STUDIES THAT HAVE examined the time course of motor

Transcription

1 1106 Robotic Therapy for Chronic Motor Impairments After Stroke: Follow-Up Results Susan E. Fasoli, ScD, Hermano I. Krebs, PhD, Joel Stein, MD, Walter R. Frontera, MD, PhD, Richard Hughes, PT, NCS, Neville Hogan, PhD ABSTRACT. Fasoli SE, Krebs HI, Stein J, Frontera WR, Hughes R, Hogan N. Robotic therapy for chronic motor impairments after stroke: follow-up results. Arch Phys Med Rehabil 2004;85: Objectives: To study the effects of robotic rehabilitation in persons with chronic motor impairments after stroke and to examine whether improvements in motor abilities were sustained 4 months after the end of therapy. Design: Pretest-posttest design. Setting: Rehabilitation hospital, outpatient care. Participants: Volunteer sample of 42 persons with persistent hemiparesis from a single, unilateral stroke within the past 1 to 5 years. Intervention: Robotic therapy for the paretic upper limb consisted of either sensorimotor active-assistive exercise, or progressive-resistive training during repetitive, planar reaching tasks, 3 times a week for 6 weeks. Main Outcome Measures: Modified Ashworth Scale, Fugl- Meyer Assessment (FMA), Motor Status Scale (MSS) score, and Medical Research Council motor power score. Results: No significant differences were found among pretreatment clinical evaluations. Statistically significant gains from admission to discharge and from admission to follow-up (P.05) were found on the FMA, MSS score for shoulder and elbow, and motor power score. Conclusions: Short-term, goal-directed robotic therapy can significantly improve motor abilities of the exercised limb segments in persons with chronic stroke that are sustained 4 months after discharge. This suggests that motor recovery can be enhanced by repetitive exercise training more than 1 year after stroke. Key Words: Cerebrovascular accident; Paresis; Rehabilitation; Upper extremity by the American Congress of Rehabilitation Medicine and the American Academy of Physical Medicine and Rehabilitation From the Departments of Mechanical Engineering (Fasoli, Krebs, Hogan) and Brain and Cognitive Science (Hogan), Massachusetts Institute of Technology, Cambridge, MA; Department of Neuroscience, Winifred Masterson Burke Medical Research Institute, Weill Medical College of Cornell University, White Plains, NY (Krebs); Spaulding Rehabilitation Hospital, Boston, MA (Stein, Frontera, Hughes); and Department of Physical Medicine and Rehabilitation, Harvard Medical School, Boston, MA (Stein, Frontera). Supported by the National Institute of Child Health and Human Development (grant nos. R01-HD-36827, R01-HD-37397, F32 HD41795) and the Burke Medical Research Institute. An organization, with which 1 or more of the authors is associated, has received or will receive financial benefits from a commercial party having a direct financial interest in the results of the research supporting this article. Reprint requests to Susan E. Fasoli, ScD, OTR/L, Massachusetts Institute of Technology, Rm 3-147, 77 Massachusetts Ave, Cambridge, MA 02139, sfasoli@mit.edu /04/ $30.00/0 doi: /j.apmr STUDIES THAT HAVE examined the time course of motor recovery after stroke have found that the greatest gains in motor function occur within the first month of onset, with some additional improvement observed up to 6 months poststroke. 1,2 Limited capacity of neural mechanisms to reorganize and regain function after stroke has been thought to account for this plateau in motor abilities, with little to no progress expected after 6 to 12 months. Recent studies of persons with chronic motor impairments after stroke have begun to challenge this expected course of recovery. Intensive therapeutic interventions have led to significant improvements in cortical reorganization and motor function in persons more than 1 year poststroke. 3-8 Empirical evidence suggests that a plateau in motor recovery after stroke may be related to the timing and intensity of stroke rehabilitation services. Recent technologic advances have made it possible to use robotic devices to provide safe and intensive rehabilitation to persons with mild to severe motor impairments after neurologic injury. A robot is capable of controlling and quantifying the intensity of practice and objectively measuring changes in movement kinematics and forces. In addition to providing new options for treatment, this technology may further our understanding of the mechanisms that underlie the recovery of motor function and neural reorganization after stroke. Other randomized controlled studies 9-12 have shown the benefits of upperlimb robot-assisted sensorimotor therapy for persons in the acute phase of stroke recovery. These studies revealed that persons who received robotic therapy had significant gains in motor coordination and muscle strength of the exercised shoulder and elbow that were not observed in the control group. Furthermore, Volpe et al 11 reported that these improvements were sustained during the 3-year period after inpatient hospital discharge. We have extended this research to persons with persistent motor impairments more than 1 year poststroke. Our initial findings with 20 subjects revealed that repetitive, goal-directed robotic therapy led to statistically significant improvements on the upper-limb subtest of the Fugl-Meyer Assessment (FMA), the Motor Status Scale (MSS) score, and motor power assessment. 8 The investigation reported here expands on previous research with persons in the chronic phase of stroke recovery, by examining the effects of robotic rehabilitation in a larger sample of subjects and by investigating whether improvements in motor abilities were sustained 4 months after the end of training. We hypothesized that robotic therapy would produce continued and sustained improvements in upper-limb motor abilities in persons more than 1 year poststroke. METHODS Participants Forty-two community-dwelling persons with chronic stroke (28 men, 14 women) met inclusion criteria and volunteered to participate. Inclusion criteria were (1) diagnosis of a single, unilateral stroke within the past 1 to 5 years verified by brain imaging; (2) sufficient cognitive and language abilities to un-

2 ROBOTIC THERAPY AND CHRONIC STROKE, Fasoli 1107 derstand and follow instructions; and (3) stroke-related impairments in muscle strength of the affected shoulder and elbow between grades 2 and 4 on the Medical Research Council (MRC) motor power scale. Subjects ranged in age from 19 to 79 years (mean standard deviation [SD], y) with an average time poststroke of months. Twenty-eight subjects had a history of right-hemisphere stroke; 14 had incurred left-hemisphere damage. None of the subjects was engaged in conventional occupational or physical therapy programs during the experimental trial, and none had received robotic therapy before this research. One subject (not included here) withdrew because of medical and personal reasons. Two subjects who completed the experimental trial were lost to follow-up (unable to be contacted). All subjects gave informed consent to take part in the study. The experimental protocol was approved by the Human Studies Committee at Spaulding Rehabilitation Hospital, where the robotic therapy was provided, and by the Committee on the Use of Human Experimental Subjects of the Massachusetts Institute of Technology. Measures After subjects provided consent, baseline clinical evaluations to establish motor stability of the involved upper limb were administered at 2-week intervals, during a 1-month observation period before robotic therapy. The same evaluation tools were used to assess the effects of robotic therapy after 3 and 6 weeks of intervention and at the 4-month follow-up visit. The Modified Ashworth Scale 13 (MAS) measured muscle spasticity by rating the resistance to passive stretch in 14 different muscle groups of the upper limb. The FMA of upperextremity function 14,15 examined for the presence of synergistic and isolated movement patterns and grasp. The MRC test of motor power 16,17 measured strength in isolated muscle groups of the involved shoulder and elbow on an ordinal scale, ranging from 0 (no muscle contraction) to 5 (normal strength). The MSS 18 provided a more complete and discrete measure of upper-limb isolated movement and motor function than is possible with the FMA, by grading motor abilities on a welldefined 6-point scale. The MSS has been divided into 2 scales: one for shoulder and elbow movements (which are exercised by the robot) and the second for wrist and finger movements (not exercised by the robot). Reliability of these clinical evaluations has been reported. 13,15,17,18 In addition, subjects were asked to report the amount of pain in the affected shoulder, wrist, and hand by using a Likert-type scale, ranging from no pain (0) to pain with any motion (3). A single therapist, who was blinded to the type of robotic therapy provided, administered all clinical evaluations, to ensure the consistency of testing procedures. In addition to these clinical measures, robotic evaluations were administered before treatment, after 3 and 6 weeks of intervention, and at the follow-up visit 4 months after completing robotic therapy. These robotic evaluations consisted of planar reaching tasks, circle drawings, and isometric holding tests. Data concerning the speed, position, and force of paretic limb movements were collected. To date, we have primarily used these data to measure changes in movement smoothness (eg, jerk, number of peaks in speed profile) over the course of therapy. Results of these analyses have been published elsewhere. 19 Intervention Robotic therapy was delivered by a successor to the MIT- MANUS robot, InMotion2, 20,21,a which was specifically designed and built for clinical, rehabilitation applications. During Fig 1. Person with stroke engaged in robotic therapy using In- Motion2. From Fasoli et al. 8 Reprinted with permission of the American Academy of Physical Medicine and Rehabilitation and the American Congress of Rehabilitation. therapy, subjects were seated comfortably at a table, and their paretic arm was placed in a customized arm support attached to the end-effector (ie, handle) of the robot arm. All subjects were asked to perform goal-directed, planar reaching tasks that emphasized shoulder and elbow movements. As subjects attempted to move the robot s handle toward designated targets, the computer screen in front of them provided visual feedback of the target location and movement of the robot handle (fig 1). A physical or occupational therapist administered each robotic therapy session, ensured proper positioning, and provided verbal instructions and cues, as needed, to orient subjects to the training tasks. Subjects received 1 hour of robotic therapy 3 times a week for 6 weeks, performing approximately 18,000 repetitive reaching movements during the course of therapy. This frequency and duration of therapy was chosen because it is similar to that delivered during conventional outpatient rehabilitation programs. The number of repetitions performed in a given treatment session (dosage) was consistent with that of prior inpatient trials 9,10 and was held constant for each treatment group described below. Two forms of robotic therapy (sensorimotor and progressive-resistive exercise) were provided based on the person s ability to move the involved limb toward targets during the robotic evaluation. 8 Subjects who were unable to move the robot handle to all target locations at admission were assigned to the sensorimotor group. During sensorimotor therapy, the robot offered movement assistance when the person was unable to reach targets independently, much like a therapist provides hand-over-hand assistance during conventional therapy. 21 If subjects were unable to move their arm toward a given target, the robot would provide passive range of motion (ROM). If the subject could initiate but not complete a reach, the robot was compliant to the person s movement attempts and gave active assist as needed. If subjects in this group were able to independently reach all targets without robot assist after 3 weeks of therapy, they were randomly assigned either to continue in the sensorimotor group or to begin progressive-resistive therapy. All robotic exercises were performed in the horizontal plane, regardless of the form of intervention provided. Therapy was directed toward improving combinations of shoulder and elbow movements as the person attempted to move from the center

3 1108 ROBOTIC THERAPY AND CHRONIC STROKE, Fasoli Table 1: ANOVAs for Pretreatment Clinical Evaluations Evaluation* (possible range) 1st Pretreatment Score (mean SD) 2nd Pretreatment Score (mean SD) 3rd Pretreatment Score (mean SD) F 2,40 P r MAS (0 56) FMA (0 66) MSS shoulder/elbow (0 40) MSS wrist/hand (0 42) MRC motor power score (0 40) *For all evaluations except the MAS, higher scores indicate better performance. Repeated-measures ANOVAs were performed on the change scores between the 1st and 2nd pretreatment evaluations, 2nd and 3rd pretreatment evaluations, and 1st and 3rd pretreatment evaluations. Raw scores (rather than change scores) are provided here to indicate initial impairment level before treatment. target to each of 8 peripheral targets (fig 1). Trained movements included forward reach with elbow extension and combinations of shoulder horizontal abduction, adduction, or extension with elbow flexion or extension. Subjects who were able to reach all targets without robot assistance at admission were randomly assigned to either the sensorimotor or the progressive-resistive therapy group. During progressive-resistive therapy, subjects performed the same repetitive, planar reaching tasks performed during sensorimotor training. However, they were required to move against an opposing force generated by the robot. The magnitude of this opposing force was determined and modified via a control algorithm that used robotic measures of the subject s muscle strength to increase or decrease the effort required to reach the targets (see Krebs et al 22 for a detailed description of these robot-based measurements). These measures were obtained at the end of each treatment session to determine the amount of force delivered by the robot during the next session. Subjects were allowed rest periods as needed throughout treatment. Of the subjects who were able to hit all targets at study admission, 7 received sensorimotor therapy and 8 received progressive-resistive therapy throughout robotic training. Of the persons who were able to reach all targets without robot assist after 3 weeks of therapy (but not at admission), 5 received only sensorimotor training and 5 received 3 weeks of sensorimotor therapy followed by 3 weeks of progressiveresistive therapy. Seventeen subjects were never able to hit all targets without robot assist and received only sensorimotor therapy for the duration of the study. The intent of our study was to examine whether persons with stable, chronic upper-limb paresis after stroke would experience sustained improvements in motor outcomes after discharge from robotic therapy, regardless of the type of robotic exercise they performed. Further analyses to compare the effects of sensorimotor versus progressive-resistive robotic therapy are being completed and will be reported separately. Data Analyses Both parametric and nonparametric analyses were performed, and each yielded similar results. For conciseness, we report our parametric analyses of the evaluation change scores here. Repeated-measures analyses of variance (ANOVAs) were used to compare changes in pretreatment scores among the initial baseline evaluations. One-group Student t tests assessed whether the change scores from admission to discharge and from admission to the 4-month follow-up were statistically significant. Nonparametric Wilcoxon signed-rank tests compared pain scores at admission, at discharge, and at follow-up. Because a slight, although nonsignificant, upward trend was noted during the 3 baseline evaluations, the third pretreatment evaluation scores were used as the admission scores for the t tests and Wilcoxon analyses. StatView b was used for data analysis. The strength, or magnitude, of our findings was determined by calculating the effect size r. 23 According to Cohen, r equal to.10 is a small treatment effect, r equal to.30 or greater represents a moderate effect, and r equal to.50 or greater is a large effect. RESULTS Baseline Evaluations No statistically significant differences were found among any of the pretreatment clinical evaluations (table 1), which suggests the stability of chronic motor impairments in this subject group. However, the slight upward trend in these baseline scores could be related to study limitations that were not controlled, such as practice effects due to repeated testing. Nevertheless, it seems unlikely that such small pretreatment effects could account for subsequent changes in clinical scores. Indeed, measures such as the FMA appeared to approach a plateau before treatment. We conclude that the changes observed between admission and discharge may be attributed largely to effects of the robotic therapy described earlier. Findings As in our earlier study, robotic therapy led to significant reductions in motor impairment of the paretic limb from admission to discharge. Statistically significant gains with large effect sizes were found on the FMA, on the MSS for shoulder and elbow, and on the MRC test of motor power (table 2). Clinically, subjects were better able to reach toward visual targets during robotic therapy and reported greater comfort and ease when attempting to move their paretic upper limb. Nonsignificant changes and small effect sizes were found after robotic therapy for muscle spasticity, as measured by the MAS, and for the MSS for wrist and hand, not exercised by the robot (table 2). At admission, 19 of 42 subjects reported mild to moderate pain in the paretic shoulder. When discharged from robotic therapy, shoulder pain was significantly reduced (z 2.35, P.02). Wrist and hand pain were negligible, with no significant differences found from admission to discharge. Follow-Up Results Follow-up evaluations indicated that persons with chronic motor impairments after stroke continued to show improved movement abilities 4 months after discharge from robotic therapy. Statistically significant gains from admission to follow-up were found for the FMA, MSS for shoulder and elbow, and motor power scores. However, the treatment effect sizes at follow-up, indicated by the Cohen r, were smaller than those

4 ROBOTIC THERAPY AND CHRONIC STROKE, Fasoli 1109 Table 2: Comparison of Admission and Scores (N 42) Evaluation* Mean SD t 41 P r MAS Admission FMA Admission MSS shoulder/elbow Admission MSS wrist/hand Admission MRC motor power score Admission *For all evaluations except the MAS, higher scores indicate better performance. MRC motor power was performed only for shoulder and elbow movements. observed at discharge (table 3). Additional 1-group t tests to examine change scores from discharge to follow-up showed the declines in FMA (t , P.02, r.35) and motor power (t , P.05, r.31) scores to reach statistical significance. Figure 2 plots the mean scores for each clinical measure across evaluation sessions. Table 3: Comparison of Admission and Follow-Up Scores (n 40) Evaluation* Mean SD t 39 P r MAS Admission FMA Admission MSS shoulder/elbow Admission MSS wrist/hand Admission MRC motor power score Admission *For all evaluations except the MAS, higher scores indicate better performance. Negative t value indicates that follow-up scores were lower than admission. MRC motor power was performed only for shoulder and elbow movements. Fig 2. Clinical scores across evaluation sessions. Abbreviations. FM, Fugl-Meyer assessment; s/e, shoulder/elbow; w/h, wrist/hand. At follow-up, shoulder pain continued to be significantly less than that at admission (z 2.90, P.004) and did not increase during the 4 months after therapy had ended (z.71, P.48). Wrist and hand pain continued to be negligible, and no significant differences were found from admission to follow-up. DISCUSSION These results extend earlier research on robotic therapy for persons in the acute phase of stroke recovery 9-12 to individuals with persistent motor impairments, and they support our hypothesis that continued improvements in motor abilities are possible in persons more than 1 year poststroke. As in our earlier studies, 9-12 improved performance on the MSS score subtests was limited to the exercised shoulder and elbow, providing further evidence that the benefits of repetitive training are specific to the limb segments exercised. Although the gains in clinical scores were modest, the treatment effect sizes indicated by the Cohen r were consistent with our previous findings 8 and reinforce the efficacy of our robotic therapy methods for persons with chronic motor impairments. These findings indicate that intensive robotic therapy may complement other approaches: it can significantly decrease chronic motor impairments in persons with moderate to severe upperlimb dysfunction, with whom techniques such as constraintinduced movement therapy could not be used. In this group of persons with chronic stroke, the significant and sustained reductions in shoulder pain were an additional benefit of robotic therapy. The percentage of persons with mild to moderate shoulder pain dropped from 45% at admission to 21% at discharge. Clinically, this finding is important because the incidence of hemiplegic shoulder pain after stroke has been reported to be as high as 70%. 24 Although the causes and treatment methods vary widely, recommendations to manage shoulder pain include proper positioning, ROM, and motor retraining exercises. 24,25 Our robotic therapy was in accordance with these recommendations: it provided subjects with the opportunity to engage in safe, well-supported ROM and motor retraining exercises that led to reduced pain and improved motor abilities in the paretic arm. There continues to be limited information available concerning the extent and time course of motor recovery after stroke. Many of the outcome studies used to guide rehabilitation practice examine the recovery of motor function in limited populations (persons receiving hospital-level rehabilitation) and for a relatively short period of time (often within the first

5 1110 ROBOTIC THERAPY AND CHRONIC STROKE, Fasoli 6mo poststroke). 2 It is not surprising that this research shows that the greatest motor improvements occur during the first months after stroke (when the person is experiencing spontaneous neurologic recovery and intensive rehabilitation), with little change observed once these events have concluded. These outcome studies generally do not examine whether the provision of exercise therapy after this acute period can effectively alter the time course of recovery and optimize motor function in the paretic arm. The fact that robotic therapy gains were observed after a period of apparent stability (baseline phase) reinforces that the receptivity to therapeutic intervention does not end after the acute phase of recovery; in fact, it extends beyond the first 6 to 12 months after stroke. Similarly, a growing number of studies 6,26 have reported that other repetitive, goal-directed therapies also improve motor function and cortical reorganization in persons with chronic stroke. As a whole, this empirical evidence challenges current rehabilitation practices, in which less intensive and shorter periods of remediation are provided to persons with substantial motor impairments after stroke. This practice may contribute to learned nonuse 4 and does little to optimize long-term cortical reorganization and motor recovery of the paretic arm. Although clinical scores at follow-up were significantly better than those at admission, small but significant declines were observed in the FMA and motor power scores from discharge to 4-month follow-up. One reason for the decline may be the type of robotic therapy these subjects performed. During this study, robotic therapy emphasized movements of the paretic shoulder and elbow during repetitive, goal-directed tasks. The focus of therapy was to improve proximal coordination and strength, rather than distal control. Motor function of the paretic arm involves not only transport of the limb during reach but also distal grasp, release, and manipulation of task objects. To address this, we have initiated pilot studies to examine the effectiveness of functionally based robotic therapy for the paretic arm after stroke, which emphasizes both proximal and distal control. In addition, our ability to elicit therapy-specific gains in shoulder and elbow movement has motivated our ongoing development of wrist and hand robots for rehabilitation. Alternatively, diminished performance on the FMA and motor power tests at follow-up may be attributed to the lack of ongoing exercise for the paretic arm after discharge from robotic therapy. Daily home exercise programs to reinforce the use of robot-trained movements were not provided. Duncan et al 27 recently found that participation in a structured home exercise program contributed to improved motor recovery in the paretic arm. In addition, Smits 28 reported evidence to support continued therapeutic exercise in a single case study of long-term motor recovery after stroke. Smits quantitatively tracked changes in motor performance (eg, time, number of errors) during a daily exercise program that addressed eye-hand coordination, object manipulation, and controlled movement of the fingers and thumb. When repetitive, daily exercises were performed for more than 1 year (eg, moving pencils from 1 container to another), motor recovery continued and did not reach a plateau. In contrast, Taub et al 4 found that persons whose motor abilities had plateaued after stroke showed significant improvements in motor function of the paretic arm after an intensive, daily exercise program. This research implies that the course of motor recovery after the first3to6 months may be related more to the cessation of exercise and inadequate attempts to functionally use the paretic limb than to neural barriers that cannot be overcome. 28 Additional studies are needed to better quantify the mechanisms that underlie gains in motor function, during both acute rehabilitation and after the period of expected recovery has ended. Several potential limitations of our study deserve mention. Factors such as attention placebo and measurement practice effects might have contributed to the slight upward trend in pretreatment evaluation scores, as well as to the changes observed from admission to discharge. In addition, this was a single-blind study, with the evaluating therapist blinded only to the type of robotic therapy provided. Therefore, assessor bias cannot be ruled out as a potential influence on our clinical data. Finally, we did not administer functional measures of motor performance (eg, the Wolf Motor Function Test 29 or Action Research Arm Test 30 ). Our treatment was focused on proximal training for the paretic shoulder and elbow, and we did not expect to find significant improvements in motor function of the unexercised wrist and hand. We believe that the desired outcome of robotic rehabilitation is not only a reduction in motor impairment but also an improvement in functional use of the paretic limb. Therefore, our future studies are being designed to better address and evaluate an individual s functional motor needs after stroke. CONCLUSIONS The results reported here reinforce our earlier findings that short-term, goal-directed robotic therapy can significantly improve motor abilities of the exercised limb segments in persons with chronic stroke. Clinical scores 4 months after discharge continued to be significantly better than those at admission, which suggests that the time course of motor recovery can indeed be influenced by repetitive exercise training more than 1 year after stroke. Future research will use robotic technology during stroke rehabilitation to quantitatively measure and identify the therapeutic mechanisms (eg, timing, intensity, and duration of therapy; type of task practiced) that contribute to the recovery and retention of functional motor abilities in the paretic limb after stroke. We expect this research to lead to a revised model of motor recovery that may guide rehabilitation professionals and persons with stroke in achieving optimal motor function in the paretic limb. Acknowledgments: We thank Anne McCarthy Jacobson, DPT, MS, NCS, for her involvement and feedback in the clinical evaluation process. References 1. Duncan PW, Goldstein LB, Matchar D, Divine GW, Feussner J. Measurement of motor recovery after stroke. Stroke 1992;23: Hendricks HT, van Limbeek J, Geurts AC, Zwarts MJ. Motor recovery after stroke: a systematic review of the literature. Arch Phys Med Rehabil 2002;83: Miltner WL, Bauder H, Sommer M, Dettmers C, Taub E. Effects of constraint induced movement therapy on patients with chronic motor deficits after stroke. Stroke 1999;30: Taub E, Miller NE, Novack TA, et al. Techniques to improve chronic motor deficit after stroke. Arch Phys Med Rehabil 1993; 4: van der Lee JH, Wagenaar RC, Lankhorst GJ, Vogelaar TW, Deville WL, Bouter LM. Forced use of the upper extremity in chronic stroke patients: results from a single-blind randomized clinical trial. Stroke 1999;30: Liepert J, Bauder H, Wolfgang HR, Miltner WH, Taub E, Weiller C. Treatment-induced cortical reorganization after stroke in humans. Stroke 2000;31: Lum PS, Burgar CG, Shor PC, Majmundar M, van der Loos M. Robot-assisted movement training compared with conventional therapy techniques for the rehabilitation of upper-limb motor function after stroke. Arch Phys Med Rehabil 2002;83:952-9.

6 ROBOTIC THERAPY AND CHRONIC STROKE, Fasoli Fasoli SE, Krebs HI, Stein J, Frontera WR, Hogan N. Effects of robotic therapy on motor impairment and recovery in chronic stroke. Arch Phys Med Rehabil 2003;84: Aisen ML, Krebs HI, Hogan N, McDowell F, Volpe BT. The effect of robot assisted therapy and rehabilitative training on motor recovery following stroke. Arch Neurol 1997;54: Volpe BT, Krebs HI, Hogan N, Edelstein L, Diels CM, Aisen ML. A novel approach to stroke rehabilitation: robot-aided sensorimotor stimulation. Neurology 2000;54: Volpe BT, Krebs HI, Hogan N, Edelstein L, Diels CM, Aisen ML. Robot training enhanced motor outcome in patients with stroke maintained over 3 years. Neurology 1999;53: Volpe BT, Krebs HI, Hogan N. Is robot-aided sensorimotor training in stroke rehabilitation a realistic option? Curr Opin Neurol 2001;14: Bohannon RW, Smith MD. Interrater reliability of a modified Ashworth scale of muscle spasticity. Phys Ther 1987;67: Fugl-Meyer AR, Jaasko L, Leyman I, Olsson S, Steglind S. The post stroke hemiplegic patient. A method for evaluation of physical performance. Scand J Rehabil Med 1975;7: Duncan PW, Propst M, Nelson SG. Reliability of the Fugl-Meyer assessment of sensorimotor recovery following cerebrovascular accident. Phys Ther 1983;63: Medical Research Council/Guarantors of Brain. Aids to the examination of the peripheral nervous system. London: Bailliere Tindall; Gregson JM, Leathley MJ, Moore AP, Smith TL, Sharma AK, Watkins CL. Reliability of measurements of muscle tone and muscle power in stroke patients. Age Ageing 2000;29: Ferraro M, Demaio JH, Krol J, et al. Assessing the motor status score: a scale for the evaluation of upper limb motor outcomes in patients after stroke. Neurorehabil Neural Repair 2002;16: Rohrer B, Fasoli S, Krebs HI, et al. Movement smoothness changes during stroke recovery. J Neurosci 2002;22: Hogan N, Krebs HI, Sharon A, Charnnarong J, inventors. Massachusetts Institute of Technology, assignee. Interactive robotic therapist. US patent 5,466, Nov Krebs HI, Hogan N, Aisen ML, Volpe BT. Robot-aided neurorehabilitation. IEEE Trans Rehabil Eng 1998;6: Krebs HI, Volpe BT, Ferraro M, et al. Robot-aided neuro-rehabilitation: from evidence-based to science-based rehabilitation. Top Stroke Rehabil 2002;8(4): Cohen J. Statistical power analysis for the behavioral sciences. 2nd ed. Hillside: Lawrence Erlbaum; Bender L, McKenna K. Hemiplegic shoulder pain: defining the problem and its management. Disabil Rehabil 2001;23: Gresham GE, Duncan PW, Stason WB, et al. Post-stroke rehabilitation. Clinical practice guideline no. 16. Rockville: US Department of Health and Human Services, Public Health Service, Agency for Health Care Policy and Research; AHCPR Publication No Whitall J, McCombe Waller S, Silver KH, Macko RF. Repetitive bilateral arm training with rhythmic auditory cueing improves motor function in chronic hemiparetic stroke. Stroke 2000;31: Duncan P, Richards L, Wallace D, et al. A randomized, controlled pilot study of a home-based exercise program for individuals with mild and moderate stroke. Stroke 1998;29: Smits JG. Recovery rate constants of recovery from stroke. J Neurovasc Dis 1997;2: Wolf SL, Catlin PA, Ellis M, Archer AL, Morgan B, Piacentino A. Assessing Wolf motor function test as outcome measure for research in patients after stroke. Stroke 2001;32: van der Lee JH, de Groot V, Beckerman H, Wagenaar RC, Lankhorst GJ, Bouter LM. The intra- and interrater reliability of the Action Research Arm test: a practical test of upper extremity function in patients with stroke. Arch Phys Med Rehabil 2001; 82:14-9. Suppliers a. Interactive Motion Technologies, 56 Highland Ave, Cambridge MA b. SAS Institute Inc, 100 SAS Campus Dr, Cary, NC