Visuo-spatial neglect: A systematic review of current interventions and their effectiveness

Transcription

1 Neuroscience and Biobehavioral Reviews ] (]]]]) ]]] ]]] Review Visuo-spatial neglect: A systematic review of current interventions and their effectiveness Jacques Luaute a,b,c,d,, Peter Halligan e, Gilles Rode a,b,c,d, Yves Rossetti a,b,c,d, Dominique Boisson a,b,c,d a Hospices Civils de Lyon, Hôpital Henry Gabrielle, Rééducation Neurologique, Mouvement et Handicap, Saint Genis Laval, F-69230, France b INSERM, U534, Espace et Action, Bron, F-69500, France c Université Lyon 1, Lyon F-69008, France d Institut Fédératif des Neurosciences de Lyon (IFNL), Hôpital Neurologique, Lyon, France e School of Psychology, Cardiff University, Cardiff, UK Received 19 October 2005; received in revised form 3 March 2006; accepted 4 March Abstract Left visuo-spatial neglect is a well-recognized predictor of poor functional outcome following right hemisphere stroke. Over the past 60 years, 18 different methods have been described and evaluated aimed at reducing the effects of this impairment. Although there are some grounds for optimism particularly in terms of short-term impairment-based effects, the range and degree of disability borne by many patients remain high and the clinical effectiveness of the different methods viewed in terms of long-lasting functional improvement (i.e. improvement of disabilities or handicap) is not clear. A systematic review of the available clinically relevant literature, using comparative and stringent levels of evidence, indicates that visual scanning training (VST), trunk rotation (TR) or repeated neck muscle vibrations (NMV) when associated with an extensive training program, mental imagery training, video feedback training and prism adaptation (PA) can be recommended for the rehabilitation of patients with left neglect. More studies however are needed to determine the optimal paradigm of limb activation (LA) eliciting a sustained functional improvement. Sensory stimulations alone and Fresnel prisms do not appear to be functionally relevant. For the other methods, the actual literature is not sufficient to conclude whether or not a long-term functional improvement can be achieved. r 2006 Elsevier Ltd. All rights reserved. Keywords: Left neglect; Stroke; Rehabilitation; Treatment; Systematic review Contents 1. Introduction Visual scanning training (VST) Limb activation (LA) Space remapping Mental imagery training Repetitive trans-cranial magnetic stimulation (rtms) Sustained attention training (SAT) Training at a functional level Feedback training Vestibular stimulation Corresponding author. Hoˆpital Henry Gabrielle, 20, route de Vourles, Saint Genis-Laval, France. Tel.: ; fax: address: jacques.luaute@chu-lyon.fr (J. Luauté) /$ - see front matter r 2006 Elsevier Ltd. All rights reserved. doi: /j.neubiorev

2 2 ARTICLE IN PRESS J. Luauté et al. / Neuroscience and Biobehavioral Reviews ] (]]]]) ]]] ]]] Optokinetic stimulation (OPK) Neck muscle vibration (NMV) and trunk rotation (TR) Fresnel prisms Eye patching Prism adaptation (PA) Music therapy Dopamine-agonists (DA) Noradrenergic agonist (NA) Previous reviews Methods Search terms Databases Inclusion and exclusion criteria Critical analysis: evidential quality of study Results Discussion and recommendation Visual scanning training (VST) and training at a functional level Limb activation (LA) Mental imagery Sustained attention training (SAT) Feedback training Sensory stimulations Eye patching Prism adaptation Conclusion Appendix A References Introduction Stroke continues to represent a leading cause of morbidity in many countries. According to Bonita (1992), at least 70% of the patients who survive one year after stroke remain disabled. Among impairments following stroke, left visuo-spatial neglect is recognized as a significant disabling deficit, clinically defined as a failure to report, respond or orient to novel stimuli presented on the side opposite to a brain lesion (Heilman et al., 1985). It is generally agreed that left neglect associated with a right hemisphere stroke is both more frequent and severe than right neglect associated with a left hemisphere stroke (Stone et al., 1991). Although some spontaneous recovery occurs in the majority of patients after a stroke (Twitchell 1951), left visuo-spatial neglect remains severe in many patients and may persist chronically (Katz et al., 1999; Hier et al., 1983; Samuelsson et al., 1997; Farne et al., 2004). Commonly associated with left hemiplegia, the presence of left visuo-spatial neglect renders motor-associated deficit more severe (Denes et al., 1982). Most clinicians recognize that left visuo-spatial neglect is one of the major factors associated with a poor functional outcome (Denes et al., 1982; Edmans et al., 1991; Fullerton et al., 1988; Jehkonen et al., 2000; Kalra et al., 1997; Stone et al., 1992; Boisson and Vighetto, 1989). Consequently, it is not surprising that over the past 60 years, many different rehabilitation techniques or treatments have been put forward to alleviate, reduce or remediate left visuo-spatial neglect (cf. Fig. 1). In the following review, the 18 different approaches are described, explained in terms of their underlying or motivating mechanisms and finally evaluated in terms of the current evidence base for their effectiveness for clinical rehabilitation. Early rehabilitation approaches were inspired by close clinical observations of the condition. The rationale and the description of the different methods and treatments are discussed briefly and are summarized in Table Visual scanning training (VST) In the early 1970s, visuo-spatial neglect was initially considered a deficit of left side visual exploration and hence the focus on visual scanning training (VST) was to bring about a re-orientation of visual scanning (i.e. nonvoluntary eye movements) toward the neglected side by means of a training program based on providing explicit instructions, in the belief that the preserved language system could be used to help direct the patients voluntary gaze control. According to the model proposed by Robertson and Murre (1999b), who distinguished five different types of approaches (non-specific stimulation, bottom-up targeted stimulation, top-down targeted stimulation, manipulation of inhibitory processes and manipulation of arousal mechanisms), VST is a top-down intervention. Another model that specified the possible mechanisms underlying therapeutic intervention was proposed by Code (2001). This model involved two different processes restoration and compensation which could

3 J. Luauté et al. / Neuroscience and Biobehavioral Reviews ] (]]]]) ]]] ]]] 3 LA Mental imagery training Dopa-agonist NMV rtms VST Fresnel prisms SAT Music therapy Training at a functional level TR Feedback Eye patching NA Space remapping Vestibular stimulation OPK Prism adaptation Fig. 1. Time-line of the first publications for the 18 different attempts to remediate visuo-spatial neglect. Abbreviations: VST: visual scanning training; LA: limb activation; rtms: repetitive transcranial magnetic stimulation; SAT: sustained attention training; OPK: optokinetic; NMV: neck muscle vibration; TR: trunk rotation; NA: noradrenergic agonist. operate at different levels: neural, cognitive and behavioural. Using this framework, VST involves a behavioural compensatory mechanism. From a practical point of view, the training programme is progressive and based on the principles of anchoring, pacing, density and feedback (Diller and Weinberg, 1977). Visual anchors involving visual cues such as a red line are located on the left part of the page and the patient is asked to look at these before commencing the exercises (e.g.: Weinberg et al., 1977). Using this paradigm, several studies reported a significant improvement of the neglect symptoms as assessed by simple paper and pencil tasks. In order to enhance visual exploration to the left, a scanning board, with a small visual target driven from one edge to the other, was sometimes used (e.g.: Young et al., 1983). This scanningbased exercises lead to further improvement on cancellation tasks, scanning strategies in reading and writing as compared to routine occupational therapy Limb activation (LA) In keeping with the results of Halligan et al. (1991) and the need to make use of a perceptual cue present on the left, Robertson and North (1992a) employed the patient s left arm as an endogenous cue. Using single case reports, latter studies showed that active left limb movements in the left hemi-space could significantly reduce visual neglect, compared with no movement, movements performed with the right hand and movements of the left hand performed in the right hemi-space (Robertson et al., 1992b). These results subsequently inspired the development of another rehabilitation technique: Limb activation (LA). The rationale underlying this technique was based on activating a poorly attended body schema by making voluntary initiated contra-lesional limb movements in the left side of the space which in turn were considered to activate corresponding areas of extra-personal space (Robertson and North, 1993a). LA is a bottom-up intervention with behavioural compensation and probably neural and/or cognitive restoration. This method also seems to take advantage of an inter-hemispheric inhibitory process as Robertson and North (1994) showed that the beneficial effects of single left LA in left hemispace could be eliminated if the right limb was simultaneously moved. In clinical practice, the patient is required to initiate movements with his/her left paretic limb in the left part of the space. Robertson et al. (1998a) developed a specific apparatus to elicit LA: the neglect alert device (NAD). This device emits a loud buzzing noise and a red light if the switch is not pressed within a predetermined time interval. The device is placed in the left part of the space and the patient is required to press the switch with his impaired left arm to turn off the buzzer during a variety of situations. Using this apparatus on a patient with severe neglect, Robertson et al. (1998a) showed a significant and lasting improvement of near peripersonal neglect Space remapping Space remapping training is another method that originated from clinical experimental trials. The idea behind this intervention originated from the observation that an elongated stick could produce a virtual extension of body space that resulted in a remapping of far space as near space (Farné and Ladavas, 2000). The principle here was to generalize the effect toward the neglected left space. Thus, this intervention involved a cognitive

4 4 ARTICLE IN PRESS J. Luauté et al. / Neuroscience and Biobehavioral Reviews ] (]]]]) ]]] ]]] Table 1 Short rationale and description of the 18 different methods that have been used to remediate left visuo-spatial neglect Interventions Purpose/rationale Mechanism Procedure VST Left side voluntary eye movement scanning Top-down Behav. compensation Left scanning exercises (sensory cueing; scanning board) LA Using the left limb as an Bottom-up Behav. compensation Movements with the left paretic limb in active cueing device Inhibitory process Neural restoration Cogn. restoration the left part of the space Space remapping Space remapping toward the Bottom-up Cogn. restoration Extension of body space using virtual left neglected side Top-down reality Mental imagery Reducing left representational Top-down Cogn. restoration Visual and movement imagery exercises neglect rtms Inhibition of the relative hyperactivity of the Inhibitory process Cogn. compensation Repetitive transcranial magnetic stimulation over the left parietal cortex unaffected left hemisphere SAT Modulating spatial attention Bottom-up Cogn. compensation Repeated loud sounds stimulation using alerting stimuli Arousal mechanism Training at a functional level Reducing a disability important for the patient Top-down Behav. Compensation Progressive training program depending on the task to be train Feedback training Increasing self-awareness of neglect Bottom-up Cogn. restoration Verbal, visual (video, mirror y) or visuo-motor feedback Top-down Behav. compensation Vestibular stimulation Recalibrating spatial coordinates frames Bottom-up Cogn. restoration Application of cold water in the left external ear canal OPK Illusory rightward Bottom-up Cogn. compensation 7 Leftward moving background displacement of stimuli. cogn. restoration? Recalibrating spatial coordinates frames NMV Recalibrating spatial coordinates frames Bottom-up Cogn. restoration? Vibration of left neck muscles with electrical stimulation TR Recalibrating spatial Bottom-up Cogn. restoration? Left rotation of the trunk coordinates frames Fresnel prisms Shifting the left visual field toward the central retina Bottom-up Behav. compensation Prisms with base directed towards patient s left Eye patching Sprague effect Bottom-up Cogn. compensation Right monocular patch or right halffield Inhibitory process patches PA Visuo-motor adaptation Bottom-up Cogn. compensation Adaptation to prisms requires a set of successive pointing movements Music therapy Simulation of cognitive?? Music processes through sensory and emotional stimulation Dopamine-agonists Stimulate perceptual and premotor systems Bottom-up Cogn. restoration Bromocriptine 15 mg daily for 3 4 weeks Top-down Noradrenergicagonists (NA) Modulation of non-spatial attentional systems Top-down Cogn. restoration Guanfacine 29 mg/kg (oral single dose) Interventions VST: visual scanning training; LA: limb activation; rtms: repetitive transcranial magnetic stimulation; SAT: sustained attention training; OPK: optokinetic; NMV: neck muscle vibration; TR: trunk rotation; PA: prism adaptation. Mechanisms behav.: behavioral; cogn.: cognitive. restoration but is difficult to classify in terms of top-down or bottom-up intervention since both mechanisms are probably implicated. For example, Castiello et al. (2004) used this method in a clinical trial in which they instructed patients with left neglect to reach and grasp a real object in the right space while simultaneously observing the grasping of a virtual object by a virtual hand located in the left space using virtual reality. Results of this study revealed significant improvement in grasping accuracy for the left side of the space following specific training Mental imagery training Other theory-based rehabilitation attempts arose from different theoretical accounts to explain visuo-spatial neglect. For example mental imagery training was inspired by the representational theory of left spatial neglect (Bisiach et al., 1979). The purpose was to reduce left-sided representational neglect by enhancing or training mental imagery and thus invokes a top-down intervention cognitive restoration mechanism. In one study, Smania et al. (1997) used visual and movement imagery exercises

5 J. Luauté et al. / Neuroscience and Biobehavioral Reviews ] (]]]]) ]]] ]]] 5 with 2 right brain damaged patients and showed enhanced left space exploration as measured by various neuropsychological and functional tasks Repetitive trans-cranial magnetic stimulation (rtms) Following Kinsbourne s mutually antagonistic orientation model (1987), a competitive relationship is assumed to exist between each cerebral hemisphere regarding spatial attention. According to Kinsbourne s model, the rightward bias elicited by the left hemisphere is naturally stronger as illustrated, for instance, by the fact that normal subjects have a tendency to veer to the right while walking in an undifferentiated spatial field. After right brain damage, the inter-hemispheric asymmetry regarding spatial orientation is accentuated resulting in a dramatic increase of the rightward attentional bias and generating left visuo-spatial neglect. Following this hypothesis, some authors investigated whether the inhibition of the relative hyperactivity of the left hemisphere could reduce left visuo-spatial neglect. This hypothesis was partially confirmed with repetitive trans-cranial magnetic stimulation (rtms) located over the left parietal cortex using high frequency stimulations (Oliveri et al., 2001). The authors of this study reported a significant improvement of contralesional neglect, as assessed by a line bisection judgment task, in the 7 patients treated by this method. Brighina et al. (2003) reported similar results on 3 neglect patients with low frequency 1 Hz rtms Sustained attention training (SAT) Another method, sustained attention training (SAT), has it origins in connectionist models of neuropsychological function and involved the interaction of two functionally discrete attentional systems with reciprocal connections: arousal and selective spatial attention (Heilman et al., 1987). By stimulating arousal through external alerting stimuli in the general arousal system, it was hoped to enhance the impaired spatial attentional system. In practice, the specific training developed by Robertson et al. (1995b), followed several sequential stages. In the first, the patient carries out a particular task and spatial errors are pointed out. In the second stage, the patient is required to carry out the task again but on this occasion the trainer knocks loudly and unpredictably on the desk on average every s and employs the words attend! in a loud voice. In the final stage, it is hoped that the patient can be trained to self-alert. When applied to 8 patients with left neglect, this specific training was followed by a significant improvement of cancellation tasks Training at a functional level In a small number of reports, authors developed individualized training programs focused on a specific functional skill (e.g. wheelchair navigation, transfer training, balance training, crossing street training, etc.). Systematic training programs were developed depending on the task to be trained and taking account of the patient s own specific strengths and deficits (Stanton et al., 1983). Small steps hierarchical programming and intense repetition of the skills to be trained were commonly employed Feedback training Since left visuo-spatial neglect is commonly associated with anosognosia lack of appropriate awareness for a neurological/neuropsychological deficit and also a recognized contributor to poor outcome some researchers have suggested the need to alleviate anosognosia before an effective training procedure can be implemented (McGlynn and Schacter, 1989). In keeping with this, specific feedback training procedures were developed involving a bottom-up mechanism to produce the feedback (i.e. aimed at restoring self awareness) and a top-down mechanism to compensate for neglect behaviour. For instance, Tham et al. (2001), administered a guided interview to four patients with left neglect during which patient s neglect behaviour was pointed out to him/to her in order to increase selfawareness. Improvement of extra-personal neglect was observed in at least 3 patients. Soderback et al. (1992) video-recorded their 4 patients in order to provide feedback of their neglect behaviour before employing a learning strategy. All these patients exhibited an improvement in their neglect behaviour as assessed by a cancellation task and 3 household tasks. Ramachandran et al. (1999) subsequently used a vertical mirror held parallel to the sagittal plane to provide feedback of the patient s neglect hemi-body. The patient was required to look in the centre of the mirror so he/she could see the reflection of visual input coming from the left side of the space. No quantitative data was reported although the authors of this study claim that the presence of the mirror seemed to help some of the patients to overcome their neglect behaviour. Robertson et al. (1995a) developed a visuo-motor feedback training following the observation that (i) the bisection of a rod was significantly more central when neglect patients grasp the rod as if to pick it up; (ii) repeated trials of grasping and lifting a metal rod at its centre produced a positive effect on several perceptual tasks in a small number of patients. In one controlled trial (Harvey et al., 2003), this training comprised rod lifting exercises over 3 days (different types of wooden rods were presented horizontally in front of the patient, with the middle of the rod in line with the patient s body midline). Patients were asked to reach, lift and balance the rods, with their right hand, at the centre, readjusting their movements until satisfied with the judged central grip. Then patients were trained to self practice. A generally positive effect for this procedure was found using the 6 tests of the conventional BIT.

6 6 ARTICLE IN PRESS J. Luauté et al. / Neuroscience and Biobehavioral Reviews ] (]]]]) ]]] ]]] 1.9. Vestibular stimulation However, as argued by Harvey et al. (2003), all these approaches require the patients to voluntary initiate and maintain attention oriented to the left side, a demanding task in its own right and one that many patients find difficult to apply in everyday life. Another approach to neglect rehabilitation involved the idea of enhancing automatic orientation toward the left space, without the requirement of language mediated attentive learning (i.e. through bottom-up targeted stimulation). Vestibular stimulation was probably the first sensory stimulation to be used in order to alleviate/modulate left visuo-spatial neglect (Silberfenning 1941). Rubens (1985), in a systematic investigation showed that caloric stimulation with leftward nystagmus significantly ameliorated left spatial neglect on several tasks albeit only for a short period. This short term remediation of left visuo-spatial neglect (Cappa et al., 1987; Bisiach et al., 1991; Rode and Perenin, 1994) as well as the demonstration of contralateral cortical activation after vestibular stimulation (Fasold et al., 2002) provides clear and powerful evidence of how simple bottom-up mechanisms can (albeit briefly) overcome high level cognitive deficits. The improvement observed in the case of neglect symptoms following vestibular stimulation depends on the modulation of cortical areas and non cortical areas implicated in or capable of influencing spatial cognition. Rode and Perenin (1994) conceptualized this improvement in terms of the restoration of the internal representation for the egocentric frame. Other sensory stimulations also proved capable of reducing left spatial neglect symptoms: these included optokinetic stimulation (OPK), neck muscle vibration (NMV), trunk rotation (TR) (for a review see Vallar et al. (1997b); Rossetti and Rode (2002)). Again the assumption here was that a specific directional effect in terms of the side of the stimulation and side of the brain would be brought about as a result of the the stimulation. Although there is less direct evidence for these than for vestibular stimulation, it seems likely that these stimulations also work by affecting the activity of cortical networks responsible for calibrating spatial coordinate s frames (Vallar et al., 1997b) Optokinetic stimulation (OPK) The underlying cognitive mechanisms put forward to explain the effect of OPK on attention, involved the displacement of the perception of the line stimuli induced by the retinal optokinetic reflex (Na et al., 2002). Indeed, the left nystagmus (slow eye movements to the left) generated by leftward moving background induces the illusion that visual stimuli are being displaced rightward and thus was considered the compensation process and improvement for left neglect. In practice, the moving background is produced by moving dots (Pizzamiglio et al., 1990; Mattingley et al., 1994) or moving stripes (Bisiach et al., 1996) displayed on a computer screen. The speed of the moving background ranges from 5.11/s (Mattingley et al., 1994) to 71.11/s (Pizzamiglio et al., 1990) and the total duration of this procedure generally last for a few minutes (Pizzamiglio et al., 1990). Different manifestations of neglect have been ameliorated using OPK: line bisection bias (Pizzamiglio et al., 1990; Bisiach et al., 1996; Mattingley et al., 1994) body orientation (Karnath, 1996; Vallar et al., 1995a) and motor neglect (Vallar et al., 1997a) Neck muscle vibration (NMV) and trunk rotation (TR) In the case of NMV and TR, Karnath et al. (1993) proposed that improvement of left visuo-spatial neglect might be explained by a re-centring of the egocentric frame of reference either by illusory (NMV) or real (TR) modification of afferent somesthesic information relative to the position of the head on the trunk. Vibration of left neck muscles can be obtained by trans-cutaneous electrical stimulation (TENS) with superficial electrodes placed on the posterior left neck below the occiput, just lateral to the spine. Frequency of stimulation is 100 Hz, with a pulse width of 100 ms 100 ms and an average intensity of 0.5 ma/ mm ma/mm 2 (Karnath et al., 1993; Vallar et al., 1995a; Guariglia et al., 2000). NMV produced a transient remission of neglect as assessed by cancellation tests (Karnath, 1995; Vallar et al., 1995c; Schindler et al., 2002; Johannsen et al., 2003), visual detection within the left hemispace (Karnath et al., 1993), straight ahead judgement (Karnath, 1994; Schindler et al., 2002) and neglect hemianesthesia (Vallar et al., 1996). Rotation of the trunk which may be obtained by a shoulder strap (Karnath et al., 1993) or by a corset (Wiart et al., 1997) elicited an improvement of visual detection and exploration toward the left hemispace (Karnath et al., 1993) as well as cancellation and line bisection tests (Schindler and Kerkhoff, 1997; Wiart et al., 1997). Other bottom-up interventions involving vision were subsequently developed using Fresnel prisms, eye patching and prism adaptation (PA) Fresnel prisms Rossi et al. (1990) used Fresnel prisms to investigate whether shifting the left visual field toward the central retinal meridian could reduce left spatial neglect. After 4 weeks, the treated group performed significantly better than controls on line bisection and cancellation. In terms of rehabilitation principles, this bottom-up intervention involves a behavioural compensatory mechanism Eye patching Other authors sought to take advantage of the classic Sprague effect. In the original study (Sprague, 1991)

7 J. Luauté et al. / Neuroscience and Biobehavioral Reviews ] (]]]]) ]]] ]]] 7 visual impairments in cats were ameliorated by destroying the superior colliculus on the side opposite to initial visual input. This had the effect of releasing the lesioned hemisphere from the collicular inhibition and thus allowing circuits on the same side as the lesion to now function. Several aspects of left visuo-spatial neglect have also been improved using patching of the patient s right eyes or right hemi-visual field (Butter and Kirsch, 1992; Serfaty et al., 1995; Walker et al., 1996; Arai et al., 1997; Beis et al., 1999; Zeloni et al., 2002). Hence in addition to a bottom-up mechanism, an inhibitory process can also be evoked. The procedure simply requires the patient to wear spectacles containing a right monocular patch or right half-field patches Prism adaptation (PA) More recently, a promising intervention PA was introduced by Rossetti et al. (1998). This took advantage of the effect of visuo-motor adaptation. PA was widely used since the end of the nineteenth century as a paradigm to demonstrate visuo-motor short-term plasticity (Redding et al., 2005). Exposure to prisms produces a lateral shift of the visual field so visual targets appear at a displaced position. Adaptation to such an optical induced shift critically requires a set of successive perceptual-motor pointing movements. While the initial movements tend to approximate to the virtual position of the target, subsequent pointing movements ensure that the pointing error rapidly decreases so that subjects can readily point towards the real target position (Rossetti et al., 1993). This initial error reduction comprises a strategic component of the reaction to prisms and does not necessarily produce adaptation at this stage (Redding and Wallace 1996). To obtain robust negative after-effects after removal of prisms further pointing movements are required. These reinforce the sensory motor adaptation and are considered characteristic of the real or true adaptive component of the adaptation (Weiner et al., 1983). The negative after-effects result from a compensatory shift in manual straight-ahead pointing in a direction opposite to the original visual shift produced by prisms, which may parallel the straight ahead shifts described in neglect patients (see Jeannerod and Rossetti, 1993). Rossetti et al. (1998) proposed that right PA with leftward negative after-effects (using the intact right hand) would improve left neglect symptoms. A significant reduction of left neglect was demonstrated across a variety of different standard tests (line bisection, cancellation, copying and reading) following a brief period (3 5 min) of PA. Subsequent studies have shown that this clinical effects could extend to numerous neglect-related processes such as straight ahead pointing (Pisella et al., 2002), visual exploration toward the left hemispace (Ferber et al., 2003), contralesional somato-sensory perception (McIntosh et al., 2002; Maravita et al., 2003; Dijkerman et al., 2004), temporal order judgment (Berberovic et al. 2004), visuo-verbal tasks (Farne et al., 2002), wheel-chair driving (Jacquin-Courtois et al., in press), postural control (Tilikete et al., 2001) and mental representation (Rode et al., 1998b, 2001b; Rossetti et al., 2004). These cross sectional observations suggest that even after acquired brain damage short-term exposure to visuo-motor adaptation is sufficient to stimulate a long term reorganization of the neural representation of space that develops autonomously after removal of the prisms. The neural basis for this effect in neglect patients has yet to be formally established. One possibility is that PA reduces left spatial neglect by facilitating the recruitment of intact brain areas responsible for controlling normal visuo-spatial output by way of short-term sensori-motor plasticity. However, as indicated by Danckert and Ferber (2006), the gap might be important between what we know about sensori-motor plasticity in normal subjects and what happens in brain damaged neglect patients. Another possibility is that the observed therapeutic effects, and the impressive generalization to high-order processes, are mediated through the modulation of cerebral areas implicated in spatial cognition via a bottom-up mechanism (Rode et al., 2003). A recent functional imaging study argues for this alternative hypothesis since the network of significant brain regions associated with improvement of left neglect performance produced by PA modulated the activity of several cortical areas such as the left temporo-occipital cortex, the left medial temporal cortex and the right posterior parietal cortex (Luaute et al., in press) Music therapy Aside from these bottom-up stimulations, some authors have proposed that music could improve left spatial neglect given that it is an important mediator of emotions and it can affect a large range of sensory and cognitive processes (Landi et al., 1997; Gilbertson and Ischebeck, 2002). However, until now, no convincing experimental data have been reported Dopamine-agonists (DA) Finally, pharmacological agents have been tried to ameliorate visuo-spatial neglect. Dopamine-agonists have been shown to ameliorate some of the classical signs of visuo-spatial neglect such as line bisection, letter cancellation and reading (Fleet et al., 1987, Hurford et al., 1998; Geminiani et al., 1998; Mukand et al., 2001). Conversely, Grujic et al. (1998) reported a worsening of contra-lesional visual exploration in 5 neglect patients treated by Bromocriptine, a dopamine agonist. Supporting evidence for the use of dopamine-agonists comes from the observation that damage to anatomical structures reported to result in left visuo-spatial neglect may be related by a common dopaminergic bond (Fleet et al., 1987). Hurford et al. (1998) and Grujic et al. (1998) justified the use of this neurotransmitter given its implication in perceptual attentional systems. For Geminiani et al. (1998),

8 8 ARTICLE IN PRESS J. Luauté et al. / Neuroscience and Biobehavioral Reviews ] (]]]]) ]]] ]]] dopamine-agonists have a potentially therapeutic utility because this medication could improve the pre-motor component of visuo-spatial neglect. Mukand et al. (2001) argues for the role of dopaminergic networks in the modulation of both perceptive and pre-motor components of visuo-spatial neglect Noradrenergic agonist (NA) Noradrenergic agonist (Guanfacine) has been recently used in 3 patients with left visuo-spatial neglect (Malhotra et al., 2006) with the underlying hypothesis that this neurotransmitter might modulate non spatial attentional processes which have been shown to interact with spatial components of neglect (Husain and Rorden, 2003). The two patients who showed improvement on pen-paper and pencil tasks as well as on visual exploration had a lesion which spared the dorso-lateral pre-frontal cortex. This was not the case for the other patient who did not improve after treatment Previous reviews Several high-quality reviews of neglect rehabilitation already exist. These include selective appraisals of the sensory stimulation s studies and theoretical implications (Vallar et al., 1997b; Chokron and Bartolomeo, 1999; Kerkhoff, 2003). In a recent review, Danckert and Ferber (2006) focused on the different studies related to PA in neglect patients. More wide ranging large reviews have reported the literature available at the time involving the different rehabilitation techniques (Robertson et al., 1993b; Robertson and Halligan, 1999; Bailey and Riddoch, 1999; Rode et al., 2001a; Pierce and Buxbaum, 2002; Rossetti and Rode, 2002). Most of these reviews understandably concluded that most of the interventions reviewed for left visuo-spatial neglect were intuitive, exploratory and as such preliminary. Few if any had been formally evaluated beyond traditional neuropsychological test outcomes. There is only one study that we are aware of that attempted to provide a systematic review which also considered the functional effectiveness of rehabilitation interventions (Bowen et al., 2002). This Cochrane review selected only those controlled trials of cognitive rehabilitation (15 studies) in which at least 75% of the sample were stroke patients and concluded that the effectiveness of rehabilitation for neglect remained unproven. In the context of stroke rehabilitation the motivations for the current review were different in several important respects; namely the need to consider: (i) the functional consequences (the ways by which associated disability can be minimized for the patient) of any existing intervention for visual neglect (ii) the chronicity or long-term effects of the intervention and finally (iii), the evidential quality currently available. For pragmatic purposes such evidence, unlike Bowen et al. (2002), was not limited to controlled trials (e.g. it is important to consider single case studies), to establish an up-to-date consensus on best practice. Indeed, single case designs are most appropriate as they take into account both spontaneous recovery and the heterogeneity of the population studied. 2. Methods Given the above aims we performed a systematic review of the literature using the following search strategy and critical analysis: 2.1. Search terms Patients: neglect/visuo-spatial neglect/visual neglect/ hemi-inattention/stroke/cerebral vascular accident. Intervention: rehabilitation/remediation/treatment/ training/video-feedback/visual scanning/sensorial stimulation/vestibular stimulation/caloric stimulation/galvanic stimulation/optokinetic stimulation/electric stimulation/ neck muscle vibration/trunk rotation/eye patching/prism/ mirror/spatiomotor cueing/mental imagery training/pa/ LA/dopamine agonist/sustained attention Databases In addition to personal databases, we carried out hand searched specific journals (American Journal of Physical Medicine, Annales de Re adaptation et de Médecine Physique, Archives of Physical Medicine and Rehabilitation, Clinical Rehabilitation, Disability and Rehabilitation, Neuropsychological Rehabilitation, Physical Therapy Journal, Scandinavian Journal of Rehabilitation Medicine) and searches of electronic databases (MEDLINE between 1966 and 2005; BIDS EMBASE between 1980 and 2005). Citation-tracking of all primary studies provided an additional search strategy; reference lists from review articles and books identified in the searches were also scanned and then assessed for inclusion Inclusion and exclusion criteria Type of participants: Stroke patients with right brain damage and evidence of neglect on clinical examination and/or classical neuropsychological tests. Type of intervention: Any type of well defined intervention aimed at alleviating neglect. Type of studies: Any type reporting an experimental or observational study. Reviews were excluded. Type of outcome measures: Given our clinical focus, we only considered functional outcome measures together with the duration of the effect. Functional outcomes such as the behavioural sub-tests of the behavioural inattention test (BIT) (Wilson et al., 1987), the Zocollotti semistructured scale (Zoccolotti et al., 1992a) and the Bergego s functional scale (Azouvi et al., 1996) and non-specific functional scales such as the Barthel index (Mahoney and Barthel, 1965), the functional independence measure

9 J. Luauté et al. / Neuroscience and Biobehavioral Reviews ] (]]]]) ]]] ]]] 9 (Granger et al., 1986) and the assessment of motor and process skills (AMPS; Fischer, 1999) were included. Using these constraints the following outcomes were considered: spatial orientation; feeding; dressing; grooming; cleaning; transfers; posture; walking strategy; stairs climbing; wheelchair navigation; communication; social interaction; problem solving; memory; utilising commonly used objects; description of figures, environment; serving tea; card sorting; map navigation; picture scanning; sentence copying; reading and setting time; telephone dialling; handling money; cooking Critical analysis: evidential quality of study Each article was analysed using the analytical grid, taken from the recognized evidence-based medicine levels of evidence used by Ball et al. (2001), but adapted for use with stroke rehabilitation studies: Level 1: This is the gold standard and typically involves well-designed randomized controlled trial (RCT) where the intervention tested is compared to no treatment or to another procedure. If the size of the sample analysed is large enough, it is possible to assume that the improvement due to spontaneous recovery as well as the heterogeneity of the patients are the same in both groups. A follow-up assessment is also required to assess the chronicity of the effect. Level 2: Cohort studies with follow-up assessment in which patients with visuo-spatial neglect received a particular treatment and were followed over time and compared with another group who was not affected by the condition (e.g. stroke without neglect). Level 3: This was assigned to case control studies, crossover studies and well designed single-case studies where spontaneous recovery and the long-term effect were taken into account. Concerning spontaneous recovery, studies in which repeated measures over a first period were used to estimate the rate of change due to spontaneous recovery (for single case methodology, see Pring (1986)). According to Katz et al. (1999), who studied spontaneous recovery in neglect patients, it tends to occur predominantly during the first 5 months after onset (For more details concerning the different patterns of spontaneous recovery in neglect patients, see also Farne et al. (2004)). It should be noted that a well designed single-case study provides crucial information and can sometimes be very helpful when integrating the heterogeneity of clinical patients throughout in terms of brain lesions and impairments. At the level of disability or handicap the heterogeneity is even greater and one could consider that the problem presented by one patient is unique and thus could only really be assessed by a single-case study approach. Concerning the long-term effect, a follow-up assessment was required. Level 4: Other types of studies without a specific assessment of spontaneous recovery, patient s heterogeneity or long-term effect. Level 5: Expert opinion without explicit critical appraisal, or based on physiology, bench research or first principles. Grades of recommendation for clinical practice (Ball et al., 2001): Grade A: Consistent level 1 studies showing long-term functional gains. Grade B: Consistent level 2 or 3 studies or extrapolations from level 1 studies showing long-term functional gains. Grade C: Consistent level 4 studies or extrapolations from level 2 or 3 studies showing functional gains but without specific assessment of spontaneous recovery, patient s heterogeneity and/or long-term effect. Grade D: Consistent level 5 studies or troubling inconsistent or inconclusive studies of any level showing functional gains but without specific assessment of spontaneous recovery, patient s heterogeneity and/or long-term effect. Grade E: Evidence for the absence of long-term functional gain. Grade non-conclusive (NC): No evidence neither for a long-term functional gain nor for the absence of effectiveness or discordant results. 3. Results One hundred and sixteen articles were found that fulfilled the criteria for this systematic search dating from 1962 until 2006 (In press studies that we are aware of were also analyzed). Thirty-five studies were excluded as they were not specifically concerned with formally assessing the efficacy of a therapeutic intervention. Five further studies were excluded as individualized data for stroke patients with left visuo-spatial neglect was not available. Twenty two studies were excluded as the outcomes did not involve functional measures. Consequently, fifty-four studies were included in the current review. The lists of references, including the breakdowns relevant for exclusion have been referred to in Table A.1. in the Appendix. Results concerning the articles included in our review are presented in Table 2. A graphic summary of the iterative stages of exclusion and inclusion for the papers reviewed is shown in Fig Discussion and recommendation Like in other domains of neuropsychological rehabilitation (e.g. memory see Wilson et al., 2001 ; aphasia see Howard, 1986 ), the heterogeneity of the population targeted, the lack of a clear patho-physiologycal understanding, and the variability of spontaneous recovery, requires well designed studies to investigate the effectiveness of any treatment procedure for left visuo-spatial neglect. On the basis of our current analysis only 21 studies (cf. studies in bold within Table 3) 7 RCT and 14 singlecase studies properly investigated the effectiveness of a treatment in stroke patients with left visuo-spatial neglect

10 10 ARTICLE IN PRESS J. Luauté et al. / Neuroscience and Biobehavioral Reviews ] (]]]]) ]]] ]]] Table 2 Evidence quality and functional outcome reported in studies aimed at alleviating left neglect Articles Ref. Interventions Evidence quality Control Patients (n) Duration Outcome Results Exp C Immediate Long term Weinberg et al. (1977) VST RCT Classical OT w Reading; Copying Improvement Improvement 41 yr* Weinberg et al. (1979) VST RCT Classical OT w Reading; Copying Improvement No follow up Young et al. (1983) VST RCT Classical OT w Reading; Copying Improvement Improvement 46 mth* Gordon et al. (1985) VST RCT Classical OT w Reading; Copying Improvement No improvement at 4 mth* Robertson et al. (1990) VST (CAT) RCT Computer activity w Reading; B-BIT No improvement No improvement* Antonucci et al. (1995) VST RCT Gle cognitive rehab w Reading; Zoccolotti Improvement No follow up Paolucci et al. (1996) VST RCT Gle cognitive rehab w Reading; Zoccolotti Improvement No follow up Kerkhoff (1998) VST (CAT) Cross over Visuo-spatial training w Reading Improvement No follow up Wagenaar et al. (1992) VST Single case MB 5 2 w WC navig. No improvement No follow up Pizzamiglio et al. VST Single case MB 13 8 w Reading; Zoccolotti Improvement Improvement 45 mth* (1992) Fanthome et al. (1995) VST Single case MB 14 4 w Reading; B-BIT Improvement No follow up Lawson (1962) VST Case report 2 6 h Reading improvement Improvement 43 mth Vallar et al. (1997a, b) VST Case report 8 8 w reading Improvement No follow-up Wiart et al. (1997) VST+TR RCT Usual rehab w FIM Improvement Improvement 41 mth* Schindler et al. (2002) VST+NMV Cross over VST w Reading; personal care spatial orientation Pizzamiglio et al. (2004) Zoccolotti et al. (1992a, b) Marshall and Maynard (1983) Improvement Improvement 42 mth VST+OPK RCT VST w BI No improvement No improvement VST+OPK Case report 1 6 w Reading; Zoccolotti Improvement No follow up VST+VS Case report 1 3 w Balance; Hygiene Improvement Improvement 42 yr Brunila et al. (2002) VST+LA Single case MB 4 3 w Reading Improvement Improvement 43 w Webster et al. (2001) Functional training Case control Gle cognitive rehab w WC navig. improvement No follow up Webster et al. (1984) Functional training Single case MB 3 4 w WC navig. Improvement Improvement 41 yr Webster et al. (1988) Functional training Single case MB 10 4 w WC navig. Improvement No follow up Gouvier et al. (1984) Functional training Case report 2 2 w WC navig. Improvement Improvement 46 w Gouvier et al. (1987) Functional training Case report 5 2 w Reading; WC navig. Improvement y No follow u Stanton et al. (1983) Functional training Case report 1 6 w Transfer Improvement Improvement Matjacic et al. (2003) Functional training Case report 1 2 w Balance Improvement No follow up Kalra et al. (1997) LA (SMc) RCT Classical OT w BI Improvement z No follow up Hospital duration Reduction Robertson et al. (2002) LA RCT VST w Motor functions Improvement Improvement42 yr BI; Bergego; B-BIT No improvement No improvement Robertson et al. (1994b) Robertson and North (1994) LA Case control No hand mvt 6 1 sess Walking strategy Improvement No follow up LA Case control R & bilat. limb mvt 2 10 w Reading Improvement No follow up

11 J. Luauté et al. / Neuroscience and Biobehavioral Reviews ] (]]]]) ]]] ]]] 11 Worthington (1996) LA Single case ABACA Robertson et al. (1998a) B: VMc C: SMc 1 10 w Reading Improvement Improvement418mth LA (NAD) Single case MB 1 18 d Combing; Navig.; BTT Improvement Improvement 49 d z Samuel et al. (2000) LA (SMc) Single case ABAB A: VST; B:SMc 2 8 w Bergego Improvement Improvement41 mth Wilson et al. (2000) LA Single case MB 1 10 d Self care routines Improvement Improvement 410 d Robertson et al. LA (NAD) Single case 1 5 d Walking trajectory improvement Improvement o3 w (1992b) Smania et al. (1997) Mental imagery Single case MB 2 8 w Zocolloti Improvement Improvement 46 mth Robertson et al. SAT Single case MB 8 5 d BTT Improvement No follow up (1995b) Soderback et al. (1992) Feedback Single case MB 4 26 d Cooking task Improvement Improvement 42 mth Tham et al. (2001) Feedback Single case MB 4 4 w AMPS; BTT Improvement Improvement 49 w Harvey et al. (2003) Feedback RCT VM training w B-BIT; BI No improvement No improvement Rubens (1985) Caloric stim. Case control L stimulation/nle min Reading Improvement Improvemento15 min Rode et al. (1998c) Caloric stim. Case control R hemiparetic/ Nle min Posture Improvement No follow up Guariglia et al. (2000) NMV Case control No NMV min Walking strategy Improvement No follow up Perennou et al. (2001) NMV Case control No NMV 6 20 min Posture Improvement No follow up Rossi et al. (1990) Fresnel prisms RCT No goggles 6 4 w BI No improvement No follow up Beis et al. (1999) R eye patching RCT No patches mth FIM Improvement No follow up Walker et al. (1996) R eye patching Single case ABC B:L patch C: R patch 9 1 sess Reading No improvement No follow up Butter and Kirsch R eye patching Case report 3 1 sess Reading Improvement No follow up (1992) Rossetti et al. (1998) Prism adaptation RCT Neutral goggles sess Reading improvement No follow up Frassinetti et al. (2002) Prism adaptation CT Gle cognitive rehab w Reading; B-BIT Improvement Improvement45 w Farne et al. (2002) Prism adaptation Single case ABA 6 1 sess Reading Improvement Improvement 41 d McIntosh et al. (2002) Prism adaptation Case report 1 3 sess Reading Improvement No follow up Rode et al. (in press) Prism adaptation Single case MB 1 1 sess Writing Improvement Improvement 448 h Jacquin-Courtois et al. Prism adaptation Single case MB 1 1 sess WC navig. Improvement Improvement 496 h (in press) Interventions VST: visual scanning training; CAT: computer-assisted training; WC: wheelchair; LA: limb activation; SMc: spatiomotor cueing; NAD: neglect alert device; Stim.: stimulation; OPK: optokinetic; TR: trunk rotation; NMV: neck muscle vibration; R: right. Design RCT: randomized control trial; MB: multiple baseline; CT: control trial. Control (procedure followed by the patients in the control group) OT: occupational therapy; spe: specific; Gle: general; rehab: rehabilitation; mvt: movement; L: left; R: right; bilat.: bilateral; VMc: visuomotor cueing; SMc: spatiomotor cueing; Nle: normal; NMV: neck muscle vibration; VM: visuomotor. Level refers to the level of evidence (cf. details in the text). Patients: n ¼ number of patients in the experimental group (Exp) and in the control group (C). Duration refers to the duration of the procedure: w: week(s); d: day(s); h: hour(s); min: minute(s); sess: session. Outcome B-BIT: Behavioral BIT (Wilson et al., 1987); Zoccolotti: Zoccolotti semi structure scale (Zoccolotti et al., 1992a); WC navig.: Wheelchair navigation; Tel.: Telephone; BI: Barthel index (Mahoney and Barthel, 1965); Bergego: Bergego s functional scale (Azouvi et al., 1996); FIM: functional independence measure (Granger et al., 1986); BTT: Baking tray task; AMPS: Assessment of motor and process skills (Fischer, 1999). Results: yr: year; mth: month(s); w: week(s); d: day(s); h: hour(s); min: minute(s). *: partial data; y: improvement for some patients but not all the group; z improvement on some tests but not all the test; z: trend toward improvement but not statistically significant. In bold: Studies with long term improvement of functional skills.