Aug. 01, 2002
Valerie Pomeroy; Raymond Tallis
Physiotherapy Research International
Lewis Thomas (1995) in his justly famous book, described medicine as `the youngest science'. Compared with physics and chemistry and even the biological sciences, it was a relative new boy on the block. At the beginning of the twentieth century, physicians prescribed empirical nostrums, few of which had any biological basis and even fewer were evidence-based in the way we understand this term nowadays. Now, at the beginning of the twenty-first century, the notion of a scientific physic is no longer a contradiction in terms.
However, the penetration of science into clinical practice has not been uniform. There are still underdeveloped areas. Foremost among them is neurological rehabilitation, a scientific adolescent which, as is the way with adolescents, has precocious spending habits. The relative underdevelopment of the science base of neurological rehabilitation is to be deeply regretted because neurodisability is the most important healthcare challenge of the next few decades.
There are, for example, over 300 000 stroke survivors in the UK, coping with often very severe levels of disability (Ebrahim and Harwood, 1999) and stroke is but one of several common disabling neurological diseases. The incidence of most of these diseases rises steeply with age. As the population is ageing (Kinsella et al., 2000), unless we improve the effectiveness of neurological rehabilitation, we shall see more, not less, neurodisability in the future. This is despite advances in prevention and acute treatments, for it has been shown that at present the gains resulting from preventative measures are offset by the ageing of the population (Torvaldsen et al., 1999).
HOW EFFECTIVE IS NEUROLOGICAL REHABILITATION?
Rehabilitation is difficult to define but, very briefly, it denotes a package of measures designed to lessen the impact of disabling conditions (Young, 1996). Damage to the right hemisphere may lead to a leftsided weakness - an impairment. The question for rehabilitators is how to prevent this weakness, this 'impairment', from translating into a 'disability', such as an inability to walk, or, if this is not possible, how to prevent an inability to walk from translating into dependence on others and limited capacity to participate in everyday life, a 'handicap'.
We shall divide rehabilitation strategies into those that aim to prevent the translation of impairments into handicaps, mainly by helping the patient to adapt to them, which we shall name `higher-level' interventions, and those that aim to reduce impairment, which we shall name `lower-level' therapies. This distinction is clearly a gross simplification because some rehabilitation strategies may have both kinds of effects. The two strategies are illustrated in Figure 1.
The present state of affairs is illustrated by stroke. Organized stroke care saves lives and reduces disability. Recent metaanalyses (Langhorne et al., 1993; Stroke Unit Triallists' Collaboration, 2000) show that patients who are managed within an organized stroke service are 30% less likely to be dead or dependent at six months post-stroke. Even more encouraging, it has now been shown that benefits are not transient. They are seen five (Indredavik and Slordahl, 1997; Lincoln et al., 2000) and even 10 years after the stroke (Indredavik et al., 1999). So much is clear. When it comes to determining the specific contribution of rehabilitation, however, things are less obvious.
There appears to be a distinctive contribution of the overall rehabilitation package (as opposed to acute medical care and the prevention of medical complications) to the observed improved mortality and outcome in terms of dependency. The Stroke Unit Triallists Collaboration (2000) identified multidisciplinary team working, commitment of all parties to the principles of rehabilitation and early occupational therapy and physiotherapy as important factors distinguishing organized stroke care from care in other, less favourable, settings. An overview of studies of stroke rehabilitation found a weak correlation between increased intensity of rehabilitation and patients' improvement in terms of function in activities of daily living (Kwakkel et al., 1997). However, generalization on the basis of the reported studies was difficult because of their poor methodological quality.
The picture becomes even murkier when we try to determine the respective contribution of `higher-level' and `lower-level' components of the rehabilitation package to the overall effectiveness of rehabilitation. For example, Walker and colleagues (1999), in a study of stroke patients not admitted to hospital, found that modest input from an occupational therapist (approximately five visits over a six-month period) produced significant improvement in outcome in terms of activities of daily living. The intervention was `higher-level' and consisted of occupational therapy designed to increase independence in personal and instrumental activities of daily living. By contrast, when Lincoln and colleagues (1999) examined the effect of increased intensity of 'lowerlevel' upper limb rehabilitation, using interventions based on the Bobath concept (the one most widely used in the UK; Sackley and Lincoln, 1996; Davidson and Waters, 2000) they found that patients who received additional, more intensive therapy fared no better either in terms of activities of daily living or specific measures of arm function than those who received the standard treatment. The picture is not entirely clear, however, as post hoc analysis of the data suggests that additional therapy consisting of repetitive functional activities might improve function in the patients who demonstrated milder upper limb function (Parry et al., 1999). Lower-level interventions were also investigated by Kwakkel et al. (1999), who reported a beneficial effect of increased intensity of activity and muscle strengthening, which included functionally orientated therapy directed at upper and lower limbs. However, the implications of the study for the generality of stroke patients are uncertain because only 3% of patients were recruited and these workers did not measure impairment A recent systematic review of publications on physical therapies to improve movement performance and functional ability post-stroke (Pomeroy and Tallis, 2000) found that a few lower-level techniques seemed to be effective, at least in the studies which met the review criteria. Even though there was sufficient evidence to justify further investigation, there was no compelling evidence that any single technique had a major sustained effect in reducing impairments. Most of the primary studies were of poor quality.
None of the above should be taken to imply that therapy is not overall beneficial. However, the lack of major effects seen in research on clinical experience from lowerlevel techniques designed to reduce impairments is worrying because the gains to be had from higher-level interventions designed to help patients cope with impairments, although not insignificant, could be limited even if the best method of delivering the current package of care is identified (the overriding preoccupation of much health services research). If impairment can be reduced then the potential for recovery of function is greater. The challenge is to build on what we can presently achieve in reducing dependency through helping patients to adapt to impairments by trying to reverse impairments. For illustrative purposes the present paper focuses on physical therapies for motor problems.
NEUROLOGICAL REHABILITATION: NOT YET A MATURE SCIENCE
In clinical practice most physical therapies to restore movement have been developed through clinical experience and have a tendency to be based on now outdated beliefs about what does and what does not promote recovery. Often, these rationales lead to recommendations that do not withstand further examination. For example, the predominant physiotherapy concept in the UK, the Bobath concept, discourages self-propelled wheelchairs, on the basis that they increase spasticity, increase unwanted associated movements and increase abnormalities of posture, such as asymmetry (Cornell, 1991). This claim has been examined directly by use of an instrumented seat which records the distribution of the pressure of the buttocks on a wheelchair seat. It was not possible to show an increase in seated asymmetry as a result of wheelchair use. Other underlying rationales to practice, such as the proposed adverse effect of progressive resistive exercises to reduce muscle weakness are challenged by research findings (Miller and Light, 1997; Brown et al., 1997). A first step towards more effective treatment of impairments, therefore, will be a critical evaluation of current rationales for treatment.
Content of therapies
Of equal concern is the vagueness with which physical therapies tend to be specified, even in the most scrupulously conducted trials. This is understandable, perhaps, because of the complexity of such therapies. This complexity has been examined in some detail. It was found that not only did physiotherapists do many different things, for example 31 ploys in mobilizing patients with higher-level gait disorders (Mickelborough et al., 1997) and no less than 175 types of intervention in the prevention and treatment of post-stroke shoulder pain (Pomeroy et al., 2001). There was little consensus as to the 'correct' intervention to use.
Characterization of the target population
The deficits being treated are rarely clearly specified. Stroke patients, for example, tend to be grouped together for trials or, at best, stratified according to very broad categories of severity. For example, a recent comparison of Bobath approaches and Movement Relearning Programme approaches for stroke classified patients only according to sex and side of stroke (Langhammer and Stanghelle, 2000). If, as seems highly likely, therapy should be tailored to the type of deficits, it is essential to specify the nature of those deficits and to address relatively homogeneous groups of patients in trials, in order to avoid missing an effect on subgroups who may have benefited, even though the group as a whole did not. We do not, after all, evaluate drugs for otherwise unspecified 'fever'. Unfortunately, as yet there is no clear idea of a suitable taxomony of lesions or impairments for defining inclusion in clinical trials, something which became apparent in a recent trial of treadmill retraining for patients with higher-level gait disorders in cerebral multi-infarct states (Liston et al., 2000). A post hoc analysis of the results suggested that patients with gait ignition or initiation problem should be distinguished from those with equilibrium problems, with the focus on the former.
Although trials use measures of movement and functional ability which are clinically appropriate these give limited information about the mechanisms which underlie clinical change. This information is important for evaluating attempts to reverse impairments, as it gives insight into what, if anything, has actually happened to the patient's nervous system.
Some studies make an attempt to report and even quantify neurological changes, but even where it is potentially illuminating, this is often unreliable. For example, it has recently been found that the clinical assessment of muscle tone by use of a simple three-point scale: flaccid, normal and spastic, proved unreliable even in the hands of expert physiotherapists and doctors (Pomeroy et al., 2000). A novel visual analogue scale proved equally unreliable. Ironically, a paper published in the same issue of the same journal, came to the opposite conclusion about the reliability of clinical measures of muscle tone (Gregson et al., 2000), although the different outcomes have more to do with which statistical tests are appropriate to analyse agreement between raters (Pomeroy et al., 2001).
Understanding the effects of physical therapies on the nervous system requires utilization of neurophysiolological tests. In the neurological rehabilitation literature, only a small number of studies have done so. In the absence of appropriate tests, trials are comparable to studies of a putative hypotensive drug and assessing its effect in terms of quality of life without seeing whether it actually lowers the blood pressure. Put another way, it seems as if physical therapies have moved to the stage of pragmatic real world trials without passing through the first phase - essential for a truly scientifically based therapy of explanatory trials.
THERAPEUTIC IMPLICATIONS OF THE SOFT-WIRED SYSTEM
For a long time pessimism and the emphasis on adaptation to as opposed to reduction of impairment, seemed to be justified by the observation that neurones exhibit limited capacity for regrowth in the adult brain. If the brain were hard-wired, death of neurones meant irreversible loss of function. Now, things look more cheerful. In the `soft-wired' nervous system, function depends on organization (connectivity); reorganization does not always require regrowth, and so function may be restored without major regrowth. New methods of imaging the activity and connectivity of the living brain have been revelatory, showing a nervous system that is a less fixed structure than was previously thought.
There are many examples of neuroplasticity revealed by functional imaging, both of normal subjects acquiring new skills and of impaired individuals recovering from brain damage (for an excellent review, see Hallett et al., 1998). Hamdy and colleagues (1996), for example, used transcranial magnetic stimulation to map the cortical representation of swallowing in normal subjects, non-dysphagic stroke patients and dysphagic stroke patients. The cortical map of swallowing in a patient who was dysphagic at presentation and who by three months post-stroke had recovered swallowing showed dramatic changes. The initial dysphagia was explained by the near complete absence of cortical representation of swallowing, due to ischaemic damage. This area in the left hemisphere had been damaged by the stroke. Subsequent recovery from dysphagia was explained by the huge expansion of the cortical representation of swallowing in the opposite hemisphere.
We need to know how to assist this process. To do this a better understanding of the drivers to reorganization is needed, at both the microscopic (dendritic and synaptic) and macroscopic (cortical maps) levels. Here, another insight is crucial: functional activity-associated afferent input appears to be an important driver to reorganization that may bring about functional recovery. As expressed by Buonomano and Merzenich (1998) in relation to the cortex: experience drives reorganization. This is not entirely straightforward, as the size of cortical representation may be associated more with the process of skill acquisition than with the use of acquired skills (Pascual-Leone et al., 1994).
However, if afferent inputs associated with normal activity are the most potent influences maintaining or restoring functional organization, there is a problem for impaired patients, as illustrated in Figure 3. Loss of normal function deprives the brain of afferent information with the resulting attenuation of precisely those experience and activity-based signals that drive adaptive reorganization as well as permitting maladaptive reorganization, commonly expressed in spasticity. There is, therefore, a vicious circle which has to be broken.
There are several possible strategies, including the use of prosthetic inputs such as electrical stimulation (Glanz et al., 1996), assisted and independent activity giving more 'natural' inputs (Potempa et al., 1995; Dean and Shepherd, 1997; Visintin et al., 1998) and, an approach that has attracted much recent attention, constraint-induced therapy, which involve encouraging the use of a paretic arm contralateral to the stroke loss by constraining the use of the nonparetic arm (van der Lee et al., 1999).
Excitatory electrotherapy aims to bring about an immediate effect, such as the contraction of a muscle or the stimulation of a excitatory or inhibitory pathway in the nervous system. Cerebellar stimulation was introduced over 30 years ago in the hope of reducing spasticity in cerebral palsy on the basis that the output from the cerebellum was largely inhibitory (Cooper, 1973). Lateral popliteal nerve stimulation (Burridge et al., 1998) and forearm stimulation (Chae et al., 1998) have also been used. More recently, there has been considerable interest in deep brain stimulation for patients with Parkinson's disease (Benabid et al., 1991). Finally, programmed muscle stimulation, as in paraplegic walking devices (for example, Marsolais and Kobetic, 1988) has also been used in clinical practice, though its place still remains uncertain after 20 years of research and development. With the exception of thalamic stimulation for Parkinson's disease and functional electrical stimulation for selected patients with foot drop, the results have been largely disappointing and the overall message seems to be that excitatory stimulation is probably not going to play a major part in the management of patients with severe neurological disease. Patients do not like being encumbered with equipment, and find it inconvenient to put on and take off. There are also considerable technical problems. Moreover, knowing what to stimulate in the case of more complex lesions is not very clear. What is needed is the kind of stimulation that brings about functional reorganization: neuroculture rather than 'puppetry'. For this, chronic stimulation designed to bring about plastic change might be more promising, such as contingency stimulation for unilateral spatial neglect which has been shown to produce specific changes in measures of spatial neglect (Prada and Tallis, 1995).
If what maintains or restores functional organization of the nervous system is the information arising from normal activity, it might be expected that quasi-natural stimuli would be more effective than, uniform frequency electrical stimulation. Over a 10-year programme of work, this notion was explored using a rather simple and much less ambitious example of excitable tissue, namely muscle. Muscle is a good place to start because it exhibits extraordinary activity-related plasticity. By use of analytical electromyographic techniques, Kidd and Oldham (1988) tracked the firing patterns of individual motor units. They observed the changes of those patterns at the extremity of exertion, in other words, under precisely those conditions in which muscles become stronger and more fatigue-- resistant. The firing pattern of an extremely fatigued motor unit is a mixture of fast and slow frequencies. This pattern was loaded into a programmable stimulator and used to treat wasted muscles. The effect was measured in terms of both muscle strength and fatiguability. This technique was sometimes effective, as in patients with wasting and weakness of the first dorsal interosseus associated with rheumatoid arthritis (Kidd and Oldham, 1988) and, less predictably, with ulnar nerve palsy (Petterson et al., 1994), but was much less effective for large muscle, such as the quadriceps (Howe et al., 1995).
The conclusion to be drawn is that we are not yet able to mimic electrically the kind of information that excitable tissue needs in order to enhance its ability to undergo adaptive reorganization after injury.
Another way of inputting afferent information is, of course, the time-honoured approach of assisted activity. This lies at the heart of many physical therapy interventions and, as mentioned above, modest benefits are seen with some exercise-based therapies (Pomeroy and Tallis, 2000). Treadmill retraining, a form of assisted walking, has been widely investigated in recent years for patients with hemiparesis (for example, Visintin et al., 1998; Hesse et al., 1999) and with higher-level gait disorders due to cerebral multi-infarct states (Liston et al., 2000). Benefits are seen consistently.
The attractiveness of prosthetic inputs such as electrical stimulation is that they can be given in large quantities 24 hours a day but they do not closely match what the healthy system normally receives. The attractiveness of natural inputs from assisted activity is that they are `of the right kind'; however, they can be given in doses that could be regarded as homeopathic. What happens during 30 minutes' uninterrupted activity on, say, a treadmill is a minute proportion of the normal activity in a normal day. Upper limb rehabilitation therapy may last only a proportion of 30 minutes, and the number of movements falls well short of the number of movements made by a normal limb in a average day. There are, however, many added advantages to active participation as opposed to passive stimulation. They seem to carry more centrally active information because the patient attends more to what is happening and it has been clearly demonstrated that attention promotes reorganization (Robertson et al., 1997).
It seems as if we can roar at the nervous system in the wrong language or whisper to it inaudibly in the right language. Put another way, we seem to be able to give patients large doses of weak medicine through prosthetic inputs, or small or homeopathic doses of strong medicine through natural inputs. An exception to this may be constraint-induced therapy a form of exercise-based therapy in which the drive to increased activity of affected limb is continuous and sustained and the consequent increased afferent input is natural.
Constraint-induced therapy is based on experimental evidence of the effects of 'forced' use in monkeys and has progressed through a series of small studies in chronic stroke patients which have found beneficial and sustained effects (for example, Taub et al., 1993; Liepert et al., 1998). The technique involves constraining the arm on the stronger side to encourage or drive the use of the weaker side. However, a recent single-blind randomized clinical trial found that, although constraint-induced therapy improved upper limb function more than intensive bimanual training, an effect that was maintained one year after treatment, the difference between the two treatments was smaller than what was considered to be the minimal clinically important difference (van der Lee et al., 1999). Nevertheless, further characterization of stroke patients suitable for this approach might be indicated, as a post hoc analysis suggested that stroke patients with sensory disorder or unilateral neglect might be more likely to benefit (van der Lee et al., 1999).
In practice, constraint-induced therapy is an amalgamation of several elements: the constraint, repetitive functional exercise and increased attention to the weak limb. Indeed, one explanation for the lack of clinically significant effect found in the study reported by van der Lee and colleagues (1999) is that constraint-induced therapy was compared to bilateral exercise. The active ingredients might therefore be repetitive functional exercise and increased attention to the paretic limb rather than the constraint. This suggestion receives some support from a before and after study of 16 chronic stroke patients who showed significant increases in motor function of the upper limb after six weeks' bilateral arm training performed for one hour, three times a week (Whitnall et al., 2000). This effect was maintained two months after the end of treatment.
In addition, even though studies of constraint-induced therapy in chronic stroke patients have found that improvements in function of the weaker upper limb are associated with cortical reorganization, similar changes have been found in studies that investigated cortical maps and brain activity on at least two separate occasions after stroke, but did not directly relate these to therapy given in the intervening period. The changes related to constraint-induced therapy were:
* Increase in size of motor representation and in amplitude of motor evoked potential for the paretic target muscle in the lesioned hemisphere (Kopp et al., 1999).
* Relocation of cortical activity during paretic finger movements from the lesioned to the non-lesioned hemisphere (Kunkel et al., 1999).
* Equalization of size of the cortical representation area for the paretic target muscle in the lesioned and non-lesioned hemisphere six months after treatment (Blanton and Wolf, 1999).
The changes related to the unspecified therapy and improvements in the function of the weaker side or good motor recovery were:
* Enlargement of cortical representational area in the lesioned hemisphere (for example, Marshall et al., 2000).
* Amplitude of lesioned hemisphere motor evoked potentials (Traversa et al., 1998).
* Activation of the non-lesioned hemisphere (Honda et al., 1997).
* Activation in brain structures remote from the lesion (Seitz et al., 1999).
Progress requires modelling studies to be undertaken to determine the effects of the elements of constraint-induced therapy, singly and in interaction, before undertaking randomized controlled trials.
THE CHALLENGE OF THE FUTURE
In summary, we need to find ways of administering appropriate and adequate inputs to promote adaptive plasticity. There is much work to be done in defining those inputs and in determining the necessary doses and the appropriate timing, especially since there is some evidence from animal studies that suggests that enhanced activity in the first week after stroke, far from promoting reorganization, may actually threaten a vulnerable ischaemic penumbra around a core lesion and it may result in increasing the size of an infarct (Kozlowski et al., 1996).
It might be objected that new approaches to neurological impairment, such as drugs to promote regrowth, or cellular implants may have more to offer than the most carefully designed activities. Most recent research, however, supports the clinical intuition that if neural circuitry is required rather than neural tangles then you cannot bypass the need for information arising out of normal activity. Feeney and Sutton (1998) noted that the beneficial effect of drugs such as amphetamines were not seen in restrained animals. Mattson and colleagues (1997) reported that implants of fetal neocortical cells produce behavioural benefits in lesion animals only in enriched environments, and Wenk et al. (1999) found that growth factor inhibition as well as stimulation was essential for recovery. In other words, it is not merely the drive to regrowth and reorganization that matters but also the control of growth and reorganization and this can come only from afferent information relating typically to normal activity. We cannot therefore evade the task of defining inputs of the right kind: what dose of what kind of afferent activity for what lesion at what time.
It is important not to be too pessimistic about the current state of neurological rehabilitation. As we have seen, organized stroke care does reduce disability and this is at least in part due to rehabilitation. Indeed, we could save even more people from becoming dependent if organized stroke care were universally available (Ebrahim and Redfern, 1999). Moreover, it would be inappropriate to dismiss the neurorehabilitation research carried out so far.
The message emerging from the present paper, however, will be sufficiently obvious: determining the right inputs to drive neuroplasticity will be very difficult. We have a huge hill to climb if we are going to make further progress in neurological rehabilitation; if, in particular, we are going to exploit in neurotherapy the new understanding coming from basic neuroscience. Even in the simple model of muscle, remote from the complexities of stroke, it has proved very difficult to translate advances in theoretical understanding into practical treatment. We need integrated research programmes in which meaningfully defined lesions are treated with well-characterized, biologically plausible, therapies and outcomes are assessed by use of genuinely informative neurophysiological and clinical measures.
This is a formidable research agenda. How is it going to delivered? First, we need a massive scaling-up of research efforts. The contrast between the amount of investment in developing drug treatments and that deployed in developing physical therapies is staggering, whereas neurological rehabilitation practice has adult spending habits, neurological rehabilitation research certainly does not. Second, our efforts will have to be better co-ordinated; there must be an end to neurological rehabilitation research as a cottage industry. We will not achieve one-tenth of the agenda if we continue to order our affairs in the haphazard way we do at present, pinning hopes of advance on a few scattered enthusiasts. Progress comes from making mistakes as quickly as possible and the rate at which the few enthusiasts have got things informatively wrong or right over the last 20 or 30 years is simply not fast enough. Finally, and most importantly, we need to have clinicians and basic scientists working more closely together, currently the dialogue between neurotherapists and neuroscientists is too intermittent.
If we read the message arising from the current state of the art correctly,
there is an outside chance that within a few decades we will have genuine
science-based therapies rather than the hit and miss remedies that we offer
our patients at present. Then, we should truly be able to build on what
we offer our patients, by adding the benefits of reversing impairments
to our current strategy of helping them to adapt to them.
© 2002, Physiotherapy Research International