanted epidural cortical electrodes is used to enhance voluntary arm function capacity. Brown et al. obtained very encouraging results, without any severe adverse events on six chronic stroke patients who had had an ischemic stroke at least 4 months prior to inclusion causing persistent to moderate weakness of the arm. The electrical stimulation was only delivered during the rehabilitation sessions, either at an intensity level corresponding to half of that able to trigger an electro-induced movement, or if there is no movement at 6,5 mA. In the control group, four chronic stroke patients receive similar rehabilitation training, without electrical stimulation. The superiority of the experimental treatment is demonstrated on the motor impairment (upper extremity Fugl-Meyer score), however the control group has a significantly higher poststroke delay constituting a valid bias . The 12-week follow-up should be extended.


These motor results are also found in a RCS with similar methodology. This study included 24 patients (moderate motor impairment, median 33 months poststroke) and compares motor retraining + corticostimulation versus motor retraining only over a 6-week period . Significant improvements to the upper extremity are described after a 4-week follow-up.


The question is: which stimulation, electrical or magnetic, transcranial or epidural, presents the best benefit/risk ratio? This question remains unanswered to this day.



Coupled stimulation


Other perspectives might be of interest: for example, facilitating the central motor command by cortical stimulation could be improved by coupling it to a peripheral nerve stimulation , or to distal electrostimulation (EMG-stim), a technique association not studied to this day.



Inhibition of the healthy hemisphere



Low-frequency transcranial magnetic stimulation



Single rTMS session on the healthy motor cortex


One single stimulation session (1 Hz, 90% of the motor threshold) performed 7 days after the ischemic stroke immediately improves hand dexterity (Nine Hole Peg Test) but not the palmar-pinch grip force. This double-blind RCS included 12 stroke patients. No adverse events were reported but no follow-up was available .


Takeuchi et al. conducted a double-blind RCS on 20 chronic stroke patients . One group received rTMS stimulation (1 Hz, at 90% of the motor threshold) on the healthy hemisphere, and the other one a placebo stimulation. The randomization was preceded by a training phase in order to determine the threshold and exclude any training impact on subsequent results. The assessment was based on the pinch task between the paretic thumb and index. Movement speed was improved but not the pinch grip force. This effect did not last more than 30 minutes after the stimulation. In parallel, they observed a clear decrease of the transcallosal inhibition. The study highlights the positive impact of rTMS on paretic hand functions, but the clinical relevance remains quite limited. It mainly shows that the efficacy of healthy cortex stimulation is due to the diminution of the transcallosal inhibition.



Repetitive sessions of rTMS on the healthy hemisphere over a 5-day period


The dual goal of this study was first to show that rTMS had a higher and more lingering effect with repetitive sessions over a 5-day period and second to validate the safety of this method. The study included 15 patients at least 1-year poststroke. They were randomly divided into two groups: rTMS (five sessions on the primary motor cortex, at 100% of the motor threshold and a frequency of 1 Hz) and placebo stimulation. The randomization was preceded by a training phase. The patients were evaluated beforehand, during and 2 weeks after the end of the treatment. We can see a decrease in the motor cortical excitability in the healthy hemisphere and an increase on the lesioned side. The functional evaluation is based on the JTT, the Purdue Pegboard Test, with simple reaction time and multiple choices. All these criteria are significantly better in the rTMS group, even day 15th of the follow-up. A correlation appears between the hand function (JTT) and changes in the cortical excitability of the lesioned hemisphere. The safety of the method is validated by a cognitive evaluation and an electroencephalogram (EEG). The authors only report one episode of headaches for one subject in each group and one subject suffering from anxiety. Thus, we can observe better results with repetitive rTMS sessions but mostly the positive effects last longer. Nevertheless, we can wonder if the numerous various tests administered in this study might not increase the α risk, and if the results presented are not fortuitous. Furthermore most patients suffered from a left subcortical stroke, thus limiting the external validity of the study.



High and low-frequency rTMS: discussion


The reviewed RCS are summed up in Table 2 .



Table 2

Transcranial magnetic stimulation (rTMS), RCS review.





















































































RCS Kim et al. 2006 a Takeuchi et al. 2005 Fregni et al. 2006
Number of subjects 15 20 15
Inclusion of patients with cortical lesions Yes No Yes
Poststroke delay > 3 months 6 months 1 year
Motor impairment Slight Slight Moderate to Slight
Exclusion criteria Tight stenosis of the internal carotid artery, implant, epilepsy, lesion of the primary motor cortex Cognitive disorders History of drug and/or alcohol abuse, neuropsychiatric disorders
Stimulation type Lesioned hemisphere Healthy hemisphere Healthy hemisphere
10 Hz 1 Hz 1 Hz
80% RMT 90% RMT 100% RMT
1 session 1 session 1 session /day; 5 days
Clinical evaluation criteria Precision and speed for performing a digital motor task Speeding up of the thumb-index pinch movement Jebsen-Taylor Hand Function Test, Purdue Pegboard Test, reaction time
Neurophysiological evaluation criteria Cortical excitability (amplitude of the motor evoked potentials) Duration of the transcallosal inhibition Cortical excitability (RMT b )
Follow-up None 30 minutes 2 weeks
Adverse side events None None 1 slight headache and 1 increased anxiety

a Crossover study: rTMS-placebo.


b RMT: relaxed motor threshold.



Stimulating the lesioned hemisphere could be beneficial by unveiling the corticospinal connections that are present but functionally silent around the lesion. The efficacy of the stimulation of the healthy hemisphere could lead to the same phenomenon by decreasing the transcallosal inhibition. In that case, we can discuss the potential inhibiting effect applied to the nondecussated corticospinal fibers, partly responsible for proximal motor capacity. Their inhibition by low-frequency rTMS could be harmful and this aspect is not at all taken into account by the two studies listed (the authors only evaluated distal motor capacity). Furthermore, we can bring up some confusion risk factors: mood improvement could have an indirect impact and thus lead to a greater motivation for stroke rehabilitation training.


rTMS can not be used in all patients, the main contraindications being pregnancy and epilepsy. However, seizures were only reported for less than 10 patients out of more than tens of thousands of patients that benefited from a stimulation . The adverse events reported in this review are benign (headaches). The therapeutic effect of a single session only lingers on for a few minutes and thus does not present a real practical benefit. Repetitive consecutive stimulation seems to lead to longer lingering effects. This effect seems to be dose-dependant (number of treatment days and number of trains of impulses by session). The studied populations are not really representative of the typical ischemic stroke populations. In fact these patients only had slight UE motor impairments.


In conclusion, we need to be cautious. The stimulation response is heterogeneous and it is necessary to conduct more studies on therapeutic outcomes but also on potential harmful effects of daily stimulations over long periods of time.



Constraint-induced (CI) therapy: physical constraint of the healthy upper limb


It would be relevant to study both the effects of intensive rehabilitation training and CI therapy in a RCS. We can discuss the efficacy of physical constraint in regards to the following results: combined use of CI therapy with physical constraint of the healthy upper limb (2 hours/day) and classic rehabilitation training does not yield any additional functional improvement . We can also observe the same results when physical constraint is proposed alone for chronic stroke rehabilitation when the patient is back home . Finally, keeping the training protocol while eliminating the physical constraint of the healthy upper limb yields significant functional improvements .



Anesthesia of the healthy upper limb


Transitory anesthesia of the healthy hand can be achieved with a venous compression-induced ischemia to the wrist. It leads to motor performance improvement of the paretic fingers in 13 chronic stroke patients. This suggests a decrease of the transcallosal inhibition applied by the healthy hemisphere .



Mixed techniques under clinical conditions



Classic rehabilitation training: impact of intense rehabilitation training on the stroke patient’s care


In the literature, classic rehabilitation training corresponds to non-standardized physiotherapy and occupational therapy rehabilitation care. It is based (in variable proportions according to different authors) on various known rehabilitation techniques (Bobath, proximal or distal functional electrical stimulation (FES), bilateral exercises, compensatory activities with healthy upper limb, etc…). On average this basic classic training has a total duration time of 10 hours: 30 minutes per day, 5 days a week and the average stay in a rehabilitation center is around 4 weeks for english speaking countries.



Early stroke rehabilitation


Increasing the total amount of classic rehabilitation care by 5 hours does not yield any functional improvements, even though the treatment starts on the 10th day poststroke . Five studies suggest that 15 to 20 additional hours of specific rehabilitation retraining, taking place during the first trimester poststroke, can lead to clinically significant improvements of the hand function on the long term. The degree of motor impairment would condition the access to a technique that would either be specifically focused on sensorimotor or functional improvements . The “Arm BASIS training” is a standardized sensorimotor training dedicated to severely impaired patients. In a multicenter RCS including 62 stroke patients (40 days poststroke), three treatments are compared. The “Arm BASIS training”, the “Bobath” therapy and the “classic rehabilitation training” . The first two groups benefited from 15 hours of experimental training on top of the classic rehabilitation training program common to the three groups. The intensification itself (+ 15 hours Arm BASIS or + 15 hours Bobath) did not yield any additional functional improvement, these improvements were even better in the control group (classic rehabilitation training only). However, the Arm BASIS group reached the best motor scores after treatment. This study underlines the limit of an intensive sensorimotor training in severely impaired stroke patients. This technique, alone, cannot improve hand functions and manual dexterity.



Chronic stroke rehabilitation


One year after stroke, undergoing 9 hours of functional retraining does not yield any clinical results (in case of moderate motor impairment) . Conversely, 57 hours of classic rehabilitation training, administered outside a specialized rehabilitation center setting, are efficient for moderate motor impairment .



Synthesis


Such contrasted results illustrate two essential points of the early stroke rehabilitation care. First, the results indicate a global lack of efficacy in case of severe motor impairment. In fact, the sensorimotor training only improve hand functions and the functional rehabilitation training leads to disappointing results in regards to the high rehabilitation costs: 50 hours of rehabilitation training . Second, the results reveal impact of the treatment duration in case of moderate motor impairment. We can estimate that 25 rehabilitation hours are needed in case of moderate motor impairment. The development of new motor retraining techniques focuses on two objectives. The first one is to increase the number of gesture repetitions during each session. The second one is to render this gesture repetition possible in case of severe motor impairment. The final goal is to shorten the hospitalization time. In chronic stroke rehabilitation, the aim is to elaborate rehabilitation programs that can be performed at home with a minimum of human intervention (physiotherapists, occupational therapists). The techniques proposed are neuromuscular stimulation, robot-aided therapy and virtual reality.



Neuromuscular electrostimulation


Conventional electrostimulation, or Functional ElectroStimulation (FES) targets the efferent nervous fibers in their intramuscular pathway. It is an external neurostimulation administered by surface electrodes placed on the muscle’s motor points. Distal electrostimulation (extensors of the wrist and fingers) is used in early and chronic stroke rehabilitation. It can be used for varying degree of motor impairments, in home setting, with good compliance. The choice of stimulation parameters remains empiric and still needs to be scientifically validated. In clinical studies we find quite similar data: biphasic current from 200 to 300 μs, frequency from 20 to 50 Hz, intensity from 30 to 45 mA in order to obtain a painless movement in the entire joint range of movement (ROM) .


Distal EMG-stim combines the detection, by surface electromyography, of weak voluntary muscle activity (starting at 50 μV) produced by the paretic muscles and the electrical stimulation of these same muscles when the threshold is crossed. This technique gives FES two bonuses: patient’s intention and effort. Distal EMG-stim, when used alone as a rehabilitation technique, improves active wrist and fingers extension as well as dexterity in chronic stroke patients with moderate motor impairments. This result is obtained when the stimulation is applied once per minute, 90 min/day, for a total of 4 days over a 2-week period . The functional improvements gained with distal EMG-stim seem superior to those gained with FES, but comparative studies alone are not enough to validate this conclusion . For a meta-analysis and a review see Bolton et al. and De Kroon et al. .


Cauraugh et al. showed that the combined use of EMG-stim and distal bilateral movements yielded noticeably better results (Box and Block Test) compared to distal EMG-stim for chronic stroke patients with moderate motor impairment. The bilateral movements of active wrist and finger extensions were performed simultaneously (in-phase) (6 hours) . For the group of patients following a training program including electro-aided bilateral movements, they observed a generalization of the distal motor improvements to the proximal musculature .


The combined use of distal EMG-stim and proximal stimulation (anterior deltoid and triceps brachii) could be more efficient on a functional level . Conversely, proximal FES (supraspinatus and posterior deltoid muscles) would be irrelevant. In fact, a RCS versus placebo conducted on 176 patients (less than 10 days after an ischemic stroke) did not report any arm function improvement after 3 months of treatment (FES during 1 hour, three times a day over a 4-week period) . These results should be interpreted with caution since the rehabilitation administered to both groups was not quantified. Yet they can also be interpreted according to the “learned disuse” concept. According to this concept, the reinforcement of the proximal muscles during acute stroke rehabilitation, while these muscles are the first ones to recover spontaneously, is detrimental to the cortical representation of distal muscles and thus would tend to limit hand’s movement recovery .


For distal electrostimulation, there does not seem to be a dose-dependant relationship, maybe due to the great inter-individual variability (skin impedance, denervation degree secondary to the central lesion) . Electrostimulation improves the voluntary motor motricity of the stimulated muscles but there is no proof of improved arm functions . The recent studies converge towards evaluating and validating distal electrostimulation as part of a functional therapeutic strategy.



Distal electrostimulation is an integral part of a functional strategy


Electrostimulation triggers a sensitive feedback which, when coupled with repetitive movements, induces a synaptic long term potentiation. And the enhanced cortical excitability facilitates the motor learning . Using electrostimulation to facilitate the opening of the hand during grasp and release exercises would optimize the functional improvements obtained with classical rehabilitation training (without electrostimulation) . These case studies results need to be validated by RCS.


Using A botulinic toxin with EMG-stim and CI therapy should be looked into some more, the forced inactivity of some spastic muscles complements the reinforcement of the useful impaired muscles . The early inhibition of syncinesia (elbow flexion and abduction-elevation of the shoulder) with a transient neuromuscular block remains an interesting therapeutic option that still needs to be looked into.



Constraint-Induced movement therapy (CI therapy)


The original technique (described above) has not been widely adopted by clinicians mainly because some noticeable changes are necessary to make it more practical for rehabilitation teams . First, it is clear that CI therapy should only be offered to highly motivated patients, free of severe cognitive disorders and not at risk for falling. These patients also need to have a minimal distal motor capacity (10° active extension of the long fingers and 20° active extension of the wrist). Some changes were suggested such as using a semi-robotized workstation freeing 75% of the physiotherapist’s time during a session or wearing a glove rather than a sling restraint in order not to impair the postural adaptations . The main change is to reduce the immobilization time to 5 hours/day, 5 days a week and the retraining exercises to 30 minutes/day, 3 days/week, this program takes place over a 10-week period .



Early stroke rehabilitation phase


Starting CI therapy before the 10th day poststroke would yield more significant functional improvements to the upper limb as opposed to classic rehabilitation training with the same amount of total training hours (15 hours) . However, the study was only conducted on 10 patients only. Moreover the classic rehabilitation training program included muscle reinforcement exercises, performing basic arm function tasks but also functional compensation with the healthy upper limb. Another study benchmarked CI therapy to classic rehabilitation training, both having the same intensity level and focusing on the paretic upper limb during the sessions (3 hours/day; over a 2-week period) . The subjects were included before the 15th day poststroke. There is no significant difference between the two methods; however they noted better results with CI therapy (Fugl-Meyer score) after the treatment and at the 3-month follow-up. The study’s limited power (23 subjects) could have masked a significant difference between the two groups . Another team (Myint et al.) found more solid results with 43 patients at less than 16 weeks poststroke. CI therapy is compared to the neurodevelopmental approach (same amount of training: 40 hrs). The resulting arm function capacities are better in the CI therapy group .


The multicenter RCS of Wolf et al. was conducted on 222 stroke patients recruited between the third and ninth month poststroke . The experimental treatment lasts 14 days and includes 6 hours/day of intensive functional training of the paretic arm while wearing a physical constraint on their healthy limb during 90% of the daytime. The control group does not perform any retraining tasks but some patients do benefit from physiotherapy. The palmar-pinch grip force and arm function capacities are measured with the Wolf Motor Function Test (WMFT). The performance in an environmental setting is measured with a retrospective self-administered questionnaire, the Motor Activity Log (MAL). The experimental treatment enhances the arm function capacities and the performances in situation, with a 2-year retention . However, for both groups no noticeable progresses were made on distal motor impairment.



Chronic stroke rehabilitation


A recent study versus placebo from Taub et al. underlines the efficacy of the technique for chronic stroke patients . Thus, the quality and frequency of use of the paretic limb in 30 daily life activities were quickly and noticeably improved after CI therapy (MAL). However, the arm functions assessment, using the WMFT, yielded only very modest results restricted to the execution speed in the proposed tasks.


The first RCS comparing CI therapy and neurodevelopment treatment included 66 patients. The treatment is not only recorded as superior for both motor impairments and arm functions but they also report that CI therapy improves the use of the paretic arm in an environmental setting .


In two RCS (CI therapy versus neurodevelopmental therapy) the kinematic movement analysis yields objective differences during a grasp/hold task. The results point towards CI therapy for better planning (reaction time) and better control of the movement in space and time (segmentation) but does not show any evolution of peak velocity (correlated to motor impairment) .


rTMS could be relevant in enhancing neurorehabilitation strategies. It was proposed in association to CI therapy in a RCS conducted on 19 chronic stroke patients. However this ipsilesional stimulation applied during 10 consecutive days (20 Hz, 90% of the motor threshold) did not yield any additional functional improvements . The frequency chosen by the authors was particularly high (no data were available on an eventual cortical localization of the lesions) without any reported adverse events and with a 6-month follow-up. During an ipsilesional stimulation at 20 Hz (110% of the motor threshold), other authors reported electromyography anomalies suggesting a non-negligible risk of seizures .



CI therapy: discussion


It is only when used in an environmental setting that CI therapy appears noticeably superior to other therapies. However the tool used (MAL) is based on a subjective appreciation (by the subject or a third party) and is retrospective. An ambulatory assessment tool is necessary to validate the MAL results .



Virtual reality


Virtual reality offers a major sensory feedback while the subjects are immersed in a virtual reality environment witnessing their own body in movement. Technological advances are expected to reduce the kinetosis linked to the time delay between the visual information received by the subjects and their movements performed in total immersion. The difficulty of the arm function exercises can be modulated according to the performance, the subject’s motivation is greater because of the playful aspect of the training . Training in a two-dimensional environment (1 hour/day, over a 4-week period) improved the arm function capacities of five chronic stoke patients compared to five control subjects who did not benefit from this virtual reality training . These exercises were associated to an augmented feedback on the performance and the result for each target-reaching try. The authors describe an ipsilesional focusing of the sensorimotor cortical activity in these five subjects after treatment.



Robot-aided therapy



Unimanual robot


The robotic assistive device seems to be the ideal sensorimotor support as it resolves the issue of human costs involved with rehabilitation. The robotic assistive device has the advantage of several modalities for facilitating the voluntary movement according to the motor command: passive, active-aided, active, counter-resistance adjusted for each session, uni- or bimanual work. The sensory feedback reinforcement is allowed by using an outside device (from an approximate target on a screen to total immersion in an interactive virtual environment) where subjects can visualize the path they describe. In the active-aided mode, the training is “errorless” since the robotic devices can complement the voluntary movement for each try.



Early stroke rehabilitation


The “NeReBot” allows for repetitive basic movements of the shoulder and elbow in three-dimensional space by eliminating the gravity effect. The forearm is strapped to a rigid horizontal pad itself suspended by cables. The patients actively move their arm towards various points, predetermined at the beginning of each session, according to their aptitudes . This rehabilitation training is offered as soon as day 7 poststroke (35 patients with severe motor impairment) . The experimental group undergoes 20 hours of robot-aided therapy on top of classic rehabilitation training. The very early intensification leads to a better proximal voluntary hand function (Fugl-Meyer) compared to the control group. The improvements yielded by this therapy are still recorded at the 8-month follow-up. The evolution of hand function capacities was not precisely studied.



Chronic stroke rehabilitation


Two successive studies have explored the relevance of resistance versus active-aided training on an “InMotion2” unimanual robot for subjects with moderate motor impairment . The first results show that working on resistance training improves the trained movements and this effect extends to wrist motor capacities. This spreading to distal motor capacities is not observed for the group working in the active-aided mode . In the second study and with a higher number of subjects there appears to be no difference between both groups . The motor improvements observed after 18 hours of therapy are not clinically significant and do not spread to distal motor capacities. This retraining specificity is in accordance with other studies on early and chronic stroke rehabilitation such as the one from Volpe et al. in 2000 .



Bimanual robot


The robot-aided therapy can underline the impact of intensive repetitive bilateral movements in the framework of a sensorimotor approach.



Early stroke rehabilitation


Distal bilateral movements . The robot-aided movements are extension-flexion and pronation-supination of the wrist. Their repetition improves the voluntary motor capacities of the arm in severely impaired patients. The experimental treatment consists of a daily 20-minute session, 5 days out of 7 over a 6-week period . The robot used (“Bi-Manu-Track”) can be adjusted for speed, ROM and resistance to movement according to the patient’s aptitude. It allows for a high number of repetitive tasks (40/min). The control treatment by distal EMG-stim does not include bilateral movements, it has the same amount of exercise hours but the intensity is lower (one electro-induced extension per minute). The two groups also receive an additional 7 hours of neurodevelopmental therapy. We should note that the Fugl-Meyer proximal upper limb motor score is also improved by both interventions, suggesting a non-specific extension to the uni- or bimanual mode.



Chronic stroke rehabilitation


Proximal bilateral movements . “Mirror image movement enabler” (MIME) robot allows for repetitive symmetrical (in-phase) bilateral movements in case of severe motor impairments with better results than neurodevelopmental therapy . The functional improvements are restricted to trained distal motor capacities and are clinically not significant (Fugl-Meyer score) in spite of 24 hours of training over a 2-month period.


The “bilateral arm training with rhythmic auditory cueing” (BATRAC) is a device similar to the MIME robot but also permits to alternate (symmetrical) in-phase and (non-symmetrical) anti-phase movements according to a rhythm guided by auditory feedback . The training is proposed 20 minutes per day, 3 days a week over a 6-week period for subjects with severe motor impairments. The 6 hours of robot-aided therapy do not bring any additional motor improvements compared to the same number of hours spent on neurodevelopmental therapy. The functional improvements on manual dexterous ability are limited to the execution speed of tasks that the patient had already mastered before treatment. To conclude, BATRAC-type training remains specific to trained motor capacities.



Synthesis


Tables 3a and 3b list all the RCS mentioned in this chapter.



Table 3a

Mixed techniques in clinical settings: RCS review. The experimental group benefits from more rehabilitation training hours than the control group.















































































































































Author, year, reference Mean poststroke delay (days) Upper limb motor impairment Treatment duration (weeks) Control group: rehabilitation training Experimental group
Rehabilitation training Additional hours of training Motor improvements Functional improvements Retained functional improvements
Rodgers et al., 2003 10 Moderate-severe 6 Classic rehabilitation training Classic rehabilitation training + 5 No No No (6 months)
Higgins et al., 2006 365 Moderate 6 Lower Limb Functional rehabilitation + 9 No No _
Feys et al., 1998, 2004 30 Moderate-severe 6 Classic rehabilitation training Sensorimotor + 15 Yes No Yes* (5 years)
Platz et al., 2005 40 Severe 4 Classic rehabilitation training ArmBASIS or NDT + 15 No No _
Winstein et al., 2004 15 Moderate-severe 5 Classic rehabilitation training Functional or sensorimotor + 20 Yes* No No (9 months)
Blennerhassett and Dite, 2004 40 Moderate 4 Lower limb Functional + 20 No Yes * Yes* (6 months)
Masiero et al., 2007 7 Severe 5 NDT Unimanual robot NeReBot + 20 Yes * _ _
Kwakkel et al., 1999 14 Severe 20 Lower limb Functional + 50 _ Yes *
Pang et al., 2006 365 Moderate-severe 19 Lower limb Classic rehabilitation training + 57 Yes* Yes * _
Wolf et al., 2006 180 Moderate 2 Placebo CI therapy + 60 No Yes * Yes* (2 years)
Taub et al., 2006 1460 Moderate 2 Placebo CI therapy + 60 _ Yes Yes (2 years)

NDT : Neurodevelopmental therapy; motor and functional improvement: yes = “statistically” more significant than the control group; yes* = the difference between both groups is equal or above the minimal “clinically” significant difference for each functional or motor test used (as a rule 10% of the total score); no = no statistical significance. Note : functional improvements in test conditions which excludes the Motor Activity Log (MAL).


Table 3b

Mixed techniques in clinical settings: review of the RCS. The total number of hours dedicated to rehabilitation training is similar in both groups.















































































































































Author, year, reference Mean poststroke delay (days) Upper limb motor impairment Treatment duration (weeks) Control group: rehabilitation training Experimental group
Rehabilitation training Total hours of training Motor improvements Functional improvements Retained functional improvements
Luft et al., 2004 900 Severe 6 NDT Bimanual robot BATRAC = 6 No Yes* _
Cauraugh and Sangbum, 2002 365 Moderate 2 Distal EMG-stim Distal EMG-stim + Distal bilateral movements = 6 _ Yes* _
Hesse et al., 2005 42 Severe 6 Distal EMG-stim Bimanual robot Bi Manu track = 10 Yes* _ _
Page et al., 2008 365 Moderate 10 NDT CI therapy = 15 No Yes* _
Page et al., 2005 10 Moderate 10 Classic rehabilitation training CI therapy = 15 Yes* Yes* _
Stein et al., 2004 365 Moderate 6 Unimanual Robot InMotion2 active-aided mode Unimanual Robot InMotion2 counter-resistance mode = 18 No No _
Lum et al., 2002 900 Severe 8 NDT Bimanual Robot MIME = 24 Yes _ _
Wu et al., 2007 365 Moderate 3 NDT CI therapy = 30 Yes _ _
Boake et al., 2007 15 Moderate 2 Classic rehabilitation training CI therapy = 30 No No _
Myint et al., 2008 112 Moderate 2 NDT CI therapy = 40 _ Yes* Yes* (3 months)
Van der Lee et al., 1999 1095 Moderate 2 NDT CI therapy = 60 No Yes Yes* (1 year)

NDT : Neurodevelopmental therapy; motor and functional improvement: yes = “statistically” more significant than the control group; yes* = the difference between both groups is equal or above the minimal “clinically” significant difference for each functional or motor test used (as a rule 10% of the total score); no = no statistical significance. Note : Functional improvements in test conditions which excludes the Motor Activity Log (MAL).


For more training hours ( Table 3a ), we observed that 25 hours of rehabilitation training yield more improvements than 10 hours of basic training in early stroke rehabilitation (paragraph 1). Should we conclude that proximal robot-aided therapy can reduce the physiotherapist time with the patient during these hours and consequently the hospitalization time? The results obtained with the NeReBot robot match these findings. However, what is the functional relevance of intensive reinforcement for the proximal muscles in early stroke rehabilitation (cortical representation of the upper limb)? During chronic stroke rehabilitation, the proximal robot-aided therapy does not yield any clinically relevant motor improvements for severely impaired subjects. And this is true whether the movements are performed in uni-, bimanual, active-aided or counter-resistance mode (InMotion2, MIME, BATRAC).


For a similar amount of training hours ( Table 3b ), clinically significant functional improvements are obtained when the distal motor capacity is specifically and intensively stimulated. A severe motor impairment should be geared towards distal bimanual training, electrostimulated or with a robotic assistive device. In case of moderate impairments, CI therapy can be considered superior to the neurodevelopmental techniques but not to classic rehabilitation training. In all cases, neurodevelopmental therapy remains inferior to experimental techniques.


On a methodological level, the Fugl-Meyer motor score (upper limb) is the evaluation criteria most often used. This score can be divided into proximal and distal sub-scores. These two parts are rarely separated in RCS when there is a distinct relationship between voluntary distal command and hand function .



Conclusion


In Table 4 , we propose a therapeutic strategy based on the severity of the motor impairment and the poststroke delay.



Table 4

Therapeutic strategy according to the stroke stage (acute or chronic) and severity of the motor impairment.


















Moderate motor impairment Severe motor impairment
Early stroke rehabilitation (< 6 months) Functional rehabilitation training (25 hours) including Distal EMG-stimulation + distal bimanual movements (6 hours) Bimanual distal robot (10 hours) or Distal EMG-stimulation + distal bilateral movements (20 hours) Then if possible: functional rehabilitation training (15 hours)
Chronic stroke rehabilitation (> 6 months) Constraint-Induced movement therapy (CI therapy) (30 hours) or Functional rehabilitation training (30 hours) (in a virtual environment setting or with verbal feedback on the performance) + Mental Imagery If the neurophysiological criteria are favorable a : classic rehabilitation training (50 hours) with trunk restraint including distal EMG-stimulation + distal bilateral movements (20 hours)

a See paragraph « Study of the corticospinal excitability ».


The neurodevelopmental approach refers to a conceptual framework rendered partly obsolete by the new data stemming from neurosciences. In fact, facilitation and muscle inhabitation techniques work alongside the spontaneous postlesion brain plasticity rather than controlling it, hoping for distal motor capacities to finally emerge. The present literature review shows that there are many ways to modify this postlesion plasticity in order to quickly improve, in priority, the UE distal motor capacities.


In conclusion we will underline four points that can serve as guidelines for the design of a post-stroke UL rehabilitation program:




  • the objective functional improvements generally occur after 25 hours of motor retraining in early stroke rehabilitation (functional rehabilitation in case of moderate motor impairment, classic rehabilitation training in case of severe motor impairment);



  • Techniques promoting distal motor capacities (EMG-stim, robot-aided therapy, bilateral movements) seem efficient regardless of the poststroke delay (in the absence of validated neurophysiological criteria);



  • an improved participation to daily tasks after CI therapy in case of chronic moderate impairment;



  • the emergence of central neuromodulation can complement motor training.






French version



Introduction


La rééducation du membre supérieur hémiplégique après accident vasculaire cérébral (AVC) fait l’objet d’une large littérature actuelle. Les nouvelles techniques en cours d’évaluation sont l’application pratique du concept de plasticité cérébrale post-lésionnelle .



Matériels et méthodes


Cette revue porte sur la rééducation du membre supérieur hémiplégique d’après les publications des cinq dernières années (MEDLINE 2004 à 2008 ; mots clés : « stroke », « upper limb », « rehabilitation »). Seuls les essais contrôlés randomisés (ECR), les revues et méta-analyses publiées en langue anglaise et dont le critère de jugement principal comportait l’évaluation de la déficience motrice et/ou de l’aptitude de préhension du membre supérieur ont été retenus. Ainsi parmi 103 études initialement répertoriées, 66 ont été sélectionnées (56 ECR, trois méta-analyses, sept revues). Les résultats sont présentés séparément pour les phases précoce (moins de six mois) et tardive (plus de six mois) après AVC. Nous avons distingué les patients sévèrement et modérément déficitaires d’après les échelles disponibles dans chaque étude (stades moteurs de Brunstromm, score Fugl-Meyer – membre supérieur, score Action Research Arm Test, etc.) ou à défaut sur la présence ou non d’une motricité distale volontaire (extension active du poignet et des doigts supérieure ou inférieure à 10°).


Dans une première partie, nous décrivons le cadre théorique. Dans une deuxième partie, nous exposons les techniques de rééducation qui font l’objet d’une évaluation expérimentale (stimulations sensorielles, activation du cortex moteur lésé, inhibition de l’hémisphère sain). La troisième partie porte sur les techniques de rééducation évaluées en condition clinique (volume horaire de rééducation classique, électrostimulation neuromusculaire, thérapie de contrainte, réalité virtuelle, thérapie robot-assistée).


Pour conclure, nous proposons une stratégie thérapeutique basée sur le stade et la sévérité de la maladie.



Les nouvelles données issues des neurosciences



Syndrome de non-utilisation acquise ( learned non-use )


La paralysie induite par la lésion cérébrale entraîne une désadaptation neuromusculaire spontanément autoentretenue et autoaggravée. C’est le syndrome de non utilisation acquise (ou amnésie motrice fonctionnelle de l’hémiplégique selon Meige ). Ce phénomène clinique est en lien avec la réorganisation somatotopique corticale postlésionnelle au sein des aires somatomotrices primaires de l’hémisphère lésé. Cette plasticité postlésionnelle est d’installation rapide puisqu’elle débute dans les premières heures post-AVC .



Concept de plasticité inadaptée ( maladaptative plasticity )


Les stratégies réadaptatives de compensation fonctionnelle par le membre supérieur sain peuvent pérenniser le syndrome de non utilisation acquise . L’inhibition de l’hémisphère sain devient donc un objectif thérapeutique (contention ou anesthésie du membre supérieur sain, stimulation magnétique à basse fréquence du cortex moteur sain…).



Limites du concept


L’hémisphère sain n’a pas un rôle univoque après AVC. La récupération neurologique spontanée au membre supérieur peut passer par le recrutement du cortex moteur contralésionnel . Cette réorganisation bihémisphérique donne un rôle favorable à l’hémisphère sain lors de l’exécution d’une tâche motrice unimanuelle côté parétique.


La facilitation motrice observée en mode bimanuel s’expliquerait par la levée de l’inhibition exercée par l’hémisphère sain sur l’hémisphère lésé . De plus, les fibres corticospinales issues de l’hémisphère sain, non décussées et destinées à la motricité proximale, pourraient être recrutées .



Le modèle rééducatif que proposent les neurosciences



Critique de l’approche neurodéveloppementale


La thérapie neurodéveloppementale selon la technique de Bobath fait consensus de manière empirique sans pour autant avoir fait la preuve de sa supériorité . Au concept bobathien qui vise à restaurer le contrôle postural comme prérequis à l’exercice d’une gestualité la plus physiologique possible, s’opposent les théories du réapprentissage moteur inspirées de Carr et Shepherd où l’entraînement se doit d’être directement orienté vers la réalisation d’une tâche fonctionnelle, quelles que soient les stratégies motrices utilisées . Un apprentissage de type écologique (c’est-à-dire avec des tâches fonctionnelles diversifiées et en lien avec les activités de la vie quotidienne) favorise la généralisation des acquis ( Tableau 1 ) . L’environnement sensoriel doit être « enrichi » de stimuli plurimodaux (proprioceptif, visuel…) .



Tableau 1

L’évolution conceptuelle de la rééducation de l’hémiplégie amène à privilégier la répétition de tâches fonctionnelles dans un environnement riche en stimuli. Les nouvelles techniques de rééducation permettent cette mise en situation, même en cas de paralysie sévère (thérapie robot-assistée, imagerie mentale…).







































Cadre conceptuel Sensorimoteur →Écologique →Fonctionnel
Thérapie neurodeveloppementale Thérapie centrée sur la déficience (impaired-oriented training) Thérapie centrée sur la répétition de tâches fonctionnelles ( task-oriented repetitive training )
Techniques de rééducation Bobath Brunstromm Arm BASIS training Thérapie robot-assistée (actif, contre résistance) Electrostimulation neuromusculaire intégrée dans une stratégie fonctionnelle Thérapie robot-assistée (actif aidé) Thérapie de contrainte Réalité virtuelle Imagerie mentale
Principes thérapeutiques Normaliser le mouvement volontaire Réduire la déficience motrice Réduire l’incapacité de préhension pour les tâches entraînées
Facilitation et inhibition musculaire Renforcement musculaire Répétition en blocs d’une tâche fonctionnelle (mass practice, shaping) a
Obtenir une rétention et une généralisation de l’apprentissage
Tâches diversifiées présentées en ordre aléatoire

a Le terme shaping (modelage) désigne ici un conditionnement positif obtenu en augmentant progressivement la difficulté de la tâche entraînée. Les situations d’échecs sont ainsi contournées, le thérapeute félicite le patient à chaque étape passée avec succès. Le mass practice désigne la répétition intensive de la tâche à chaque palier de difficulté.




Étude de l’excitabilité corticospinale ipsilésionnelle


Cliniquement, le meilleur facteur pronostique de récupération de la fonction de préhension est l’état de la motricité volontaire du membre supérieur un mois après l’AVC . La localisation de la lésion cérébrale, son type et sa taille sont également des facteurs pronostiques connus de la qualité finale de la récupération fonctionnelle . Il est possible d’évaluer l’excitabilité corticospinale dans la première semaine post-ictus. Dans ce délai, la présence de potentiel évoqué moteur (PEM) au niveau des muscles intrinsèques de la main est de meilleur pronostic que son absence. Cependant, la valeur prédictive négative de l’absence de PEM est faible, la réapparition d’une réponse évoquée corticale peut s’étaler sur plusieurs mois et le lien entre récupération fonctionnelle distale et PEM distaux au membre supérieur est complexe .


L’étude de la fraction d’anisotropie (FA) de la capsule interne en imagerie par résonance magnétique (IRM) par tenseur de diffusion fournit une quantification précise de l’intégrité anatomique du faisceaux corticospinal (FCS) . Chez l’hémiplégique chronique, la diminution du ratio FA (hémisphère lésé/sain) est corrélée au degré de motricité distale volontaire . Stinear et al. ont montré que la réponse à un réapprentissage moteur instauré en phase tardive post-AVC peut être prédite par l’évaluation de l’excitabilité du FCS et la mesure de la FA . La persistance d’une réponse évoquée motrice distale au membre supérieur autorise des gains fonctionnels notables jusque trois ans post-ictus. En l’absence de réponse évoquée motrice, la perte de plus de 75 % de la FA annonce un effet très limité du réapprentissage.


Ces indicateurs paracliniques ne font pas partie de la pratique médicale quotidienne. Les études citées soulignent leur intérêt en tant qu’outils d’élaboration de stratégies thérapeutiques (restauration ou compensation de la déficience motrice) basées sur la connaissance de la fonctionnalité de la voie corticospinale.



L’évaluation en condition expérimentale



Les stimulations sensorielles


Au cours du réapprentissage moteur, l’environnement sensoriel du sujet peut être modulé par la réduction ou l’amplification d’informations d’ordre kinesthésique, extéroceptif, visuel et attentionnel. Le rétrocontrôle ( feedback ) « augmenté », tous supports confondus, est très vraisemblablement utile à la rééducation de la préhension .



Informations attentionnelles


Le rétrocontrôle peut être donné verbalement par le thérapeute lors de la réalisation d’une tâche motrice. En cas de déficit moteur modéré, renseigner le patient sur la qualité de ses coordinations interarticulaires serait plus efficient que de lui renvoyer une information régulière sur la réussite de la tâche . Ce résultat est cohérent avec ceux obtenus par contrainte physique des mouvements de compensation du tronc lors de déficits moteurs sévères .


La déafférentation-déefférentation partielle du membre supérieur peut être réalisée par bloc anesthésique sélectif des racines sensitivomotrices supérieures du plexus brachial. Cette technique, couplée au réapprentissage moteur distal, permettrait d’obtenir des gains fonctionnels notables à la main (série de sept cas chroniques) . Le principe serait de réorganiser les cartes sensorimotrices à la faveur de l’extrémité distale sous-représentée après AVC. Dans l’étude, l’exploration par PEM semble corroborer cette hypothèse. On entrevoit l’intérêt de cette technique en phase précoce lorsqu’il existe une motricité distale minimale rendue inopérante par un schéma syncinétique proximal.



Physiothérapie


Deux heures de neurostimulation sensitive transcutanée suffisent à améliorer la dextérité manuelle de sujets hémiparétiques chroniques. La stimulation est délivrée sur le contingent sensitif des nerfs médian, ulnaire et radial au poignet parétique . Administrée avant une séance de rééducation, cette stimulation optimiserait l’effet du réapprentissage moteur, probablement par induction d’une désinhibition intracorticale ipsilésionnelle .


Le recours à une stimulation thermique sub-douloureuse avec consigne de retrait du membre supérieur lorsque l’inconfort apparaît pourrait favoriser la récupération de la motricité proximale en phase précoce .


Les effets de l’acupuncture sur la récupération motrice au membre supérieur ne sont guère probants . En revanche, l’électroacupuncture, technique proche de la neurostimulation sensitive transcutanée, accroît les gains de la rééducation classique en phase précoce post-ictus .



Thérapie par miroir


La thérapie par miroir consiste à créer l’illusion d’une synchronisation bimanuelle parfaite. Initialement appliquée aux douleurs de membres fantômes de l’amputé, elle a fait l’objet de quelques publications de cas post-AVC et récemment d’un ECR . Les patients réalisent des mouvements bimanuels de flexion-extension des poignets et des doigts (30 min/jour, quatre semaines). Leur membre parétique n’est pas visible : soit caché par un simple panneau (groupe témoin) soit caché par un miroir qui reflète les mouvements du membre sain (groupe expérimental). Les deux groupes reçoivent par ailleurs un traitement neurodéveloppemental. La thérapie par miroir autorise de meilleurs scores de motricité et d’autonomie pour les activités impliquant le membre supérieur, avec maintien des acquis à six mois. Cette étude inclut 40 hémiparétiques présentant tous une lésion de l’hémisphère dominant, une caractéristique qui favoriserait la réponse à cette thérapie bimanuelle . L’interprétation des résultats reste limitée par le délai post-ictus très variable (trois mois à un an) et par le manque de contrôle du temps de rééducation classique (deux à cinq heures par jour).



L’activation du cortex moteur lésé



Thérapie de contrainte : répétition du mouvement


Les expérimentations menées chez le singe hémidéafférenté ont illustré la réversibilité du phénomène de non usage appris lors de l’usage forcé du membre parétique . Ainsi, la thérapie motrice de contrainte décrite par Taub et al. constitue l’application la plus complète du paradigme de la tâche fonctionnelle. L’entraînement du membre parétique est intensif (60 heures, six heures par jour, dix jours consécutifs). Les exercices sont appliqués selon le concept de mass practice, c’est-à-dire par décomposition préalable d’une tâche de préhension en éléments simples travaillés séparément puis progressivement complexifiés et répétés un grand nombre de fois. La restriction des mouvements du membre supérieur sain par contention diurne force l’usage du membre parétique pour réaliser les gestes de la vie courante. L’étude de Liepert et al. (2000) recourt à la cartographie par stimulation magnétique transcrânienne et montre une modification durable de l’activité corticale après thérapie de contrainte . D’autres observations suggèrent une corrélation entre les modifications d’activité corticale et la réponse à la thérapie de contrainte. Ces séries de cas révèlent une grande variabilité interindividuelle quant aux patterns d’activation induits par l’entraînement . Cette plasticité horizontale, en particulier l’élargissement de la représentation de la main parétique au delà des aires somatomotrices primaires est non spécifique puisqu’elle est observée après réapprentissage bimanuel .



Imagerie mentale


L’imagerie mentale ( mental practice ) peut être définie comme la représentation consciente d’une action et se base sur une activation subliminale du système neuronal moteur. Ce dernier est impliqué non seulement dans la production du mouvement mais également dans l’imagination des actions, la reconnaissance d’outils, l’apprentissage par observation ou même la compréhension du comportement d’autrui . L’entraînement par imagerie mentale est pratiqué chez le sportif pour perfectionner ou acquérir de nouvelles habilités motrices. Il faut distinguer les exercices d’imagerie mentale réalisés à la première et à la troisième personne. Dans le premier cas le sujet est acteur de l’exécution imaginaire du mouvement ( kinesthesic imagery ), dans le deuxième il est spectateur d’une représentation mentale de son propre corps en mouvement ( visual imagery ).


Chez le sujet hémiplégique, plusieurs études argumentent en faveur de cette technique ( kinesthesic imagery ) mais la revue récente de Braun et al. pondère ces résultats : les ECR sont rares, les protocoles appliqués divergent d’une étude à l’autre bien qu’il s’agisse le plus souvent de répéter mentalement des gestes de la vie quotidienne que le sujet aura préalablement pu voir exécutés par un tiers . L’effet positif de l’imagerie mentale se traduirait en termes de capacités fonctionnelles de préhension (Action Research Arm Test) mais pas en termes de déficience motrice (force de la prise digitopalmaire) . Cet effet positif serait restreint aux tâches pratiquées mentalement et physiquement, ce qui ferait de l’imagerie mentale une technique d’appoint . Un ECR versus placebo récent suggère que l’imagerie mentale, pratiquée à la suite des séances de rééducation physique, améliore de façon significative la déficience motrice et l’aptitude de préhension . Les patients inclus sont en phase chronique et présentent un déficit moteur modéré ; ils sont indemnes de trouble cognitif sévère. Cependant, une évaluation cognitive plus spécifique serait souhaitable pour distinguer les sujets capables de réaliser correctement l’imagerie mentale et ceux qui la réalisent mal ( chaotic motor imagery ) ou pas du tout . Des outils existent pour mesurer cette aptitude et suivre la compliance au traitement (ex : Motor Imagery Questionnaire). Enfin, l’efficacité en phase précoce et/ou en cas de déficit moteur sévère est peu étudiée. Or c’est dans ces conditions que l’imagerie mentale pourrait palier au manque d’option thérapeutique.



Mouvements bimanuels


L’aptitude à coordonner les deux membres supérieurs lors d’une tâche bimanuelle en phase est partiellement conservée chez l’hémiplégique . Le réapprentissage bimanuel, bien que reconnu valable dans la méta-analyse de Stewart et al., n’apparaît pas clairement supérieur ou même aussi efficace que le mode unimanuel dans certaines études . Plusieurs facteurs pourraient expliquer ces résultats divergents : le délai post-ictus, le degré de déficience motrice, le type d’entraînement bimanuel proposé (proximal ou distal, fonctionnel ou sensorimoteur) et le nombre de répétition du mouvement.



Stimulation magnétique cérébrale à haute fréquence


La stimulation magnétique transcrânienne répétitive (rTMS) module l’excitabilité du cortex moteur. Son effet inhibiteur ou facilitateur dépend directement de la fréquence de stimulation choisie comme l’illustre la technique princeps de double stimulation ( paired-pulse stimulation ) décrite par Kujirai et al. . Il s’agit d’une première stimulation infraliminaire conditionnante suivie d’une stimulation test supraliminaire. Un délai bref (1–5 ms) entre les deux stimulations inhibe la réponse musculaire normalement évoquée par la seconde. Cette inhibition est le fait des interneurones intracorticaux inhibiteurs à transmission GABAergique. Un intervalle long (6–15 ms) rend la première stimulation facilitatrice (interneurones glutamatergiques). De même, lors de stimulations répétées, la fréquence de stimulation module l’excitabilité corticale : une fréquence inférieure à 1 Hz renforce l’inhibition intracorticale, à l’inverse une fréquence de plus de 5 Hz facilite l’excitabilité corticale. Les effets de la rTMS peuvent perdurer transitoirement après stimulation. Cette rémanence reposerait sur l’induction de phénomène de dépression et de potentialisation synaptique à long terme . La rTMS fait l’objet de nombreux essais thérapeutiques en neuropsychiatrie : dans la dépression pharmacorésistante où elle pourrait offrir une alternative à la sismothérapie, dans le traitement des acouphènes, des douleurs de désafférentation, de l’aphasie, des mouvements anormaux et plus récemment après AVC . Dans ce dernier cadre, les essais versus placebo sont possibles en appliquant, pour le groupe témoin, une stimulation magnétique inférieure à 10 % du seuil moteur de repos. Le patient perçoit le bruit et les vibrations produites par la stimulation magnétique et ressent l’effet du faible courant électrique induit au niveau du scalp.



Séance unique de rTMS ipsilésionnelle


Cette étude en cross-over rTMS-placebo porte sur 15 hémiparétiques chroniques avec déficit moteur léger. La séance unique de rTMS ipsilésionnelle (10 Hz, 80 % du seuil moteur) comporte huit trains de stimulation, chaque train étant immédiatement suivi de la répétition d’une tâche motrice complexe avec les doigts parétiques. Après rTMS, la précision et la vitesse d’exécution de la tâche motrice sont immédiatement améliorées, ce résultat est corrélé à l’amélioration de l’excitabilité corticale ipsilésionnelle (amplitude des potentiels évoqués moteurs). Aucun effet indésirable n’est rapporté mais il n’y a pas de suivi. Cette étude suggère que la rTMS facilite le réapprentissage moteur, cependant le choix d’une tâche motrice complexe induit un facteur de confusion : l’amélioration pourrait être liée à des paramètres attentionnels.



Stimulation électrique cérébrale


La stimulation électrique transcrânienne du cortex moteur ipsilésionnel est proposée en vue d’améliorer l’aptitude de préhension chez l’hémiparétique chronique. Hummel et al. ont ainsi mené un essai comparatif en double insu versus placebo rendu possible par le fait que le courant délivré est de si faible intensité dans le groupe placebo (1 mA) que le sujet ne peut pas différencier, après quelques secondes, l’arrêt ou la poursuite de l’électrothérapie . Pour les six sujets testés, l’AVC date d’au moins deux ans et intéresse le cortex cérébral dans un cas seulement. Dans tous les cas le cortex moteur primaire est épargné par la lésion. Le déficit moteur du membre supérieur est particulièrement léger (96 % du score Fugl-Meyer conservé). La stimulation est appliquée durant 20 minutes en regard de l’aire motrice primaire de la main parétique. Les patients progressent dans le temps de réalisation des épreuves de dextérité du Jebsen Hand Taylor Test après stimulation uniquement. L’excitabilité corticospinale ipsilésionnelle est parallèlement augmentée. L’effet clinique perdure 25 minutes après la séance mais disparaît au suivi à dix jours. Ces résultats fonctionnels sont comparables à ceux obtenus par rTMS.


La stimulation électrique du cortex moteur primaire délivrée par électrodes implantées au niveau épidural facilite la motricité volontaire du membre supérieur. Brown et al. ont obtenu des résultats très encourageants, sans effets indésirables graves, auprès de six hémiplégiques chroniques présentant un déficit moteur modéré. Le courant de stimulation est délivré uniquement au cours des séances de rééducation, soit à une intensité moitié moins que celle permettant de déclencher le mouvement électro-induit soit, à défaut de mouvement, à 6,5 mA. Dans le groupe témoin, quatre sujets hémiplégiques reçoivent une rééducation similaire, sans stimulation électrique. La supériorité du traitement expérimental est indiscutable sur la déficience motrice (score Fugl-Meyer-membre supérieur), néanmoins le groupe témoin présente un délai post-ictus significativement supérieur, ce qui constitue un biais non négligeable . Le suivi de 12 semaines demande à être prolongé.


Ces résultats moteurs sont reproduits dans un ECR de méthodologie similaire. L’essai porte sur 24 patients (déficit modéré, en moyenne 33 mois post-ictus) et compare réapprentissage + corticostimulation versus réapprentissage seul durant six semaines . Des gains fonctionnels significatifs sont décrits au membre supérieur à quatre semaines de suivi.


La question est : quelle stimulation, électrique ou magnétique, transcrânienne ou épidurale, présente le rapport bénéfice-risque le plus intéressant ? Cette question est aujourd’hui sans réponse.



Stimulation couplée


D’autres perspectives peuvent s’ouvrir : par exemple, la facilitation de la commande motrice centrale recherchée par stimulation corticale pourrait être optimisée en association à une stimulation nerveuse périphérique , voire ce qui n’a pas été étudié, à une électrostimulation distale (EMG-stim).



L’inhibition de l’hémisphère sain



Stimulation magnétique cérébrale à basse fréquence



Séance unique de rTMS sur le cortex moteur sain


Une seule séance de stimulation (1 Hz, 90 % du seuil moteur) réalisée sept jours après AVC permet d’obtenir une amélioration immédiate de la dextérité manuelle (temps de réalisation du Nine Hole Peg Test) mais pas de la force de la prise digitopalmaire. Cet ECR en double insu portait sur 12 sujets hémiparétiques. Aucun effet indésirable n’est rapporté mais il n’y a pas eu de suivi .


L’ECR en double insu de Takeuchi et al. porte sur 20 hémiparétiques chroniques . Un groupe reçoit une stimulation par rTMS (fréquence de 1 Hz, à 90 % du seuil moteur) sur l’hémisphère sain, et l’autre une stimulation placebo. La randomisation est précédée d’une phase d’entraînement afin d’obtenir un plafond et d’exclure tout effet d’apprentissage dans les résultats ultérieurs. L’évaluation porte sur une tâche de pincement entre le pouce et l’index parétiques. L’accélération du mouvement est améliorée mais pas la force de la pince. Cet effet ne dure pas plus de 30 minutes après la stimulation. On observe en parallèle une nette diminution de la durée de l’inhibition transcalleuse. L’étude met donc en évidence un effet de la rTMS sur la motricité de la main parétique, mais dont la pertinence clinique reste assez limitée. Elle permet surtout de démontrer que l’efficacité de la stimulation du cortex sain est due à la diminution de l’inhibition transcalleuse.



Séances répétées de rTMS sur l’hémisphère sain pendant cinq jours


Le but de l’étude est de mettre en évidence un effet plus important et surtout plus prolongé de la rTMS avec des séquences répétées sur cinq jours. Il s’agit également de vérifier l’innocuité de cette méthode. L’essai porte sur 15 patients à plus d’un an post-ictus. Ils sont répartis de façon aléatoire en deux groupes : rTMS (cinq sessions sur le cortex moteur primaire, à 100 % du seuil moteur et à une fréquence de 1 Hz) et stimulation placebo. La randomisation est précédée d’une phase d’entraînement. Les patients sont évalués avant, pendant, puis deux semaines après la fin du traitement. On observe une diminution d’excitabilité corticale de l’hémisphère sain et une augmentation du côté atteint. L’évaluation fonctionnelle porte sur le Jebsen-Taylor Hand Function Test (JTT), le Purdue Pegboard Test, le temps de réaction simple et à choix multiples. Tous ces critères sont significativement meilleurs dans le groupe rTMS, y compris au suivi à 15 jours. Une corrélation apparaît entre la fonction (JTT) et le changement d’excitabilité corticale dans l’hémisphère lésé. L’innocuité est objectivée par évaluation cognitive et électroencéphalographique. Les auteurs signalent seulement un épisode de céphalées chez un patient de chaque groupe et un état d’anxiété. On observe donc de meilleurs résultats avec des sessions répétées de rTMS, et surtout une durée beaucoup plus longue des effets. Néanmoins, on peut se demander si la multiplication des tests réalisés dans cette étude n’entraîne pas une inflation du risque α, et si les résultats présentés ne sont pas dus au hasard. De plus, les sujets étaient majoritairement atteints d’AVC sous-corticaux et gauches, ce qui limite la validité externe de l’étude.



rTMS haute et basse fréquence : discussion


Les ECR revus sont résumés dans le Tableau 2 .


Apr 23, 2017 | Posted by in PHYSICAL MEDICINE & REHABILITATION | Comments Off on Rehabilitation of arm function after stroke. Literature review

Full access? Get Clinical Tree

Get Clinical Tree app for offline access