American Journal of Physical Medicine and Rehabilitation

July 2007, Volume 86, Issue 7 , pp. 513-601

Special Section Introduction 513 Poststroke and Brain Injury Rehabilitation: Spatial Neglect. Anna M. Barrett, MD; Charles E. Levy, MD; Leslie J. Gonzalez Rothi, PhD Special Section Research Article: Brain Injury 515 Influence of Moving Background on Line Bisection Performance in the Normal Elderly Versus Patients with Hemispatial Neglect. Kyung Mook Choi, MS; Byung Hwa Lee, MA; Sang Chan Lee, MD; Bon D. Ku, MD; Eun-Joo Kim, MD; Mee Kyung Suh, MS; Yong Jeong, MD; Kenneth M. Heilman, MD; Duk L. Na, MD 527 Amantadine Treatment of Hemispatial Neglect: A Double-Blind, Placebo- Controlled Study. Laurel J. Buxbaum, PsyD; Mary Ferraro, PhD; John Whyte, MD, PhD; Arthur Gershkoff, MD; H Branch Coslett, MD Research Article Cerebral Palsy 538 Kinematic Aiming Task: Measuring Functional Changes in Hand and Arm Movements After Botulinum Toxin-A Injections in Children with Spastic Hemiplegia. Eugene A.A. Rameckers, PT, MRes; Lucianne A.W.M. Speth, MD; Jacques Duysens, PhD; J S. Hans Vles, PhD, MD; Bouwien C.M. Smits-Engelsman, PhD 548 Effect of Treadmill Training with Body Weight Support on Gait and Gross Motor Function in Children with Spastic Cerebral Palsy. Rong-Ju Cherng, PhD, PT; Chuan-Fei Liu, MS, PT; Tin-Wai Lau, MS, PT; Rong-Bin Hong, MD Research Article Hemiparesis 556 Gait and Balance Performance Improvements Attributable to Ankle-Foot Orthosis in Subjects with Hemiparesis. Ray-Yau Wang, PT, PhD; Pei-Yi Lin, PT; Chao-Chung Lee, PT, MS; Yea-Ru Yang, PT, PhD Research Article Trauma 563 Predicting Discharge of Trauma Survivors to Rehabilitation: A Sampling Frame Solution for a Population-Based Trauma-Rehabilitation Survey. Marie-Josée Sirois, PhD; André Lavoie, PhD; Clermont E. Dionne, PhD

Research Article Walking 574 The Reliability of Walking Tests in People with Claudication. Inácio Teixeira da Cunha-Filho, PT, PhD; Danielle Aparecida Gomes Pereira, PT; André Maurício Borges de Carvalho, MD; Leilane Campedeli, PT; Michelle Soares, PT; Joyce de Sousa Freitas, PT Literature Review Physiology 583 Space Flight Rehabilitation. Michael W.C. Payne, MD, MSc; David R. Williams, MD, MSc; Guy Trudel, MD, MSc Case Report Stroke 592 Neuromuscular Electrical Stimulation During Task-Oriented Exercise Improves Arm Function for an Individual with Proximal Arm Dysfunction After Stroke. Lois D. Hedman, PT, MS; Jane E. Sullivan, PT, DHS; Marjorie J. Hilliard, PT, MS; Deborah M. Brown, DPT Case Report Neuropathy 597 Sciatic Neuropathy after Lower-Extremity Trauma: Successful Treatment of an Uncommon Pain and Disability Syndrome in an Adolescent. John M. Saroyan, MD, FAAP; Christopher J. Winfree, MD; William S. Schechter, MD, MS, FAAP; David Roye, MD; Arnold P. Gold, MD Visual Vignette 601 A Thoracic Disc Herniation Presenting as an Abdominal Hernia. Myron M. LaBan, MD, MMSc; Gabriel Gorin, MD

Authors: Anna M. Barrett, MD Charles E. Levy, MD Special Section Leslie J. Gonzalez Rothi, PhD

Afﬁliations: From the Kessler Medical Rehabilitation Research and INTRODUCTION Education Center, West Orange, New Jersey (AMB); Departments of Physical Medicine and Rehabilitation and Neurology/Neurosciences, University of Medicine and Dentistry, Poststroke and Brain Injury New Jersey–New Jersey Medical School, Newark, New Jersey (AMB); Rehabilitation Brain Rehabilitation Research Center (CEL, LJGR), Physical Medicine and Spatial Neglect Rehabilitation Service (CEL), and Geriatric Research, Education and Clinical Center, Malcolm Randall VA Medical Center, Gainesville, Florida (LJGR); and the Departments of ABSTRACT Neurology (LJGR) and Occupational Barrett AM, Levy CE, Gonzalez Rothi LJ: Poststroke and brain injury rehabilita- Therapy (CEL), University of Florida, Gainesville, Florida. tion: spatial neglect. Am J Phys Med Rehabil 2007;86:513–514.

Correspondence: All correspondence and requests for reprints should be addressed to Dr. n last month’s issue (June 2007), we introduced the first of a four-issue special Anna M. Barrett, Stroke/KMRREC, I 1199 Pleasant Valley Way, West series of The American Journal of Physical Medicine & Rehabilitation, focused 1 Orange, NJ 07079. on innovative, physiologic treatments for stroke and traumatic brain injury. These disorders are leading causes of adult disability in the United States today, Disclosures: accounting for tremendous personal, social, and financial costs for survivors, Funded in part by support to Dr. caregivers, and society. Barrett by the National Institute of Brain injury survivors with spatial neglect (defined as a failure to report, Neurological Disorders and Stroke respond, or orient to unilateral stimuli, associated with functional disability) and the Henry H. Kessler face special challenges.2,3 They may receive no evaluation relevant to their Foundation, and by support to Dr. visual-spatial disorder, or only a brief screening for cognitive dysfunction.4,5 Gonzalez Rothi by the Rehabilitation Research and Development Service of They may receive little counseling or management in the short- or long-term the Department of Veterans Affairs. stages of recovery—and even if we assume that only 10% of people with spatial neglect have chronic symptoms, more than 200,000 United States stroke sur- 0894-9115/07/8607-0513/0 vivors may still have neglect-related disability.6,7 American Journal of Physical In this issue, we are very pleased to present two articles on treatment innova- Medicine & Rehabilitation tion for spatial neglect. Choi and colleagues8 discuss the effect of optikinetic Copyright © 2007 by Lippincott stimulation on an impairment measure in spatial neglect, suggesting that a simple Williams & Wilkins computerized visual stimulus may reduce the manifestations of this disabling

DOI: 10.1097/PHM.0b013e31806e79fc disorder. Although not standard care, this low-risk, inexpensive intervention de- serves further study. Buxbaum and colleagues9 examine treatment of spatial neglect with amantadine, an NMDA antagonist with anticholinergic-like effects that is also used to treat viral infections and Parkinson disease. Their study does not support the wide use of this agent to improve neglect symptoms, but they report important information on methodology for successful neglect treatment trials. In the August and September issues, we will present reports on pharmaceutical augmentation of rehabilitation and on reﬁning existing physiologic treatment methods; we invite you to look for the continuation of this special series in those issues.

REFERENCES 1. Barrett AM, Levy CE, Rothi LJG: Treatment innovation in rehabilitation of cognitive and motor deﬁcits after stroke and brain injury: physiological adjunctive treatments. Am J Phys Med Rehabil 2007;86:423–5

July 2007 Spatial Neglect 513 2. Heilman KM: Neglect and related disorders, in Heilman KM, tistics—2006 Update. Dallas, Tex, American Heart Associa- Valenstein E (eds): Clinical Neuropsychology,ed1.New tion, 2006 York, NY, Oxford University Press, 1979 7. Barrett AM, Buxbaum LJ, Coslett HB, et al: Cognitive reha- 3. Barrett AM, Burkholder S: Monocular patching in subjects bilitation interventions for neglect and related disorders: with right hemisphere stroke affects perceptual-attentional moving from bench to bedside in stroke patients. J Cogn bias. J Rehabil Res Dev 2006;43:337–46 Neurosci 2006;18:1223–36 4. Foerch C, Misselwitz B, Sitzer M, Berger K, Steinmetz H, 8. Choi KM, Lee BH, Lee SC, et al: Inﬂuence of moving back- Neumann-Haefelin T: Difference in recognition of right and ground on line bisection performance in the normal elderly left brain stroke. Lancet 2005;366:392–3 versus patients with hemispatial neglect. Am J Phys Med Rehabil 2007;86:515–26 5. Chodosh J, Petitti DB, Elliott M, et al: Physician recognition 9. Buxbaum LJ, Ferraro M, Whyte J, Gershkoff A, Coslett HB: of cognitive impairment: evaluating the need for improve- Amantadine treatment of hemispatial neglect: a double- ment. J Am Geriatr Soc 2004;52:1051–9 blind, placebo-controlled study. Am J Phys Med Rehabil 6. American Heart Association. Heart Disease and Stroke Sta- 2007;86:527–37

514 Barrett et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 Authors: Kyung Mook Choi, MS Byung Hwa Lee, MA Brain Injury Sang Chan Lee, MD Bon D. Ku, MD Eun-Joo Kim, MD Mee Kyung Suh, MS Yong Jeong, MD RESEARCH ARTICLE Kenneth M. Heilman, MD Duk L. Na, MD

Afﬁliations: From the Department of Neurology, Influence of Moving Background on Samsung Medical Center, Sungkyunkwan University School of Line Bisection Performance in the Medicine, Seoul, Korea (KMC, BHL, EJK, MKS, DLN); Department of Normal Elderly Versus Patients with Neurology, Dongeui Medical Center, Busan, Korea (SCL); Department of Hemispatial Neglect Neurology, Kwandong University College of Medicine, Myongji Hospital, Gyeonggi, Korea (BDK); Department of BioSystems, Korea ABSTRACT Advanced Institute of Science and Technology, Daejeon, Korea (YJ); and Choi KM, Lee BH, Lee SC, Ku BD, Kim E-J, Suh MK, Jeong Y, Heilman KM, Na the Department of Neurology, DL: Influence of moving background on line bisection: performance in the normal University of Florida and Veterans elderly versus patients with hemispatial neglect. Am J Phys Med Rehabil 2007; Affairs Medical Center, Gainesville, Florida (KMH). 86:515–526. Objective: Previous studies have shown that when healthy young partic- Correspondence: ipants bisect stationary lines on a moving background (MB) or optokinetic All correspondence and requests for stimulation, they perceive the stationary line moving in the opposite direction reprints should be addressed to Duk L. Na, MD, Department of Neurology, of the MB (illusory motion [IM]), and they displace their bisection mark in the Sungkyunkwan University, Samsung direction of the IM. This study attempted to learn whether IM also influences Medical Center, 50 Ilwon-dong Kangnam-ku Seoul, 135-710 Korea. attentional biases of the healthy elderly and patients with hemispatial neglect. Design: In experiment 1, healthy elderly participants and patients with Disclosures: neglect bisected lines in conditions where IM was absent or present. To This study was supported by better understand the MB dichotomy between the healthy elderly and Samsung Medical Center Clinical Research Development Program grant neglect patients, in experiment 2, participants’ eye movements were (CRS-104-04-2) and by a grant from recorded using an infrared eye tracker. the Korea Health 21R&DProject, Ministry of Health & Welfare, Results: In experiment 1, healthy elderly participants’ biases occurred Republic of Korea (A050079). in the opposite direction of MB when IM was present but in the same 0894-9115/07/8607-0515/0 direction of MB when IM was absent. In contrast, neglect patients’ biases American Journal of Physical occurred in the same direction of MB regardless of conditions. Eye Medicine & Rehabilitation movements reflect the spatial direction of attention. In experiment 2, the Copyright © 2007 by Lippincott Williams & Wilkins healthy elderly participants were able to selectively attend to the line, whereas neglect patients were impaired in that they fixated on the line. DOI: 10.1097/PHM.0b013e31806e0870 Conclusions: These results suggest that the healthy elderly can selectively fixate on a line, and with MB, they perceive the stationary line moving, resulting in a bisection bias in the direction of the IM. In contrast, when there is an MB, the patients with neglect are impaired in that they fixate on the line. Thus, they do not perceive IM; instead, they are primarily influenced by the MB.

Key Words: Attention, Line Bisection, Illusory Motion, Optokinetic Stimulation

July 2007 Line Bisection on Moving Background 515 the portion of a line to which healthy participants Hemispatial neglect is the failure to report, overtly attend when they are watching lines in real orient to, or respond to novel or meaningful stim- motion or with IM. These investigators found that uli presented to the side opposite a brain lesion, with actual line movement, participants’ visual fix- when this failure cannot be attributed to either ations centered on the leading part of the line. The sensory or motor defects.1 Many different maneu- participants’ attention skewed in the direction of IM, vers have been reported to reduce the spatial bias further supporting the postulate that IM can modu- associated with hemispatial neglect. The mecha- late the direction of healthy people’s attention. nisms underlying these therapeutic interventions Studies of patients with neglect performing have not been fully elucidated, but a subset of line-bisection tests with MBs have revealed that interventions seems to induce an illusion of stimulus these patients focus in the direction of the MB.8,9,12 or environmental movement in viewer-centered If the MB had elicited IM, their attention would space. These include caloric stimulation of the ear,2 have been directed away from the MB. In these neck–muscle vibration,3,4 transcutaneous electrical studies, however, the MB might not have been nerve stimulation,5–7 and optokinetic stimulation specifically designed for producing IM. Alterna- (OKS) or moving background (MB).8–14 tively, the ability of these patients to attend to a One of the first treatments, cold-water irriga- stationary object on an MB might be altered; thus, tion of the left ear in patients with left hemispatial they might perform differently on this task com- neglect, causes deviation of the eyeball to the left pared with healthy controls. The aim of this study and rapid corrective eye movements toward right, was to investigate whether the influence of IM on presumably shifting the viewer’s center reference the attentional biases of healthy people is different frame to the left, and resulting in temporary im- from the attentional biases of patients with hemis- provement of the signs associated with left ne- patial neglect, and, if so, what mechanisms might glect.2 Similarly, a leftward MB or OKS can also account for these differences. induce similar eye movements (slow leftward and 8 rapid corrective rightward eye movements), sug- EXPERIMENT 1 gesting that an MB may have the same beneficial effects on neglect as caloric stimulation. Neck– Experiment 1 examined the influence of MB on muscle vibration and transcutaneous electrical line-bisection performance in the healthy elderly nerve stimulation also reduce the spatial bias asso- vs. patients with hemispatial neglect. ciated with neglect; these treatments might work in a similar manner.3,4 MATERIALS AND METHODS Several studies have suggested that movement Participants of the background can influence the spatial bias as- Participants consisted of 22 healthy volunteers sociated with hemispatial neglect and can even in- and nine patients with left hemispatial neglect duce attentional biases in healthy individuals.8–14 For from right-hemisphere cerebral infarctions. The example, when patients with neglect were asked to healthy participants were the spouses of neurology bisect stationary horizontal lines superimposed on an outpatients at Samsung Medical Center, Seoul, Ko- MB, the MB caused their attempts at line bisection to rea and had no history of neurologic or psychiatric deviate, compared with no movement condition, in illnesses. Patients with stroke confined to the right the direction of MB.8,9,12 hemisphere were consecutively recruited for a When people look at a stationary object on an 6-mo period from the neurology inpatient service MB, the stationary object might seem to move in at Samsung Medical Center, Seoul, Korea and at the direction opposite the background motion, a Dongeui Medical Center, Busan, Korea. Patients phenomenon called illusory motion or induced who had decreased consciousness, severe cognitive motion (IM). Na et al.15 have demonstrated that in deficits, inability to understand test instructions, a condition where IM is absent, line-bisection er- or who refused to give consent were excluded from rors (LBEs) of healthy participants occurred in the the study. All participants were right-handed as same direction of MB, replicating the results of assessed by the Edinburgh handedness question- previous studies; when the same participants bi- naire.18 The participants demonstrated binocular sected the lines in conditions where IM was visual acuity above 20/30, either uncorrected or present, LBEs occurred in a direction opposite that corrected. An informed consent, approved by the of the MB and in the direction of the IM. institutional review board of each hospital, was Choi et al.16 have replicated and extended obtained from each participant. these results, showing in healthy young partici- Patients with stroke received a series of tests pants that the degree of IM correlates with the for neglect, including 1) solid-line bisection, 2) magnitude of line-bisection biases. Jeong et al.,17 letter-line bisection, 3) star-line bisection, 4) Albert using an infrared eye tracker, attempted to learn line cancellation, 5) star-cancellation task, and 6)

516 Choi et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 copying pictures (two daisy and the modified Og- rounded by a large MB. To examine whether IM den picture). The time interval from the onset of influences line-bisection performances of partici- stroke to the neglect assessment ranged from 1 to pants, we had participants bisect lines in the con- 183 days (mean 54.7 Ϯ 59.5). Details of adminis- dition where IM is presumably present (large-back- tration and scoring methods have been described ground condition) or absent (small-background previously.19 The presence of hemispatial neglect condition) (Fig. 1). was defined on the basis of norms established pre- The large background, as illustrated in Figure viously from 81 healthy participants.19 Briefly, 2A, consisted of alternating yellow and black ver- each score from three bisection and two cancella- tical stripes that were 1.85 cm (2.65 degrees/sec) in tion tasks was converted to a 10-point scale that width. A stationary horizontal line that was red and ranged from Ϫ10 to ϩ10, with the plus and minus 12.5 ϫ 0.2 cm (18.81 ϫ 0.29 degrees) in size was values representing leftward and rightward bias. centered on this background. The background Seven points were given to the two figure-copying moved leftward or rightward at speeds of 1.26 cm/ tasks. There are four objects in the modified Ogden sec (1.81 degrees/sec) (S1), 11.07 cm/sec (15.76 scene and three objects within the two-daisy figure degrees/sec) (S2), or 20.87 cm/sec (29.25 degrees/ (i.e., left daisy, right daisy, and pot). For each sec) (S3). A previous study has shown that the ϩ object, we scored 1 for the left scene-based ne- background moving at these speeds induces IM of ϩ glect, and 0.5 for the left object-centered neglect. the stationary line centered in the background.21 Ϫ Similarly, 1 was given for the right scene-based In addition to these six conditions (three different Ϫ omissions and 0.5 for the right-sided, object- speeds in two directions), there was a control con- centered neglect. A final score was obtained by dition that was identical to the moving conditions, summing scores of the two drawings, yielding except that the vertical stripes were absent and Ϫ ϩ scores that ranged from 7to 7. Thus, when were replaced by a gray background. Therefore, summing the scores of these tests, the maximum participants performed line bisections in a total of possible score was ϩ57, and the minimum possible seven conditions. score was Ϫ57. Therefore, left hemispatial neglect When the MB is smaller than the stationary was defined as a total score that exceeds 2 SD of object, IM does not occur. Thus, in the second controls’ total score. According to these criteria, background condition (small-background condi- nine patients were diagnosed as having left hemis- tion), the width of the background was made patial neglect. Controls and patients did not differ smaller than that of the horizontal line. As pre- in sex (percentage of men: controls, 45%; patients, sented in Figure 1B, the size of the MB was reduced 78%; exact test, P ϭ 0.132) or age (controls: 66.4 Ϯ to 32.5 ϫ 24.1 cm (44.22 ϫ 33.53 degrees)—80% 5.3; patients: 63.7 Ϯ 9.0; t ϭ 1.06, P ϭ 0.300). of the size used in the large-background condi- Participants’ demographic data, the results of the neurological examinations, and the neglect assess- tion—but the width of the vertical stripes was ments are presented in Table 1. unchanged. The horizontal line was increased in length to 38.5 cm (51.40 degrees)—approximately Lesion Analysis three times as long as the line used in the large- All patients were imaged with either computed background condition, but with same thickness. MB tomography (4/9) or magnetic resonance imaging speed conditions were the same as in the large-back- (5/9). The lesions identified on axial scans were ground condition. There was also a control condition traced on the best-fitting template provided by in which the background was replaced by a gray Damasio and Damasio.20 A neurologist who was background. Therefore, participants performed line unaware of patients’ clinical information coded le- bisections in a total of seven conditions (three speeds sion site as involving the following areas: frontal, in two directions, and a control condition). temporal, parietal, occipital lobes, thalamus, basal The horizontal line on the computer screen ganglia (caudate nucleus, putamen, globus palli- was presented at the level of the participants’ eyes, dus), corpus callosum, internal capsule (anterior with a viewing distance of 40 cm, and the midline limb, posterior limb), and corona radiata. The local- of the screen was aligned with the midsagittal ization of lesions is presented in Table 1, and the plane of each subject’s head and body. Most Korean illustrations of these lesions are presented in Figure 1. elderly participants are not accustomed to using a computer mouse, so participants bisected the line Experimental Apparatus on the computer screen with a pen held in their Stimuli were programmed by Microsoft Visual right hand, and then the examiner clicked on the Cϩϩ (version 6.0) and Direct X (version 6.0) and mark after aligning the cursor with the mark. were presented on a 19-inch (40 ϫ 30 cm: visual Clicking the computer mouse was immediately fol- angle 53.13 ϫ 41.11 degrees) computer monitor. lowed by the next trial, and LBEs were automati- IM is elicited when a stationary object is sur- cally computed. Negative values represent leftward

July 2007 Line Bisection on Moving Background 517 518 hie al. et Choi

TABLE 1 Results of neurological examination, neglect assessment, and lesion analysis in patients with hemispatial neglect participated in experiment 1 (n ϭ 9) and experiment 2 (n ϭ 3) Neglect Tests

Subject No. Days since Neurological Solid-Line Star-Line Letter-Line Line Star Two Modiﬁed Sex/Age Onset Examination Bisection, mma Bisection, mma Bisection, mma Cancellationb Cancellationb Daisyb Ogdenb Lesion Sites

1. M/69 9 M, S 12.0 18.2 21.5 18/18 27/25 ϪϩP, T, CR 2. M/67 28 M, S, N, G 14.7 31.4 16.5 0/17 2/16 ϪϪF, P, T, insula 3. F/51 183 M, S Ϫ14.0 8.3 23.7 18/18 24/26 ϩϩF, P, T, insula 4. F/73 64 M, S, N, G 38.6 67.1 78.7 0/17 0/15 ϩϩF, P, T, CR, border zone 5. M/67 68 M 4.9 21.5 13.8 18/16 27/27 ϩϪF, BG (CN), CR, border zone 6. M/65 107 M 7.1 12.7 16.0 18/18 27/27 ϪϪBG (CN), CR 7. M/70 26 M, S 5.8 53.9 68.2 18/18 1/25 ϪϪF, T, insula, CR 8. M/65 1 M, V, N, G 103.3 111.7 113.9 0/7 0/6 ϩϩF, P, T, BG (CN, (Exp 2-1) Pu, GP), IC, m .Py.Md Rehabil. Med. Phys. J. Am. insula 9. M/46 6 M Ϫ6.0 17.1 39.6 18/18 25/27 ϪϪBG (Pu), insula Exp2-2 M/82 22 M, S, V 35.1 23.1 23.1 18/15 12/23 ϩϩO, CR border zone Exp2-3 M/70 4 M, V, G 74.7 109.5 112.2 0/13 0/7 ϩϩF, T, O, Th, splenium of CC M, motor weakness; S, sensory impairment; N, neck deviation; G, gaze preference; V, visual ﬁeld defect; F, frontal; T, temporal; P, parietal; O, occipital; Th, thalamus; BG, basal ganglia; CN, caudate nucleus; Pu, putamen; GP, globus pallidus; CC, corpus callosum; IC, internal capsule; CR, corona radiata. a Lines were 242 mm long. Negative values indicate leftward errors from the true midpoint, and positive values rightward errors. b The maximum possible score is 18/18 for line cancellation and 27/27 for star cancellation. Two daisy and modiﬁed Ogden were rated as presence (ϩ) or absence (Ϫ) of left neglect. ● o.8,N.7 No. 86, Vol. FIGURE 1 An illustration for lesion sites in patients with neglect.

deviation from the true midpoint, and positive val- gray background). The total of 28 trials was divided ues indicate rightward errors. into three blocks. There was a 2-min break between the second and third blocks. The first block consisted Experimental Line-Bisection Task of four control trials. The second and the third blocks The mean interval between the neglect tests each consisted of 12 trials of the three leftward or and the experimental task was 2.6 Ϯ 5.8 days. In three rightward movement conditions (S1, S2, and both the large- and the small-background condi- S3). The first block always preceded the second and tions, participants performed four bisection trials third blocks, but the second and the third blocks in each of the seven conditions (three different alternated in sequence so that half of the participants background speeds, two different directions, one performed the leftward MB block first, and the other

FIGURE 2 Stimuli used in experiment 1. A, Large-background condition. B, Small-background condition.

July 2007 Line Bisection on Moving Background 519 half performed the rightward MB block first. In the ward moving speed condition were as follows: second and the third blocks, the sequence of 12 trials Ϫ0.65 Ϯ 1.88 mm (1.05%) for S1, 1.63 Ϯ 2.63 mm was randomized. There were four practice trials be- (2.61%) for S2, and 1.31 Ϯ 3.28 mm (2.09%) for fore the experiment. Half the participants performed S3. The LBEs for the different rightward MB speed the large-background condition first, and the other conditions were as follows: Ϫ1.08 Ϯ 3.93 mm half of the participants performed the small-back- (Ϫ1.73%) for S1, Ϫ1.68 Ϯ 3.21 mm (Ϫ2.69%) for ground condition first. It took about 15 mins for each S2, and Ϫ0.62 Ϯ 4.88 mm (Ϫ0.98%) for S3. To test subject to complete experiment 1. whether these LBEs differed from those of the control conditions, t tests with Bonferroni correc- RESULTS tion were performed. The results show that the LBEs Line-Bisection Performances of the Healthy at S1, S2, and S3 background speeds with the left- Elderly ward MB condition deviated significantly to the right with respect to the control condition (S1: t ϭ 2.83, Large-Background Condition P ϭ 0.030; S2: t ϭ 3.35, P ϭ 0.009; S3: t ϭ 2.63, P ϭ In the large-background condition, the mean 0.047), but the LBEs at S1, S2, and S3 background LBE in the control condition deviated slightly to speeds with the rightward MB did not differ signifi- ϭϪ the left (Ϫ0.88 Ϯ 1.68 mm, Ϫ1.40%), which dif- cantly from the control condition (S1: t 0.26, Ͼ ϭϪ ϭ ϭϪ fered significantly from the veridical midpoint (t ϭ P 0.999; S2: t 1.37, P 0.556; S3: t 0.30, Ͼ Ϫ2.45, P ϭ 0.023). In the large-MB condition, the P 0.999). mean LBE (1.20 Ϯ 2.26 mm, 1.92%) in the left- For both the rightward and leftward conditions, ward MB conditions (all speed conditions averaged) the MB speed had no significant effect on line-bisec- ϭ ϭ deviated significantly to the right from that of the tion performance (leftward: F 1.84, P 0.171; ϭ ϭ control condition (t ϭ 3.18, P ϭ 0.004). The mean rightward: F 1.80, P 0.178). LBE (–1.13 Ϯ 3.76 mm, –1.80%) in the rightward MB conditions did not deviate significantly from Small-Background Condition the control condition (t ϭϪ0.36, P ϭ 0.723). In the small-background condition, the mean With regard to the specific speed conditions, as LBE for the control condition was 0.68 Ϯ 4.97 mm presented in Figure 3A, mean LBEs in each left- (0.35%), which did not differ significantly from the

FIGURE 3 Results of line-bisection errors (LBE) in controls (A) and in patients with neglect (B). LBEs in conditions with large (left column) and small (right column) backgrounds moving at various speeds (S1: 1.26 cm/sec, 1.81 degrees/sec; S2, 11.07 cm/sec, 15.76 degrees/sec; S3, 20.87 cm/sec, 29.25 degrees/sec). The x-axis indicates background speed, the y-axis indicates LBEs translated into percentage [(deviation from the true midpoint/half of total line length) ϫ 100], and the vertical bars represent SEs. Squares indicate leftward background conditions, and triangles indicate rightward conditions.

520 Choi et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 true midpoint (t ϭ 0.65, P ϭ 0.526). In the (Ϫ14.81%) for S3 in the leftward conditions; and small-MB condition, LBEs deviated to the same Ϫ19.48 Ϯ 14.89 mm (Ϫ31.17%) for S1, 27.11 Ϯ direction of the MB (Fig. 3A, right column). That 15.97 mm (43.37%) for S2, and 27.78 Ϯ 18.62 mm is, the mean LBE (Ϫ11.81 Ϯ 11.44 mm, Ϫ6.13%) (44.44%) for S3 in the rightward conditions. The in the leftward MB conditions (all speed conditions results of t tests with Bonferroni correction showed averaged) deviated significantly to the left from that LBEs at rightward S1, S2, and S3 deviated that of the control condition (t ϭϪ4.64, P ϭ significantly rightward from those of the control 0.000). Likewise, the mean LBE (17.33 Ϯ 14.76 condition (S1: t ϭ 3.79, P ϭ 0.016; S2: t ϭ 4.57, mm, 9.00%) in the rightward MB conditions deviated P ϭ 0.005; S3: t ϭ 3.68, P ϭ 0.019), but the LBEs significantly to the right with respect to the control at leftward S1, S2, and S3 did not differ signifi- condition (t ϭ 5.45, P ϭ 0.000). The mean LBEs in cantly from those of the control condition, except each speed condition were as follows: Ϫ5.73 Ϯ 5.61 for S3 (S1: t ϭϪ1.66, P ϭ 0.406; S2: t ϭϪ2.71, mm (Ϫ2.98%) for leftward S1, Ϫ12.80 Ϯ 13.60 mm P ϭ 0.080; S3: t ϭϪ3.11, P ϭ 0.043). When the (Ϫ6.65%) for leftward S2, and Ϫ16.89 Ϯ 17.37 mm leftward and rightward conditions were analyzed (Ϫ8.77%) for leftward S3; and 10.98 Ϯ 9.25 separately to find whether the speed affected LBEs, mm (5.71%) for rightward S1, 19.04 Ϯ 16.92 mm the speed effect was significant in the leftward (9.89%) for rightward S2, and 21.96 Ϯ 20.36 mm condition (F ϭ 5.41, P ϭ 0.016); LBEs increased (11.41%) for rightward S3. The results of t tests with significantly from S1 to S2 and from S2 to S3 (F ϭ Bonferroni correction show that LBEs of all MB con- 6.54, P ϭ 0.034). In the rightward condition, the ditions differed significantly from those of the control speed effect was marginally significant (F ϭ 3.45, condition, and all were in the direction of the MB P ϭ 0.057); LBEs tended to increase from S1 to S2 with respect to the control condition (leftward, S1: and from S2 and S3 (F ϭ 3.76, P ϭ 0.088) (Fig. t ϭϪ4.60, P ϭ 0.000; S2: t ϭϪ4.28, P ϭ 0.001; S3: 3B). t ϭϪ4.46, P ϭ 0.001; rightward, S1: t ϭ 5.17, P ϭ 0.000; S2: t ϭ 5.20, P ϭ 0.000; S3: t ϭ 5.07, P ϭ Small-Background Condition 0.000). When leftward and rightward conditions were In the small-background condition, the mean analyzed separately to find whether the speed affected LBE of the control condition was 27.59 Ϯ 49.40 LBEs, the speed effect was significant for both left- mm (14.33%), which was not significantly different ward and rightward MB conditions (leftward: F ϭ from the vertical midpoint (t ϭ 1.68, P ϭ 0.132). In 11.19, P ϭ 0.000; rightward: F ϭ 10.67, P ϭ 0.000); the MB condition, as in the large-background con- the absolute magnitude of LBEs increased from S1 to dition, LBEs occurred in the same direction of the S2 and from S2 to S3 (leftward: F ϭ 12.42, P ϭ 0.002; MB. That is, the mean LBE of the leftward condi- rightward: F ϭ 11.20, P ϭ 0.003) (Fig. 3A). tion (Ϫ8.08 Ϯ 60.22 mm, Ϫ4.20%) deviated sig- In summary, healthy participants had LBEs in nificantly leftward from that of the control condi- the opposite direction from the MB in the large- tion (t ϭϪ3.90, P ϭ 0.005). Likewise, the mean background condition but in the same direction as LBE of the rightward condition (64.06 Ϯ 47.40 the MB in the small-background condition. mm, 33.28%) deviated significantly to the right with respect to the control condition (t ϭ 3.88, P ϭ Line-Bisection Performances of Patients 0.005). Mean LBEs in each speed condition were as with Hemispatial Neglect follows: 7.55 Ϯ 55.13 mm (3.92%) for S1, Ϫ8.38 Ϯ 61.01 mm (Ϫ4.35%) for S2, and Ϫ23.42 Ϯ 69.72 mm Large-Background Condition (Ϫ12.17%) for S3 in the leftward conditions; and In the large-background condition, the mean 49.46 Ϯ 50.71 mm (25.69%) for S1, 72.83 Ϯ 48.80 LBE of the control condition was 7.12 Ϯ 8.00 mm mm (37.83%) for S2, and 69.91 Ϯ 47.20 mm (11.4%), which deviated significantly to the right (36.32%) for S3 in the rightward conditions. The from the vertical midpoint (t ϭ 2.67, P ϭ 0.028). In results of t tests with Bonferroni correction showed the MB condition, the mean LBE of the leftward that the leftward S2 and S3 conditions (S2: t ϭ condition (all speed conditions averaged) was Ϫ4.44, P ϭ 0.006; S3: t ϭϪ3.53, P ϭ 0.023) and Ϫ4.96 Ϯ 18.58 mm (Ϫ7.94%), which deviated sig- the rightward S2 and S3 conditions (S2: t ϭ 3.80, nificantly to the left with respect to the control P ϭ 0.016; S3: t ϭ 4.19, P ϭ 0.009) differed condition (t ϭϪ2.71, P ϭ 0.027). Likewise, the significantly from the LBEs of the control condi- mean LBE of the rightward condition (24.79 Ϯ tion. In the remaining conditions, deviations did 15.41 mm, 39.66%) deviated significantly to the not differ from those of the control condition. right with respect to the control condition (t ϭ When the leftward and rightward conditions were 4.38, P ϭ 0.002). As presented in Figure 3B, mean analyzed separately to find whether speed affected LBEs in each speed condition were as follows: 0.16 Ϯ LBEs, the speed effect was significant in the left- 18.68 mm (0.26%) for S1, Ϫ5.80 Ϯ 19.76 mm ward condition (F ϭ 5.81, P ϭ 0.013); LBEs in- (Ϫ9.27%) for S2, and Ϫ9.25 Ϯ 19.29 mm creased significantly from S1 to S2 and from S2 to

July 2007 Line Bisection on Moving Background 521 S3 (F ϭ 8.64, P ϭ 0.019). In the rightward condi- the postulates that a large background induces IM tion, the speed effect also was significant (F ϭ 6.63, and that IM modulates healthy participants’ atten- P ϭ 0.008); LBEs increased from S1 to S2 and tion, we requested that these same elderly partici- decreased slightly from S2 to S3 (F ϭ 7.69, P ϭ pants bisect lines in the small-MB condition where 0.024) (Fig. 3B). IM was absent; as expected, in this MB condition, In summary, patients with neglect showed that the healthy elderly showed biases in the same di- LBEs deviated in the same direction as the back- rection of MB. ground in both the large- and small-MB condi- In the large-background condition, although tions. the mean biases of the healthy elderly occurred in the opposite direction of the MB, it was only on the Accuracy of Line Bisection in Patients with leftward MB condition where the bisection biases Hemispatial Neglect significantly differed from the biases in the control So far, we have investigated the directions in conditions. This asymmetry might suggest that the which patients’ bisection biases occurred according leftward MBs elicit greater IM than the rightward MBs. This account accords with the results of to the size and direction of the MB. However, from 17 the treatment perspective, the accuracy of line bi- Jeong et al.’s study, which investigated healthy section rather than the direction of bias should be participants’ overt attention using an infrared eye considered. Thus, with the absolute value of LBEs, tracker while participants were watching a station- we analyzed how close the LBEs were to the verti- ary line on an MB. This study also has revealed that cal midpoint. The data from the three different overt attention was distributed in the opposite di- speed conditions were averaged. rection of MB, and more so with leftward than with rightward MB. The underlying cause of this asym- Large-Background Condition metry is unknown, but it might be related to a right-hemispheric attentional dominance such The mean absolute LBEs of the control, left- that a leftward MB has a greater influence than a ward, and rightward conditions were 7.78 Ϯ 7.28, rightward MB, because with leftward movement, 12.34 Ϯ 14.20, and 24.79 Ϯ 15.41 mm, respec- the MB is more strongly attended to than with tively. The absolute LBEs for the control condition rightward MB. Against this hypothesis is the obser- differed significantly from zero (t ϭ 3.21, P ϭ vation that in the small-MB condition, where sub- 0.013). The absolute LBEs of the leftward MB did jects deviated in the direction of the MB, normal not differ from those of the control condition (t ϭ subjects did not deviate more to the left than to the 1.01, P ϭ 0.340), whereas the absolute LBEs of the right. Alternatively, this asymmetry might be re- rightward MB were greater than those of the con- lated to an asymmetry of IM such that IM is more trol condition (t ϭ 4.51, P ϭ 0.002). These results easily developed going from left to right than vice suggest that the leftward MB did not improve the versa, and this asymmetry might be related to the accuracy of line bisection, but the rightward MB left-to-right scanning habits (i.e., reading habit) of aggravated the accuracy of line bisection. participants.22 In contrast to controls, patients with neglect Small-Background Condition showed biases in the same direction of background The mean absolute LBEs of the control, left- motion regardless of whether the background was Ϯ ward, and rightward conditions were 29.31 larger or smaller than the stimulus. These results Ϯ Ϯ 48.28, 38.16 45.41, and 64.06 47.40 mm, suggest either that neglect patients do not develop respectively. The absolute LBEs of the control con- IM in the large-background condition, or that ϭ ϭ dition did not differ from zero (t 1.82, P these patients are not influenced by IM. Patients 0.106). As in the results of the large-background with neglect might be unable to perceive IM be- condition, the absolute LBEs of the leftward MB cause the brain regions that mediate motion per- did not differ from those of the control condition ception (e.g., MT area) have been injured. However, ϭ ϭ (t 0.78, P 0.460), whereas the absolute LBEs of most patients with neglect in our study were not the rightward MB were greater than those of the injured in MT. Furthermore, functional magnetic ϭ ϭ control condition (t 3.83, P 0.005). resonance imaging studies have shown that brain regions that mediate real motion and induced mo- DISCUSSION tion are the same.21 In addition, the fact that ne- In the healthy elderly, large-background con- glect patients’ attention was strongly affected by ditions induced biases in the direction opposite the the background motion also argues against this background motion, suggesting that MBs induced hypothesis. IM effect is maximized when the par- IM in these participants. These results replicate the ticipants selectively foveate the line. Thus, the dif- results of Na et al.15 and Choi et al.16 involving ferential effect of MB on healthy participants as young participants. To provide further support for opposed to neglect patients might have been asso-

522 Choi et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 ciated with the fact that neglect patients have a deﬁcits, who could not cooperate for the eye re- limited attention capacity and a decreased ability to cording, or who refused to give consent. The par- selectively attend.23 Unlike healthy participants, ticipants demonstrated binocular visual acuity who may be able to selectively attend the line, the above 20/30, either uncorrected or corrected. All patients with neglect might allocate most of their participants were right-handed and provided the attention to the MB. informed consent approved by the institutional re- It may be argued that the difference between view board. the large- and small-background conditions might The patients underwent the same neglect tests have been related to the difference in line length. described in experiment 1. The time interval from Previous studies24–26 have shown that in normal the onset of stroke to the neglect assessment individuals or in patients with neglect, LBEs on ranged from 1 to 26 days (mean 11.0 Ϯ 13.2). longer lines (about 5–27 cm) can occur in the Demographics and results of neurologic and ne- opposite direction from those of shorter lines glect assessment are presented in Table 1. (0.5–5 cm)—a paradoxical phenomenon called the crossover effect. However, the line lengths (12.5 or Apparatus for Line Bisection 38.5 cm) used in our study were within the range Stimuli used in experiment 2 were the same as of longer lines from previous studies. Thus, it is the stimuli used in experiment 1, but only the large less likely that the difference in line length explains background moving at 5.00 cm/sec (5.72 degrees/ the difference between the large- and small-back- sec) was used. The stimuli were presented on a ground conditions. 17-inch (34 ϫ 27 cm, visual angle 37.56 ϫ 30.43 degrees) computer monitor with a viewing distance EXPERIMENT 2 of 50 cm. The actual midpoint of each line was aligned with the monitor’s center and the midsag- Experiment 2 examined eye-movement pat- ittal plane of the viewer’s body. terns during a line-bisection task in the healthy elderly vs. patients with neglect. Line-Bisection Procedure and To help better understand the mechanisms Eye-Movement Recording that might account for the discrepancy between healthy participants and neglect patients, we per- As in experiment 1, there were three condi- formed experiment 2, in which overt eye move- tions: control, leftward, and rightward MB. One ment of the participants was recorded using an trial was conducted for each condition. As illus- infrared eye tracker. In general, people move their trated in Figure 4, the control trial preceded the eyes in the direction to which they are attending. If MB trials, and the sequence of rightward and left- patients with neglect cannot sustain attention and ﬁxate on the stimulus lines, then, compared with healthy participants, their gaze will be more frequently directed to the background rather than the stationary line.

MATERIALS AND METHODS Participants Participants consisted of three (two men and one woman; mean age, 61.3 Ϯ 11.2 yrs) healthy elderly participants and three patients (two men and one woman; mean age, 72.3 Ϯ 8.7 yrs) with left hemispatial neglect from right-hemispheric infarctions. One of the patients participated in experiment 2 as well as experiment 1 (patient 8 in experiment 1, Table 1). Healthy participants and patients did not differ in age (Mann-Whitney test, P ϭ 0.275). The healthy participants were spouses of neurology outpatients at Samsung Medical Center, Seoul, Korea and had no history of neurologic and FIGURE 4 Stimuli used for eye-movement recording psychiatric illness. The three patients with left ne- during line bisection (experiment 2). Be- glect were consecutively recruited during a 3-mo tween each stimulus, there was a pause of period from neurology inpatients at Samsung Med- about 10 secs, during which subjects were ical Center, after excluding patients who had de- asked to close their eyes. Total experi- creased levels of consciousness or severe cognitive ment time was about 125 secs.

July 2007 Line Bisection on Moving Background 523 TABLE 2 Mean values of vertical eye-movement ranges during line-bisection performances (experiment 2) Time

Group Background Conditions Total First 7.5 secs Second 7.5 secs

Controls No background 7.62 7.23 1.72 Leftward 7.73 7.39 3.18 Rightward 8.32 7.97 2.22 Patients with neglect No background 25.71 25.71 9.54 Leftward 25.11 23.83 7.06 Rightward 22.94 22.94 7.60

ward trials was counterbalanced. To determine the between pre- and postcalibrations for the control portion of the line that each participant fixated, participants were 1.18 degrees (control 1), 2.66 before the bisection trials, calibrations were con- degrees (control 2), and 4.34 degrees (control 3), ducted (precalibration); after the bisection trials, and the mean was 2.73 degrees. The differences for the calibration procedure was repeated (postcali- the patients were 4.63 degrees (patient 1), 14.08 bration). The calibration point was a red circle, 6 degrees (patient 2), and 1.90 degrees (patient 3), mm in diameter, with a central, 2-mm-diameter and their mean was 5.72 degrees. Because this hole that was presented at four locations on the pre–post test change in calibration exceeded our monitor screen: 5.35 cm rightward and leftward, as operational standard of 1.5 degrees, we cannot pre- well as 5.75 cm above and below the geographic cisely know the parts of the stimulus to which center of the screen. These calibration points were participants were looking. However, as demon- presented one at a time for 5 secs at each location strated in Figure 5 and Table 2, the ranges of in the following order: rightward, leftward, above, vertical eye positions were greater for the patients and below (Fig. 4). than for the controls, suggesting that the patients For eye-movement recording, each subject was were looking at the background more frequently seated on an adjustable chair with his or her head and for a longer time than were the healthy con- and chin stabilized using a frame with a chin rest trols, and, therefore, the patients also viewed the and forehead brace. The pre- and postcalibrations stimulus line less often than did the healthy con- (20 secs per calibration) and the three bisection trols. To learn the variation of vertical position as a trials (15 secs per trial) were performed as a block. function of time, we analyzed the range of vertical There was a 10-sec pause between trials or between position during the first the first 7.5 secs and the the trial and the calibration. Thus, the entire block next 7.5 secs. The results show that patients had took about 125 secs. The participants were in- greater SDs and ranges than controls in both the structed to avoid head movements. Before and after first and second 7.5 secs (Table 2). each bisection trial, the participants were asked to close their eyes while the stimulus was switched to GENERAL DISCUSSION the next stimulus (about 10 secs). In each bisection The results of experiment 2 suggest that com- trial, participants were asked to find the horizontal pared with the patients, healthy controls are better line on opening their eyes, to estimate the mid- able to selectively fixate on the stationary line. The point of this line, and then fixate on the subjective inability of the patients with neglect to maintain midpoint until asked to stop (close their eyes). The fixation on the line might account for these pa- time interval from eye opening to closing in each tients’ inability to perceive the IM. According to trial was 15 secs. During this interval, an infrared Jeong et al.’s17 study, participants’ attention is bi- eye tracker (iViewX Hi-Speed system, SensoMo- ased to the leading part of moving objects regard- toric Instruments, Berlin, Germany) tracked the less of whether the movement is real or induced. horizontal and vertical position of the subject’s left Thus, controls who were able to fixate on the line eye with a spatial resolution of 0.025 degrees and a had the illusion of the line moving in the direction temporal resolution of 240 Hz. opposite the MB, and their bisection biases occurred toward the induced motion of the line, away RESULTS from the direction of the MB. In contrast to the We performed pre- and postcalibration to as- control participants, the biases of the patients with sure that the eye-tracker settings remained un- neglect were primarily influenced by the real back- changed during the bisection trials. The differences ground movement.

524 Choi et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 FIGURE 5 Traces of eye movements during bisection trials were recorded using an infrared eye tracker and were superimposed on the stimulus. A, Performance of controls (C1, C2, and C3). B, Performance of patients with neglect (P1, P2, and P3).

July 2007 Line Bisection on Moving Background 525 Why the patients with neglect were unable to neglect by neck muscle vibration and spatial orientation of maintain fixation on the line is not entirely known. trunk midline. Brain 1993;116:383–96 The observation that these patients had trouble 4. Karnath HO: Subjective body orientation in neglect and the interactive contribution of neck muscle proprioception and fixating the line even when there was no back- vestibular stimulation. Brain 1994;117:1001–12 ground movement would suggest that it was not 5. Karnath HO: Transcutaneous electrical stimulation and vi- the background movement that distracted them. bration of neck muscles in neglect. Exp Brain Res 1995; Patients with neglect have been shown to have 105:321–4 reduced arousal, and they might even be inatten- 6. Vallar G, Rusconi ML, Barozzi S, et al: Improvement of left visuo-spatial hemineglect by left-sided transcutaneous elective to stimuli on the side ipsilateral to their le- trical stimulation. Neuropsychologia 1995;33:73–82 1 sion. People who have reduced arousal–attention 7. Guariglia C, Lippolis G, Pizzamiglio L: Somatosensory are less vigilant, and their inability to focus their stimulation improves imagery disorder in neglect. Cortex attention on the line might be a manifestation of 1998;34:232–42 this arousal–attentional disorder. 8. Pizzamiglio L, Frasca R, Guariglia C, et al: Effect of optokinetic stimulation in patient with visual neglect. Cortex OKS or MB may be less invasive compared 1990;26:535–40 with other vestibular or IM-related methods. It is 9. Mattingley JB, Bradshaw JL, Bradshaw JA: Horizontal visual also easy to administer and cost-effective. However, motion modulates focal attention in left unilateral spatial the OKS can be administered only via visual mo- neglect. J Neurol Neurosurg Psychiatry 1994;57:1228–35 dality; thus, it can be effective only when patients 10. Vallar G, Guariglia C, Magnotti L, et al: Optokinetic stimulation affects both vertical and horizontal deficits of posi- properly fixate on the target or background stim- tion sense in unilateral neglect. Cortex 1995;31:669–83 ulation. Our results in experiment 2 demonstrate 11. Karnath HO: Optokinetic stimulation influences the dis- that patients with neglect might have difficulty turbed perception of body orientation in spatial neglect. properly fixating on the target or even the MB (e.g., J Neurol Neurosurg Psychiatry 1996;60:217–20 fixations occurred far beyond the boundary of the 12. Bisiach E, Pizzamiglio L, Nico D, et al: Beyond unilateral neglect. Brain 1996;119:851–57 MB in patient 2 in Fig. 5). Furthermore, the results 13. Vallar G, Guariglia C, Nico D, et al: Motor deficits and of experiment 1 show that in neglect patients, optokinetic stimulation in patients with left hemineglect. leftward OKS did not improve the accuracy of line Neurology 1997;49:1364–70 bisection and the rightward OKS aggravated it, 14. Nico D: Effectiveness of sensory stimulation on tactile ex- although OKS did affect the direction of biases. tinction. Exp Brain Res 1999;127:75–82 Thus, OKS is a less promising rehabilitation inter- 15. Na DL, Son Y, Kim CH, et al: Effect of background motion on line bisection performance in normal subjects. Cortex vention than other methods. 2002;38:787–96 Experiment 2 has several limitations. The 16. Choi KM, Ku BD, Jeong Y, et al: The influence of illusory presence of vertical neglect was not investigated in motion on line bisection performance in normal subjects. patients with neglect. Thus, the possibility that J Int Neuropsychol Soc 2005;11:881–8 greater variability in ranges of vertical eye posi- 17. Jeong Y, Lee BH, Ahn HJ, et al: Attentional bias induced by viewing actual or illusionary movements. Neurology 2004; tions in patients than in controls was confounded 62:1333–7 by the fact that vertical neglect cannot be com- 18. Oldfield RC: The assessment and analysis of handedness: the pletely excluded. Second, the eye-tracker settings Edinburgh inventory. Neuropsychologia 1971;9:97–113 did not remain constant during bisection trials, 19. Lee BH, Kang SJ, Park JM, et al: The character-line bisec- which failed to precisely demonstrate the parts of tion task: a new test for hemispatial neglect. Neuropsycho- logia 2004;42:1715–24 the stimulus to which participants were looking. 20. Damasio H, Damasio AR: Lesion Analysis in Neuropsychol- Finally, the small number of participants, and our ogy. New York, NY, Oxford University Press, 1989 having recruited only patients with severe neglect, 21. Tootell RB, Reppas JB, Dale AM, et al: Visual motion after- may limit the generalization of data. effect in human cortical area MT revealed by functional magnetic resonance imaging. Nature 1995;375:139–41 ACKNOWLEDGMENT 22. Speedie LJ, Wertman E, Verfaellie M, et al: Reading direction and spatial neglect. Cortex 2002;38:59–67 The authors thank Sue Yeon Lee for her assis- 23. Rapcsak SZ, Verfaellie M, Fleet WS, et al: Selective attention tance in the preparation of the manuscript. in hemispatial neglect. Arch Neurol 1989;46:178–82 24. Marshall JC, Halligan PW: When right goes left: an inves- REFERENCES tigation of line bisection in a case of visual neglect. Cortex 1. Heilman KM, Watson RT, Valenstein E: Neglect and related 1989;25:503–15 disorders, in Heilman KM, Valenstein E (eds): Clinical Neu- 25. Mennemeier M, Vezey E, Lamar M, et al: Crossover is not a ropsychology, ed 4. New York, NY, Oxford University Press, consequence of neglect: a test of the orientation/estimation 2003, pp 296–346 hypothesis. J Int Neuropsychol Soc 2002;8:107–14 2. Rubens AB: Caloric stimulation and unilateral visual ne- 26. McCourt ME, Jewell G: Visuospatial attention in line glect. Neurology 1985;35:1019–24 bisection: stimulus modulation of pseudoneglect. Neuro- 3. Karnath HO, Christ K, Hartje W: Decrease of contralateral psychologia 1999;37:843–55

526 Choi et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 Authors: Laurel J. Buxbaum, PsyD Mary Ferraro, PhD Brain Injury John Whyte, MD, PhD Arthur Gershkoff, MD H. Branch Coslett, MD

Afﬁliations: RESEARCH ARTICLE From the Moss Rehabilitation Research Institute, Philadelphia, Pennsylvania (LJB, MF, JW, HBC); Thomas Jefferson University, Philadelphia, Pennsylvania (LJB, JW, Amantadine Treatment of AG); MossRehab, Philadelphia, Pennsylvania (AG); and University of Hemispatial Neglect Pennsylvania, Philadelphia, Pennsylvania (HBC). A Double-Blind, Placebo-Controlled Study

Correspondence: All correspondence and requests for ABSTRACT reprints should be addressed to Laurel J. Buxbaum, Moss Buxbaum LJ, Ferraro M, Whyte J, Gershkoff A, Coslett HB: Amantadine treat- Rehabilitation Research Institute, ment of hemispatial neglect: a double-blind, placebo-controlled study. Am J Phys Korman 213, 1200 W. Tabor Rd., Med Rehabil 2007;86:527–537. Philadelphia, PA 19141. Objective: The resemblance of some aspects of the hemispatial ne- Disclosures: glect syndrome (hypokinesia, decreased arousal) to aspects of Parkinso- Supported by an award to the ﬁrst nian syndromes, and the success of amantadine in treating disorders of author from the James S. McDonnell attention, prompted a placebo-controlled, double-blind trial of amantadine, Foundation. an inhibitor of the N-methyl d-aspartate (NMDA) glutamate receptor that 0894-9115/07/8607-0527/0 modulates dopamine transmission, in four patients with chronic hemispa- American Journal of Physical tial neglect. Medicine & Rehabilitation Copyright © 2007 by Lippincott Design: Patients received placebo or 100 mg of amantadine twice a Williams & Wilkins day in an ABA design. Dependent measures of drug effect included an

DOI: 10.1097/PHM.0b013e31806e3392 extensive battery of tests assessing arousal, hemiinattention, hemihypoki- nesia, personal neglect, disability, anosognosia, family burden, and naturalistic action. Results: There was no evidence of increased adverse effects with the treatment drug compared with placebo. Of the 17 measures used to assess treatment response in the four patients (68 measures total), linear regressions revealed significant positive treatment effects on very few (four) measures (uncorrected for multiple comparisons), and scattered negative responses to treatment were evident on three measures. The vast majority of measures showed no change in response to treatment. Conclusions: Possible reasons for failure of treatment effects in the present study are discussed. Additional study will be required to determine whether there are neglect patients who may benefit from amantadine.

Key Words: Amantadine, Neglect Treatment, Pharmacology of Hemispatial Neglect, Hemispatial Neglect Treatment, NMDA, Glutamate Antagonists, Parkinsonian Syndromes

July 2007 Amantadine Treatment of Neglect 527 imals, and the DA blocker spiroperidol reverses this Hemispatial neglect is a complex syndrome treatment effect.16,18 characterized by failure to report, orient toward, or These data suggest that DA agonists might respond to stimuli on the contralesional side of effectively treat neglect in humans. Results sup- space that cannot be attributed to primary motor porting this possibility were obtained by Fleet et or sensory dysfunction.1 Lateralized deficits in spatial al.,19 who have shown that 15 mg/day of the DA attention and motor intention/action, hypoarousal, agonist bromocriptine was effective in reducing and unawareness of deficit are also commonly en- neglect in two patients with large right-hemi- countered. Neglect is a relatively frequent conse- sphere strokes. For one subject, neglect worsened quence of lesions to the right hemisphere, occur- on discontinuation of the medication. Similarly, ring in approximately 50% of patients who have Geminiani et al.20 have shown positive effects of suffered right-hemisphere cerebral vascular acci- treatment with 2 mg of apomorphine in four ne- dent,2 and persisting chronically in at least mild glect patients. Methylphenidate (Ritalin), a medi- form in approximately 75% of patients.3 The dis- cation that enhances the activity of several mono- order may occur subsequent to lesions to the infe- amine neurotransmitters, including DA, and that rior parietal lobe, frontal lobe, thalamus, basal gan- has been shown to improve the speed of mental glia (striatum), and/or superior temporal gyrus.4 processing in patients with traumatic brain inju- Neglect has a significant impact on rehabilitation, ry21 and attention-deficit disorder,22 also has been disability, and family burden beyond that predicted used as a treatment for neglect. Hurford et al.23 by lesion size.2,5 assessed the efficacy of 20 mg/day of methylpheni- There are several subtypes of neglect that may date in a patient with chronic neglect caused by occur together or, much more rarely, in isolation. middle cerebral artery stroke. Methylphenidate was For example, neglect may primarily affect the left as effective as 30 mg/day of bromocriptine, and of the body (personal neglect), the left of near or far improvements were maintained after the medica- space (peripersonal or extrapersonal neglect), move- tions were discontinued. Finally, Mukand et al.24 ments in and toward the left side of space (motor have reported that carbidopa/levodopa in doses of intentional neglect), or detection of targets on the 25/100 mg, three times daily, improved neglect in left, irrespective of the type of response required to three of four patients with left neglect secondary to indicate detection (sensory neglect).2,6–13 One of the right-hemisphere stroke. most robust observations about neglect patients is However, less encouraging results have also been that they exhibit hypokinesia (decreased motor re- obtained in a number of studies. One recent study sponsiveness), even for movements in ipsilesional examined the performance of seven right-hemisphere space, along with more pronounced slowing or re- cerebral vascular accident patients who exhibited duction in movements of the head, eyes, limbs, and asymmetric target-detection performance on a com- torso in a contralesional direction (directional hy- puterized target-search paradigm; many of the pa- pometria or hypokinesia).8,10 These disorders of tients exhibited decreased exploration and prolonged movement raise the possibility that dopamine (DA) reaction times in the contralesional hemispace with depletion may play a role in the disorder. In fact, 2.5 mg of bromocriptine.25 This finding is difficult to there is evidence that dopaminergic receptors may interpret, because bromocriptine may inhibit DA re- be asymmetrically lateralized to the right hemi- ceptors at low doses.20,26 However, it is of note that all sphere.14 This suggests that prominent aspects of of the poor responders had lesions that included the the neglect syndrome might result from damage to striatum, and one patient, whose response was posi- predominantly dopaminergic right-hemisphere tive, did not have striatal damage. Barrett et al.27 subcortical and cortical structures that mediate have reported a patient with motor neglect caused both general arousal and attention to the contrale- by right middle cerebral artery stroke, whose per- sional hemispace.15 formance on a line-bisection task declined on 20 There are several lines of evidence that decreased mg/day of bromocriptine. The investigators sug- levels of DA may be implicated in neglect. Data from gest that bromocriptine may have differentially ac- studies with animals indicate that ascending DA sys- tivated the intact dopaminergic system of the left tems projecting to the frontal cortex, neostriatum, hemisphere, with relatively less activation of the and cingulate gyrus are involved in motor intention, damaged dopaminergic system of the right hemi- sensory orienting, and arousal. DA levels also corre- sphere, thereby exacerbating the ipsilesional bias. late with neglect severity in rats.16 Neglect can be They also suggest that the presence of right sub- induced by injections of the dopaminergic neurotoxin cortical involvement may have prevented DA trans- 1-methyl-4-phenyl-1,2,3,6,-tetrahydropyride into the mission to the right parietal spatiomotor system. caudate nucleus of the monkey.17 The DA-receptor In this study, we assessed the efficacy of aman- agonist apomorphine reduces neglect in lesioned an- tadine, a glutamate antagonist having modulatory

528 Buxbaum et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 effects on DA transmission. Selection of this med- cial effects on arousal and attention in the neglect ication was guided in part by theories of the patho- syndrome. physiology of Parkinson disease28,29 (PD) and in To assess the efficacy of amantadine in treating part by evidence that amantadine has been success- neglect, we performed a double-blind, placebo-con- ful in treating disorders of attention in nonfocal trolled exploratory study in which patients were pathologies such as traumatic brain injury and administered a placebo (A1), then amantadine (B), multiple sclerosis. then another placebo (A2). An important aspect of PD may result from loss of DA neurons in the our procedure was a careful attempt to establish a substantia nigra, thus reducing DA levels in the stable baseline of performance during the A1 putamen. There are two populations of DA recep- phase, to ascertain that any changes during the B tors in the putamen, both of which are inhibitory. phase were not just random variability. Addition- The first are D1 receptors, on which DA has an ally, because neglect is a multifactorial syndrome, excitatory effect, and the second are D2 receptors, we used a thorough battery of neglect tests de- on which DA has an inhibitory effect. The opposite signed to permit assessment of possible differential effects of DA depletion on the two putamen sub- treatment effects on motor, sensory, personal, and populations are important in the pathophysiology peripersonal neglect. In addition to using these of PD. D1 receptors become hypoactive and, in various tasks to assess treatment-related changes, turn, provide less inhibition of the globus pallidus we also were interested in assessing how these tests (GP). D2 receptors are overactive; consequently, performed over repeated administrations. Other they induce net excitation of the GP. The GP exces- goals of the trial were to establish the safety of the sively inhibits a portion of the thalamus, reducing medication in this population of patients, and the excitatory output to the supplementary motor area. feasibility of performing additional studies if appro- This loss of excitatory output to the cortex seems to priate. underlie the hypokinetic deficits of PD.28,29 In at least some patients with neglect, striatal METHODS damage that mitigates the positive effects of DA therapy may result from mechanisms that are sim- Subjects ilar to those that occur in PD. A possible treatment Subjects were recruited from a pool of seven strategy involves targeting the subthalamic nu- MossRehab and Bryn Mawr Rehabilitation Hospital cleus, which receives excitatory glutamatergic pro- chronic right-hemisphere stroke patients who had jections from the motor and premotor cortex and, been identified in a previous study2 as having at in turn, has excitatory glutamatergic projections to least moderate neglect, and who expressed interest the GP. To effect a reduction in GP activity (and, in participation. Patients were considered eligible hence, a reduction of thalamic and cortical inhibi- for the present study if they had sustained a right- tion), a possible pharmacologic intervention could hemisphere hemorrhagic, embolic, or thrombotic ce- inhibit excitatory effects of the subthalamic nu- rebrovascular accident (right-hemisphere cerebral cleus on the GP or of the cortex on the subthalamic vascular accident) 6–36 mos previously. Patients with nucleus. Amantadine is a low-potency glutamate history of previous head injury, left-hemispheric antagonist thought to counteract these excitatory stroke, or other neurologic disorder were excluded, effects. Glutamate antagonists such as amantadine as were those suffering from DSM-IV axis I disorder have antiakinetic effects in animal models of PD,30 (e.g., major depression, psychosis, dementia36)at and amantadine has long been used as an anti- the time of the study. Pregnant women were ex- Parkinsonian in humans (see, for example, Thomas cluded. Inclusion was also based on three behav- et al.31). ioral criteria gleaned from review of the medical It is also possible that amantadine may have chart and prior testing data: a) language compre- general or “nonspecific” beneficial effects on atten- hension adequate to understand task instructions; tion and/or cognition. The medication has been b) motor capacity, hearing, and visual function used as a treatment of fatigue in patients with sufficient to perform the testing protocol; c) multiple sclerosis, although the effects have not arousal and behavioral control adequate for a 60- been robust (e.g., reference 32). Similarly, it has min testing session. All subjects gave informed been used in the treatment of traumatic brain consent to participate in the study in accordance injury, albeit with mixed results; Meythaler et al.33 with the guidelines of the institutional review have reported significant recovery in a double- board of Albert Einstein Healthcare Network blind crossover design (but see Giacino et al.34 for (MossRehab) and Lankenau Health System (Bryn a critique), whereas Schneider et al.35 have re- Mawr Rehabilitation Hospital). ported no benefit from amantadine in a traumatic Preliminary testing was conducted to confirm brain injury rehabilitation setting. It remains pos- that neglect was indeed still present. Neglect was sible that amantadine might prove to have benefi- defined by performance below cutoff scores on at

July 2007 Amantadine Treatment of Neglect 529 least one of five standard neglect tests—the bell were all performed at the same time of day for each test,37 and four subtests of the Behavioral Inatten- subject. tion Test (BIT)38: 1) letter cancellation, 2) picture At the start of each testing session, blood pres- scanning, 3) menu reading, and 4) line bisection sure, pulse, daily caffeine intake, daily cigarette (see Test Protocol section, and Buxbaum et al.2, for use, sleep start and end times, and intake of daily test descriptions and scoring procedures). With the study medication were monitored; all participants exception of the line-bisection test, for which ne- were determined to be stable. Participants and glect was determined by BIT guidelines, patients caregivers kept a daily calendar listing the times were considered to exhibit neglect if their left– they took the study drug, as well as their daily sleep right difference scores were greater than 20% of start and end times. When necessary, participants the total number of items on each side of the were called daily to remind them to take morning array.38 There were five patients who met all crite- and evening study medication. ria for participation. One of these patients exhib- The study physician was always aware of the ited mild neglect and proved to benefit substan- participant’s drug status. The tester automatically tially from practice: after three baseline-phase relayed daily blood pressure and pulse readings sessions, he performed at ceiling levels (i.e., at outside a specified baseline range to the study maximum possible performance) on most of the physician. Once a week, a side-effect questionnaire experimental tasks; he was subsequently dropped was administered to all participants and caregivers, from the study. Patient characteristics of the re- asking them to cooperatively rate the presence of maining four patients who met all inclusion crite- 19 symptoms (e.g., drowsiness, headaches, short- ria can be seen in Table 1, and lesioned regions are ness of breath) on a four-point scale (none, mild, shown in Figure 1. moderate, severe). The results of the questionnaire were routinely reported to the study physician as Treatment Protocol well. Adverse effects are reported in Table 2. No Participants were given a comprehensive phys- adverse effects were reported more frequently in ical examination by the study physician (A.G.), in- the amantadine phase than in the placebo phases. cluding blood work to assess renal function. An ABA treatment design was used. After starting on Testing Protocol placebo (taken in unlabeled 100-mg oral syringes Daily Probes twice daily; A1 phase), subjects were titrated to 38 target-dose amantadine (200 mg, taken in unla- 1. Letter cancellation, a subtest of the BIT, re- beled 100-mg oral syringes twice a day; B phase) quired cancellation of the letters E and R and subsequently titrated down to placebo again printed on 8.5 ϫ 11-inch paper. A difference (A2 phase). The titration dosing schedule entailed score was obtained by subtracting the number 100 mg (50 mg, twice a day) for 3 days before the of targets cancelled on the left of the array from target dose and after completion of the B phase. the number cancelled on the right. Performance at each testing session was used to 2. Line bisection, another subtest of the BIT, re- assess the stability of scores (described later); con- quired demarcation of the center of three lines sequently, the number of testing sessions in the presented on a sheet of paper. The mean dis- treatment phase of the study varied across the tance (in millimeters) of responses from the participants, as follows: subject 1: four sessions in 10 days; subject 2: five sessions in 10 days; subject true midpoint was calculated. 3: 10 sessions in 23 days; subject 4: four sessions in 3. Large-letter cancellation (Buxbaum and Coslett, 8 days. Participants were tested approximately unpublished test, 1993) required cancellation of three times per week by a research assistant who 80 Ts among the letters E and C in a large array was blinded to the study phase. Testing sessions (34 ϫ 22 inches). The difference in the number

TABLE 1 Subjects Education, Months Visual Field Age Handedness yrs after CVA Lesion Deﬁcit Hemiparesis

S1 75 Right 20 35 Thal, IC No Yes S2 74 Right 12 31 F, T, P, BG, IC Yes Yes S3 73 Right 20 24 F, T, P, O, Thal Yes Yes S4 78 Right 18 6 F, T, O Yes Yes

530 Buxbaum et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 FIGURE 1 Reconstructed lesions of subject 1 (top row), subject 2 (second row), subject 3 (third row), and subject 4 (bottom row).

of items canceled on left vs. right in 2 mins was ton balls on a blindfolded participant’s left side calculated. at the shoulder, chest, elbow, forearm, wrist, 4. The Bell test37 required cancellation of bells pre- and hip (for a similar test, see Cocchini et al.39). sented in a dense array on 8.5 ϫ 11-inch paper. A On removal of the blindfold, the participant was left–right difference score was calculated. instructed to locate and remove the cotton 5. Personal neglect was assessed with the Fluff balls. The number of detected targets was test, which was administered by placing six cot- tallied (0–6).

TABLE 2 Adverse effects (n ϭ 4) No. of Participants ever Mean Proportion of Weeks Reporting Adverse Effect in Which Adverse Effects Were Adverse Effect on Amantadine/Placebo Reported on Amantadine/Placebo

Dizziness 1/2 1/0.42 Insomnia 1/1 0.5/0.5 Mental confusion 1/2 1/0.6 Drowsiness 1/3 0.25/0.37 Depression 0/2 0/0.19 Anxiety 0/2 0/0.23 Hallucinations 1/1 1/1 Loss of appetite 1/1 1/0.33 Dry mouth 3/3 0.58/0.5 Constipation 1/2 0.75/0.33 Swelling of ankles 0/3 0/0.36 Low blood pressure 0/1 0/0.17 Headaches 2/2 1/0.8 Blurred vision 1/1 0.5/0.2 Skin rash 1/1 0.5/0.17 Chest/arm pain 2/1 0.63/1 Shortness of breath 1/0 0.5/0

July 2007 Amantadine Treatment of Neglect 531 6. The Sustained Attention to Response Test and right response areas. In both tasks, nine (SART40) was use to assess attention. Subjects practice trials were followed by 20 trials each viewed 225 single digits on a computer screen with left, right, and center trials, in randomized during a 4.3-min period. Each digit was presented order. Mean difference in response time for left for 250 msecs, followed by a 900-msec mask. Par- vs. right trials was calculated for each subject. ticipants responded by pressing a mouse key for all 9. Anosognosia questionnaire. At the end of each digits except the number 3, which appeared 25 testing session, a five-question Anosognosia ques- times in a quasi-randomized fashion. The number tionnaire2 was administered. Two questions ad- of correctly withheld responses to the number 3 dressed awareness of sensorimotor impairment was tallied (maximum ϭ 25). (scores reflect deviation of actual from reported 7. The Dual-Task test is a measure of simple re- impairment; range, 0–2), and three questions ad- sponse time with and without a secondary task dressed general awareness of deficit (range, 0–2). load.41 In the baseline condition (DT base), sub- Outcome Measures jects responded with a key press to 64 black 1. The Naturalistic Action Test (NAT)43,44 mea- circles 1 cm in diameter appearing in random- sures everyday action impairment through a ized locations on a computer screen at intervals standardized three-level performance test. Par- of 500–2000 msecs throughout 64 trials. In the ticipants are asked to prepare toast and coffee, dual-task condition (DT dual), subjects per- wrap a present, pack a lunchbox, and fill a formed the DT base task in conjunction with an schoolbag. Performance is scored for key steps oral digit-repetition task conducted at span. Tri- accomplished and errors committed. The pro- als in which patients did not concurrently per- portion of left-sided objects used and/or touched form both tasks were discontinued and rerun. was also tallied. The NAT has excellent scoring Interference associated with the secondary task reliability (median weighted kappa for accom- was measured by subtracting the mean response plishment score ϭ 0.98), internal consistency latencies on DT base from DT dual (DT decre- (Cronbach alpha ϭ 0.79), concurrent criterion ment). Mean response latency to ipsilesional validity with the Functional Independence Mea- (right sided) targets in the baseline condition sure (FIM) measure (r Ϸ 0.5), construct validity was used as a measure of simple sensorimotor (r ϭϪ0.68 with a measure of processing speed/ response time (DT base right). Decrements in arousal, and 0.61 with a measure of visuospatial response to contralesional targets under dual- attention), and predictive validity. A significant task load was measured by subtracting mean relation was found between discharge NAT and a response latencies to left targets in DT base follow-up instrumental activities of daily living from DT dual (L dual Ϫ L base). scale in bivariate analysis (r ϭ 0.58), and in 8. The Lateralized Target and Lateralized Re- multiple-regression analyses. The multiple-re- 42 sponse tasks measured response latencies in gression analyses showed that NAT’s success in two different stimulus/response conditions. In predicting instrumental activities of daily living the lateralized target test of perceptual neglect, was independent of its correlation with FIM subjects viewed three horizontally arrayed, score, age, or the various attention measures. 5-mm dots on a computer monitor. They were NAT outperformed the FIM cognitive score in asked to fixate on the center dot, and the exam- predicting the instrumental activities of daily iner monitored fixation. After 2000 msecs, a living score. It also outperformed each of the downward-pointing arrow replaced one of the attention tests. Finally, the correlation between three dots. Participants responded by pressing a second NAT administered at the same time as the spacebar on the computer keyboard when a the instrumental activities of daily living mea- left or a right arrow was detected. In the later- sure was also strong (r ϭ 0.64), as was the alized response test of motor or intentional ne- correlation of the second NAT and the discharge glect, subjects viewed three horizontally arrayed NAT (0.66).43,44 Participants performed the NAT in dots. They were asked to fixate on the center dot every other testing session. and the examiner-monitored fixation. After 2. The transfers and mobility items from the FIM 2000 msecs, a left-, right- or downward-point- instrument45 were assessed once during each ing arrow replaced the center dot. Subjects re- study phase. Clinicians who had been trained sponded accordingly by pressing a key on a and certified according to Uniform Database partially shielded keyboard with left, central, System procedures rated the items on a scale of

532 Buxbaum et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 FIGURE 2 Performance of the four participants in the baseline, treatment, and posttreatment phases of the study on a combined score from the bell test and letter-cancellation test. Each graph depicts data from a single participant.

1 (total assistance required) to 7 (complete in- sources and Stress) is a 10-point true/false ques- dependence). tionnaire designed to assess stress placed on the 3. The Family Burden Questionnaire (adapted family of the stroke patient (e.g., “_____” is hard from Friedrich et al.’s46 Questionnaire on Re- to live with; “_____” doesn’t do as much as s/he

July 2007 Amantadine Treatment of Neglect 533 TABLE 3 Mean accuracy scores in each treatment phase (percent correct) S1 S2 S3 S4

Measure A1 B A2 A1 B A2 A1 B A2 A1 B

Bell cancellation 0.86 0.80 0.91 0.47 0.51 0.55 0.12 0.12 0.11 0.51 0.59 Letter cancellation 0.93 0.93 0.95 0.82 0.82 0.81 n/a n/a n/a 0.88 0.90 Large-letter cancellation 0.81 0.87 0.92 0.60 0.61 0.68 0.13 0.17 0.19 0.81 0.67 Fluff test 0.72 0.92 0.95 0.50 0.57 0.62 0.17 0.18 0.12 0.72 0.88 SART correct go 0.78 0.76 0.82 0.72 0.75 0.67 0.66 0.75 0.84 0.97 0.95 SART correct no go 0.44 0.49 0.51 0.82 0.30 0.05 0.83 0.68 0.68 0.98 0.96 Lateralized target: left 1.00 1.00 1.00 0.95 0.96 1.00 0.44 0.15 0.29 1.00 1.00 correct response Lateralized response: left 0.99 1.00 1.00 1.00 0.94 0.97 0.72 0.92 0.99 0.94 0.79 correct response Mean line-bisection error, 9.00 16.10 14.50 16.50 14.90 16.90 71.00 72.40 78.70 65.90 65.90 mm

should do”). The Family Burden Questionnaire On the basis of this calculation, the number of was administered to a family member once dur- sessions in each phase varied across subjects, as ing each study phase. follows: baseline (A1): subject 1 ϭ 13 sessions; subject 2 ϭ 8 sessions; subject 3 ϭ 5 sessions; Determination of Performance Stability subject 4 ϭ 10 sessions; treatment (B): subject 1 ϭ Stability in each study phase (A1, B, A2) was 4 sessions, subject 2 ϭ 5 sessions, subject 3 ϭ 10 determined using a combined score from the bell sessions, subject 4 ϭ 4 sessions; posttreatment test and the letter-cancellation task (Fig. 2). The (A2): subject 1 ϭ 4 sessions, subject 2 ϭ 5 sessions, total numbers of items cancelled (of 74) for three subject 3 ϭ 3 sessions. Unfortunately, subject S4 consecutive testing sessions were averaged. A mov- could not be tested in the A2 phase because of a ing average was computed that compared the change in the therapy schedule that precluded con- fourth testing session’s total score with the mean tinued participation. of the three previous sessions. Fourth-session In addition, we conﬁrmed stability in the base- scores greater than the mean of the previous ses- line phase with Young’s C statistic,47,48 an indicator sions were considered unstable. Fourth-session of serial dependency. The C statistic is not reliable scores more than 10 points below the mean were with fewer than eight data time-points; therefore, also considered unstable. The cutoff of 10 percent- only data from subjects 1, 2, and 3 were evaluated. age points was determined from evaluation of pilot Baseline-phase data from all three subjects showed subject performance. Fourth-session scores below signiﬁcant serial dependency (i.e., stability): subject 1: the mean yet within 10 percentage points were z ϭϪ9.9, P Ͻ 0.0001; subject 2: z ϭϪ6.1, P Ͻ considered stable, and the study phase was ended. 0.0001; subject 4: z ϭϪ6.4, P Ͻ 0.0001.

TABLE 4 Mean response latency in each treatment phase, msecs S1 S2 S3 S4

Measure, msecs A1 B A2 A1 B A2 A1 B A2 A1 B

Dual-task baseline 437.3 443.3 406.0 497.3 1487.0 626.3 771.3 590.5 662.7 563.7 445.5 right Left dual Ϫ left base 1639.0 932.8 677.7 3507.0 2703.0 2071.3 2881.3 2287.6 2321.3 1194.3 3085.8 Dual-task baseline 1595.3 974.0 715.0 3818.0 2908.8 2240.7 4243.7 3412.3 3545.7 1016.3 3019.3 decrement Sustained attention 492.3 473.7 428.0 628.0 520.8 501.3 677.3 569.0 538.7 559.3 504.3 to response test Lateralized target 455.5 539.5 441.0 1076.7 1523.6 1109.7 1504.0 1374.4 1009.7 993.0 952.3 left Lateralized response 1530.0 1563.8 1537.0 1893.3 1696.8 1768.7 2929.0 2713.6 2157.7 3203.7 2605.0 left

534 Buxbaum et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 TABLE 5 Mean test scores on Naturalistic Action Test (NAT), Functional Independence Measure (FIM), and questionnaires S1 S2 S3 S4

Measure A1 B A2 A1 B A2 A1 B A2 A1 B

NAT total score (18 possible) 7 8 11 8.3 10 10 1 1 1 8.5 9 NAT left objects touched/used, % 70 72 72 72 56 64 0 0 0 70 55 FIM locomotion ratings (7 possible) 1 1 1 2 2 2 1 1 1 6 6 FIM transfer ratings (7 possible) 3 3 3 5 5 5 2 2 2 7 7 Family Burden questionnaire (no. true/10) 6 6 4 4 4 5 3 3 2 4 4 Anosognosia questionnaire (maximum ϭ 8) 3.3 3 3 3 3 3 1.7 1.8 2 1.5 1.3

RESULTS responses on DT base response time (standard co- Tables 3 and 4 provide the mean performance efficient ϭ 0.81, P ϭ 0.001) and SART no-go cor- in each study phase on probe tests. Table 5 presents rect responses (standard coefficient ϭϪ0.32, P ϭ data from the outcome measures (NAT, Anosogno- 0.01). Subject 3 showed a positive response on sia questionnaire, FIM instrument, and Family SART mean response time (standard coefficient ϭ Burden Questionnaire). Ϫ0.31, P ϭ 0.02) and a negative response on later- Linear regressions were performed with many alized mean response time (standard coefficient ϭ of the probe tests to determine whether amanta- 0.47, P ϭ 0.04). Subject 4 showed a positive response dine improved performance. The measures pre- on lateralized mean response time (standard coeffi- sented in Tables 2 and 3 were used as dependent cient ϭϪ1.3, P ϭ 0.002). variables, with the exception of the bell, letter, and large-letter tasks, for which we employed right–left DISCUSSION difference scores (these difference scores had dem- We performed a double-blind, placebo-con- onstrated sensitivity to neglect in a previous trolled exploratory study of the effects of amanta- study).2 Session number and drug status (off or on dine, a glutamate antagonist, in four patients with amantadine) were used as predictor variables. Vi- moderate to severe chronic neglect. Participants sual inspection of the data revealed large practice were varied in their underlying neuropathology: effects in some subjects; therefore, only the last one had a pure subcortical lesion, two were mixed three placebo sessions were included. Sessions in cortical and subcortical, and one was purely corti- which drug dose was being titrated up or down cal. Despite careful efforts to establish stable per- were not included. The session variable permits formance baselines before entering the drug-treat- assessment of whether linear trends are present in ment phase of the study, there were no participants the data (e.g., practice effects resulting in increas- for whom amantadine had an unambiguously pos- ingly better performance with each session), itive result, and the few positive responses on in- whereas the drug variable, if significant, indicates dividual tests were quite modest. There were also that on-amantadine sessions differ significantly scattered negative responses. Encouragingly, most from off-amantadine sessions. The direction of any adverse effects were reported infrequently, with significant effects (positive or negative) must then comparable frequencies between drug and placebo be ascertained by examination of the data.* conditions. Thus, additional studies of amantadine With the threshold for significance set at an in this population seem both safe and feasible. arguably liberal P value of Ͻ0.05 (i.e., uncorrected There are several possible reasons for the fail- for multiple comparisons), subject 1 showed nei- ure of amantadine to ameliorate neglect in the ther negative nor positive effects on any measure. present group of patients. First, it might be argued Subject 2 showed a significant positive response on that lesions were too extensive in the present SART mean response time (standard coefficient for group, affecting numerous brain regions aside drug effect ϭϪ0.43, P ϭ 0.04) and on lateralized from the putatively critical striatum. This is indeed response percent correct (standard coefficient for true of subjects 2 and 3; however, subjects 1 and 4 drug effect ϭϪ0.61, P ϭ 0.04), and negative had relatively small lesions. Second, it may be the case that the dosage of amantadine was inappro- priate, despite our adherence to medication guide- *Before performing regression analyses, the Durbin– Watson statistic, a measure of autocorrelation, was de- lines. Nevertheless, additional testing with a differ- termined for each measure. None of the measures ent (likely, larger) dose of medication would be showed significant autocorrelation. required to address this possibility. Third, it may be

July 2007 Amantadine Treatment of Neglect 535 that amantadine would function effectively as an 2. Buxbaum LJ, Ferraro M, Veramonti T, et al: Hemispatial adjuvant to other behavioral treatment. neglect: subtypes, neuroanatomy, and disability. Neurology 2004;62:749–56 Given the serious effect of neglect on patient 3. Farne A, Buxbaum LJ, Ferraro M, et al: Patterns of sponta- and caregiver functioning and quality of life, con- neous recovery of neglect and associated disorders in acute siderable efforts have been devoted to exploring right brain-damaged patients. J Neurol Neurosurg Psychi- potential therapeutic and treatment approaches to atry 2004;75:1401–10 the disorder. Pierce and Buxbaum49 provide a de- 4. Vallar G, Perani D: The anatomy of unilateral neglect after right-hemisphere stroke lesions. A clinical/CT scan corre- tailed review of treatment approaches, which in- lation study in man. Neuropsychologia 1986;24:609–22 clude treatments targeting arousal, phasic alerting 5. Katz N, Hartman-Maeir A, Ring H, Soroker N: Functional treatments, hemispheric activation approaches, disability and rehabilitation outcome in right hemisphere constraint-induced therapy, mental imagery train- damaged patients with and without unilateral spatial neglect. Arch Phys Med Rehabil 1999;80:379–84 ing, prism treatment, eye patching, caloric stimu- 6. Adair JC, Williamson DJ, Jacobs DH, Na DL, Heilman KM: lation, optokinetic stimulation, neck-vibration Neglect of radial and vertical space: importance of the therapy, and trunk-rotation therapy. With the pos- retinotopic reference frame. J Neurol Neurosurg Psychiatry sible exception of prism treatment, which shows 1995;58:724–8 promise,50,51 many of these treatment approaches 7. Adair JC, Na DL, Schwartz RL, Heilman KM: Analysis of primary and secondary influences on spatial neglect. Brain have had equivocal, contradictory, or quickly fad- Cogn 1998;37:351–67 ing effects. Further study will be required to deter- 8. Bisiach E, Geminiani G, Berti A, Rusconi ML: Perceptual mine whether amantadine falls into this category and premotor factors of unilateral neglect. Neurology 1990; as well. At the least, the failure of amantadine to 40:1278–81 improve performance in any of the current patients 9. Bisiach E, Perani D, Vallar G, Berti A: Unilateral neglect: personal and extrapersonal. Neuropsychologia 1986;24: indicates that any potential benefit is not universal. 759–67 An unintended but important product of this 10. Coslett HB, Bowers D, Fitzpatrick EF, Haws B, Heilman study was a striking illustration of the extreme KM: Directional hypokinesia and hemispatial inattention in day-to-day variability in baseline (off treatment) neglect. Brain 1990;113:475–86 performance that may be observed in neglect pa- 11. Guariglia C, Antonucci G: Personal and extrapersonal space: 52,53 a case of neglect dissociation. Neuropsychologia 1992;30: tients. It is well appreciated that nonlateralized 1001–10 deficits in attention and arousal are a prominent 12. Vallar G: Extrapersonal visual unilateral spatial neglect and component of the neglect syndrome.54–57 It is also its neuroanatomy. Neuroimage 2001;14:S52–8 likely that levels of arousal may fluctuate in neglect 13. Vallar G: Spatial frames of reference and somatosensory in response to a number of internal and external processing: a neuropsychological perspective. Philos Trans R Soc Lond B Biol Sci 1997;352:1401–9 variables. For example, performance on neglect 14. Larisch R, Meyer W, Klimke A, Kehren F, Vosberg H, HW tasks is improved by the presentation of simple M-G: Left-right asymmetry of striatal D2 receptors. Nucl alerting tones, and by self- or other instructions to Med Commun 1998;19:781–7 “wake up.”58 In the present study, it is of interest 15. Nadeau SE, Crosson B: Subcortical aphasia. Brain Lang that day-to-day fluctuations seemed to decrease 1997;58:355–402 with practice (see Fig. 2), suggesting that atten- 16. Marshall J, Gotthelf T: Sensory inattention in rats with 6-hydroxydopamine-induced degeneration of ascending do- tional variability may be modulated by the automa- paminergic neurons: apomorphine-induced reversal of def- ticity with which tasks can be performed. The sen- icits. Exp Neurol 1979;65:398–411 sitivity of neglect to general attentional factors may 17. Miyashita N, Hikosaka O, Kato M: Visual hemineglect in- obscure subtle treatment effects, or, conversely, duced by unilateral striatal dopamine deficiency in mon- keys. Neuroreport 1995;6:1257–60 they may inflate the apparent effects of a given 18. Corwin J, Kanter SL, Watson RT, Heilman KM, Valenstein treatment. In addition, not all of the tests fluctu- E, Hashimoto A: Apomorphine has a therapeutic effect on ated together (i.e., improvement on one task was neglect produced by unilateral dorsomedial prefrontal cor- sometimes accompanied by worsening on an- tex lesions in rats. Exp Neurol 1986;94:683–9 other), suggesting that factors other than atten- 19. Fleet W, Valenstein E, Watson RT, Heilman KM: Dopamine agonist therapy for neglect in humans. Neurology 1987;37: tional variations affected performance. The present 1765–70 results thus underscore the need for caution in 20. Geminiani G, Bottini G, Sterzi R: Dopaminergic stimula- interpreting positive (or negative) treatment re- tion in unilateral neglect. J Neurol Neurosurg Psychiatry sults when studies do not employ an ABA design 1998;65:344–7 with careful baseline assessment. The data also 21. Whyte J, Hart T, Schuster K, Fleming M, Polansky M, Coslett H: Effects of methylphenidate on attentional func- suggest that when conducting medication-treat- tion after traumatic brain injury: a randomized, placebo- ment studies with neglect patients, a lengthy time controlled trial. Am J Phys Med Rehabil 1997;76:440–50 frame spanning numerous weeks may be necessary. 22. Mehta M, Calloway P, Sahakian B: Amelioration of specific working memory deficits by methylphenidate in a case of adult attention deficit/hyperactivity disorder. J Psychophar- REFERENCES macol 2000;14:299–302 1. Heilman K, Van Den Abell T: Right hemisphere dominance for 23. Hurford P, Stringer AY, Jann B: Neuropharmacologic treat- mediating cerebral activation. Neuropsychologia 1979;17:315–21 ment of hemineglect: a case report comparing bromocrip-

536 Buxbaum et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 tine and methylphenidate. Arch Phys Med Rehabil 1998;79: “Oops!” Performance correlates of everyday attentional fail- 346–9 ures in traumatic brain injured and normal subjects. Neu- 24. Mukand JA, Guilmette TJ, Allen DG, et al: Dopaminergic ropsychologia 1997;35:747–58 therapy with carbidopa L-dopa for left neglect after stroke: 41. McDowell S, Whyte J, D’Esposito M: Working memory im- a case series. Arch Phys Med Rehabil 2001;82:1279–82 pairments in traumatic brain injury: evidence from a dual- 25. Grujic Z, Mapstone M, Gitelman DR, et al: Dopamine ago- task paradigm. Neuropsychologia 1997;35:1341–53 nists reorient visual exploration away from the neglected 42. Buxbaum LJ, Permaul P: Hand-centered attentional and hemispace. Neurology 1998;51:1395–8 motor asymmetries in unilateral neglect. Neuropsychologia 26. Horowski R: Differences in the dopaminergic effects of the 2001;39:653–64 ergot derivatives bromocriptine, lisuride, and d-LDS as 43. Naturalistic Action Test (NAT). Reading, England, Thames compared with apomorphine. Eur J Pharmacol 1978;51: Valley Test Company, 2003 157–66 44. Schwartz MF, Segal M, Veramonti T, Ferraro M, Buxbaum 27. Barrett AM, Crucian GP, Schwartz RL: Adverse effect of L: The Naturalistic Action Test: a standardized assessment dopamine agonist therapy in a patient with motor-inten- for everyday action impairment. Neuropsychol Rehabil sional neglect. Arch Phys Med Rehabil 1999;80:600–3 2002;12:311–39 28. DeLong M: Primate models of movement disorders of basal 45. Guide for Uniform Data Set for Medical Rehabilitation, ganglia origin. Trends Neurosci 1990;13:281–5 version 5.0. Buffalo, NY, State University of New York at 29. Wichmann T, DeLong M: Physiology of the basal ganglia and Buffalo, 1996 pathophysiology of movement disorders of basal ganglia ori- 46. Friedrich W, Greenberg M, Crnic K: A short form of the gin, in Watts R, Koller W, (eds): Movement Disorders: Neuro- Questionnaire on Resources and Stress. Am J Ment Defic logic Principles and Practice. New York, NY, McGraw Hill, 1983;88:41–8 1997 47. Suen H, Ary D: Analyzing quantitative behavioral observa- 30. Lange K, Kornhuber J, Riederer P: Dopamine/glutamate tion data. Hillsdale, NJ, Lawrence Erlbaum Associates, 1989 interactions in Parkinson’s disease. Neurosci Biobehav Rev 48. Young L: On randomness in ordered sequences. Annals 1997;21:393–400 Math Stats 1941;12:293–300 31. Thomas A, Iacono D, Luciano A, Armellino K, Di Orio A, 49. Pierce S, Buxbaum LJ: Treatments of unilateral neglect: a Onofrj M: Duration of amantadine benefit on dykinesia of review. Arch Phys Med Rehabil 2002;83:256–68 severe Parkinson’s disease. J Neurol Neurosurg Psychiatry 50. Farne A, Rossetti Y, Toniolo S, Ladavas E: Ameliorating 2004;75:141–3 neglect with prism adaptation: visuo-manual and visuo- 32. Tomassini V, Pozzilli C, Onesti E, et al: Comparison of the verbal measures. Neuropsychologia 2002;40:718–29 effects of acetyl L-carnitine and amantadine for the treat- 51. Rossetti Y, Rode G, Pisella L, et al: Prism adaptation to a ment of fatigue in multiple sclerosis: results of a pilot, rightward optical deviation rehabilitates left hemispatial randomised, double-blind, crossover trial. J Neurol Sci neglect. Nature 1998:166–9 2004;218:103–8 52. Levy D, Blizzzard R, Halligan P, Stone S: Fluctuations in 33. Meythaler J, Brunner R, Johnson A, Novack T: Amantadine visual neglect after stroke? Eur Neurol 1995;35:341–3 to improve recovery in traumatic brain injury-associated diffuse axonal injury: a pilot double-blind randomized trial. 53. Small N, Ellis S: Brief remission periods in visuospatial J Head Trauma Rehabil 2002;17:300–13 neglect: evidence from long-term follow-up. Eur Neurol 1994;34:147–54 34. Giacino J, Whyte J: Letter to the Editor. J Head Trauma Rehabil 2003;18:4–5 54. Coslett HB, Bowers D, Heilman K: Reduction in cerebral activation after right hemisphere stroke. Neurology 1987; 35. Schneider W, Drew-Cates J, Wong T, Dombovy M: Cognitive 37:957–62 and behavioral efficacy of amantadine in acute traumatic brain injury: an initial double-blind placebo-controlled 55. Heilman KM, Schwartz HD, Watson KT: Hypoarousal in study. Brain Inj 1999;13:863–872 patients with the neglect syndrome and emotional indiffer- ence. Neurology 1978;28:229–32 36. Diagnostic and Statistical Manual of Mental Disorders, ed 4. Arlington, Va, American Psychiatric Association, 1994 56. Robertson IH: Do we need the “lateral” in unilateral neglect? Spatially nonselective deficits in unilateral neglect 37. Gauthier L, Dehaut F, Joanett J: The bell test: a quantitative and their implications for rehabilitation. Neuroimage 2001; and qualitative test for visual neglect. Int J Clin Neuropsy- 14:S85–90 chol 1989;11:49–54 57. Robertson IH: The relationship between lateralised and 38. Wilson B, Cockburn J, Halligan P: Behavioral Inattention non-lateralised attentional deficits in unilateral neglect, in Test. Titchfield, England, Thames Valley Test Company, Robertson IH, Marshall JC (eds): Unilateral Neglect: Clini- 1987 cal and Experimental Studies. Hove, UK, Lawrence Erl- 39. Cocchini G, Beschin N, Jehkonen M: The fluff test: a simple baum, 1993, pp 257–78 task to assess body representation neglect. Neuropsychol 58. Robertson IH, Mattingley JB, Rorden C, Driver J: Phasic Rehabil 2001;11:17–31 alerting of neglect patients overcomes their spatial deficit in 40. Robertson IH, Manly T, Andrade J, Baddeley BT, Yiend J: visual awareness. Nature 1998;395:169–72

July 2007 Amantadine Treatment of Neglect 537 Authors: Eugene A.A. Rameckers, PT, MRes Lucianne A.W.M. Speth, MD Cerebral Palsy Jacques Duysens, PhD J.S. Hans Vles, PhD, MD Bouwien C.M. Smits-Engelsman, PhD

Afﬁliations: RESEARCH ARTICLE From the Rehabilitation Foundation Limburg (SRL), Valkenburg, The Netherlands (EAAR, LAWMS); Department of Rehabilitation Medicine, Radboud University of Kinematic Aiming Task Nijmegen, The Netherlands (EAAR, JD, BCMS-E); Academic Hospital Measuring Functional Changes in Hand and Arm Maastricht (azM), The Netherlands Movements After Botulinum Toxin-A Injections in (EAAR, LAWMS, JSHV); Avansϩ, University for Professionals, Breda, Children with Spastic Hemiplegia The Netherlands (EAAR, BCMS-E); Motor Control Lab, Department of Kinesiology, K.U. Leuven, Belgium (BCMS-E); and Sint Maartenskliniek- ABSTRACT Research, Development and Rameckers EAA, Speth LAWM, Duysens J, Vles JSH, Smits-Engelsman BCM: Education, Nijmegen, The Netherlands (EAAR, JD, BCMS-E). Kinematic aiming task: measuring functional changes in hand and arm movements after botulinum toxin-a injections in children with spastic hemiplegia. Am J Correspondence: Phys Med Rehabil 2007;86:538–547. All correspondence and requests for Objective: To describe different aspects of a kinematic aiming task reprints should be addressed to E. (KAT) as a quantitative way to assess changes in arm movements within Rameckers, Rehabilitation Foundation Limburg (SRL)-Radboud 2 wks after botulinum toxin-A (BTX-A) injections in children with spastic University, Department of hemiplegia. Rehabilitation Medicine, Reinier Postlaan 4, Nymegen, NL 6525GC Design: Intervention study randomized clinical trial; follow-up within 4 The Netherlands. wks after baseline measurement. 0894-9115/07/8607-0538/0 Results: The KAT gave a high intraclass correlation on movement American Journal of Physical Medicine & Rehabilitation time, spread of end points (END), and index of performance effective Copyright © 2007 by Lippincott (IP-E). After BTX-A, a significant increase of END and IP-E was shown if Williams & Wilkins precision demand in the KAT was high, whereas the inverse occurred

DOI: 10.1097/PHM.0b013e31806dbc3c when speed was more important. These functional changes coincided with a significant decrease of the maximum voluntary contraction of the flexor muscles of the forearm. Muscle tone measured with the Ashworth scale did show a nonsignificant decrease of muscle tone, as did the stretch restricted angle and the active and passive ranges of motion of the elbow and wrist. Conclusions: Muscle force decreased immediately after BTX-A, showing the direct effect of BTX-A. The KAT is an adequate, reproducible way to quantify functional changes after BTX-A in the upper limb. BTX-A has an inverse effect in the precision task when accuracy is important, and it has a positive effect when speed prevails.

Key Words: Cerebral Palsy, Kinematic Aiming Task, Spasticity, Upper Limb, BTX-A, Ashworth Scale, Tardieu Scale

538 Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 tasks that are close to daily life behavior. Related to ABBREVIATIONS the leg, such studies are performed using gait pa- AS ϭ Ashworth scale rameters to quantify change in walking performance.11 In accordance with the walking perfor- ϭ CP cerebral palsy mance, we developed a visual guided aiming task TT ϭ Tardieu scale with quantitative parameters as speed and accuracy to measure whether differences in the parameters SRA ϭ stretch restricted angle of the arm movements could be reproduced reli- PRA ϭ passive restricted angle ably over time, and whether the task would be MVC ϭ maximum voluntary contraction sensitive enough to measure change after intervention. This task, which has been called the kine- ϭ KAT kinematic aiming tasks matic aiming task (KAT), uses two simple visual PSM ϭ percentage successful movements guided aiming movements, with low memory and 17,18 ϭ cognitive load. In many ADL tasks, the reach- END end point spread ing, grasping, and pointing of objects are part of MT ϭ movement time the task. Therefore, this task focuses on grasping and holding an object as well as transferring it with IP-E ϭ index of performance effective one hand to a new target position and adapting the movement to the task requirements. In the present study, we are specifically interested in subtle quantitative changes after botulinum toxin-A (BTX-A) Children with congenital spastic hemiplegia injections, because of the ascribed effect in reduc- show poorer manual skills with their affected side.1 ing spasticity and muscle force.19,20 To test our The most frequently mentioned causes for this loss KAT, we measured changes in the performance of of function are related to degrees of spasticity and children with spastic hemiplegia with the affected increased muscle tone,2 decreased muscle power,3 hand, before and after BTX-A injection in the mus- and decreased muscle length.4 A lot of research has cles of the wrist and elbow, compared with a non- been directed to the functional relation between treatment group. spasticity and manual skills in children with cere- We wanted to test the following questions: bral palsy (CP).2,6 Level of spasticity is most often graded by the Ashworth scale (AS) and the Tardieu 1. How reproducible is the current KAT in the test (TT).7,8 Reduction of spasticity, however, is not nontreatment group of CP children? automatically related to improvement in manual 2. Are the main outcome measures of the present skills.9 Indeed, low correlations have been found KAT related to spasticity as measured with AS between levels of spasticity and outcome measures and SRA, range of motion, and maximum vol- of skilled motor control.10–12 Several explanations untary contraction (MVC)? for this low correlation have been put forward. 3. Does MVC of the finger flexors decrease in the Both AS and TT score the increased resistance injected arm? against a high-velocity stretch. The TT uses the 4. Do the KATs detect changes after treatment difference between the joint angle at the moment (BTX-A)? of the high-velocity resistance (stretch restricted angle [SRA]) and the low-velocity angle (passive METHODS restricted angle [PRA]) as a measure of the dynamic component of muscle tone abnormality.12 Participants However, this increased resistance is probably not For this study, the same children were en- just a measure of spasticity. It is also related to rolled as in the study of Speth et al.21: namely, 20 myogen and collagen stiffness.4,14,15 Scholtes et children with CP (aged 4–16 yrs [mean 9.5] with al.12 even have stated that the solitary use of the diagnoses of hemiplegia according to Hagberg clas- SRA for evaluation of treatment of spasticity is sification22 and the hand function according to probably better because of the interest in high- Zancolli classification (pattern I, IIa, or IIb).23 Zan- velocity stretch as a measurement of velocity-de- colli patterns are described as pendent resistance. Damiano et al.16 have shown that the TT seems to be somewhat task dependent. Zancolli I: active finger extension with less than TT was measured by resistance torque at three 20 degrees of wrist palmar flexion; velocities, during passive knee flexion and exten- Zancolli IIA: active finger extension with more sion, combined with EMG. TT showed weak to than 20 degrees of wrist palmar flexion and wrist moderate relationships with knee motion during dorsal flexion possibility with fisted hand; and gait. This underlines the need to have performance Zancolli IIb: active finger extension with more

July 2007 Kinematic Aiming Task in Spastic Hemiplegia 539 TABLE 1 Baseline characteristics and scores of all 20 children with CP22 (10 ؍ Nontreatment Group (n (10 ؍ Baseline Median Scores Treatment Group (n

Side paresis, frequency 9 right; 1 left 3 right; 7 left Mean age, yrs 9.4 (5 to 16) 9.7 (4 to 16) Sex, frequency Female 5 6 Male 5 4 Zancolli, frequency I45 IIA 4 1 IIB 2 4 Ashworth wrist, frequency 075 124 211 Ashworth elbow, frequency 033 165 212 Stretch restricted angle—wrist, deg 60 (30 to 90) 52.5 (15 to 90) Stretch restricted angle—elbow, deg 160 (100 to 180) 127.5 (50 to 180) Active dorsal flexion—wrist, deg Ϫ18 (Ϫ51 to 45) Ϫ6(Ϫ55 to 55) Active extension—elbow, deg 170 (135 to 180) 170 (140 to 180) Passive dorsal flexion—wrist, deg 70 (65 to 90) 67 (50 to 90) Passive extension—elbow, deg 180 (150 to 180) 180 (160 to 190) Maximal voluntary contraction flexors— 9.9 (2.1 to 23) 8.9 (2 to 19.5) wrist/fingers, N

than 20 degrees of wrist palmar flexion and no Injection Technique and Dosage possibility of dorsal flexion of active wrist. Botox from Allergan was used (dilution 5 U/0.1 ml). Dosage was 2–3 U/kg body weight per Children with hand function according to Zan- muscle in the upper arm, 1–2 U/kg body weight colli III were excluded because of their inability to per muscle in the forearm, with a maximum of 50 extend the wrist and actively grasp and hold an U at any one injection site, with an overall maxi- object with the affected hand. No children with mum dose of 400 U per total body weight.24 Pre- asymmetric mild quadriplegia or diplegia were in- dominantly, the m. adductor pollicis (10 U), the m. cluded in this study. Table 1 provides an overview flexor carpi ulnaris (2 ϫ 20Uto2ϫ 40 U), and the of the baseline characteristics of the participants. m. pronator teres (30–50 U) were injected, and less The parents gave informed consent, and the med- frequent injections were given in the m. flexor ical ethics committee of the Rehabilitation Foun- carpi radialis (30 U to 2 ϫ 30 U), m. biceps bra- dation Limburg approved the study. chialis (2 ϫ 20Uto2ϫ 50 U), m. brachioradialis (40Uto2ϫ 40 U), and m. flexor pollicis brevis (5 Design U). The number of units per total body weight Matching according to age and Zancolli level varied from 2.9 to 5.8 U. For a more detailed description of dose and resulted in ten pairs of children. One child of every 21 pair was randomly allocated to either the treatment localization, see Speth et al. or the nontreatment group.21 Each child was tested Outcome Measures two times. First, the baseline measurement was administered. Then, after 1 mo, each child was Outcome measures are divided according to tested again. For the treatment group this meant measures of the level of function and the level of the baseline measurement was taken about 14 days activity, according to the International Classifica- 25 before the child received BTX-A, and the second tion of Functioning, Disability and Health. measurement was taken about 14 days after BTX-A. The examination consisted of spasticity assess- Function Level ment (AS and SRA), passive and active range of Muscle Tone motion (PROM and AROM), MVC, and the visually Spasticity was measured in supine position guided aiming task (KAT). using the AS and SRA.

540 Rameckers et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 The muscle tone of wrist the and elbow were measured with the original ﬁve-point AS.7 The SRA was assessed by moving the wrist and elbow as quickly as possible (within 1 sec) through the whole range of motion.26 SRA was measured with manual standardized goniometry (MIE Medi- cal Research Ltd clinical goniometry).27

Range of Motion FIGURE 1 The experimental setup with a digitizer as color LCD monitor. The tablet is placed AROM and PROM of the elbow and wrist were horizontally, directly in front of the par- measured in a sitting position, using the same ticipant. Adaptations in equipment have goniometry been made if disablement was too severe to hold the puppet. Muscle Force Muscle force was evaluated using a method that has been described previously.28 The MVC of the finger and wrist flexors was measured (in new- button. After a random period (0.5–1.5 secs), a tone tons) in a task in which subjects had to apply sounded and the target appeared on the digitizer. maximum pressure onto the end of a high-quality This was the go signal for the child, who was then strain lever. The pressure was transmitted onto a required to move as quickly and as accurately as force transducer. possible to the target by shifting the puppet to the target that had appeared on the other side. After Activity Level placing the puppet in the target, the child had to stop and wait for a new auditory go signal. Ten Kinematic Aiming Tasks movements were made in each condition. The four The kinematic aiming tasks (KAT) consisted of conditions were presented in the same order with two different kinds of tasks: the discrete task and increasing difficulty (conditions 1, 2, 3, and 4). the continuous task. After a practice session, the experiment started. One task was called the discrete task and fo- After a practice session, the experiment cused on precision in performance. This task is started. expected to be sensitive to adaptations in force The second task, termed the continuous task, recruitment. In the discrete task, the children had focused on movement speed; within 20 secs, as a custom-made puppet (7-cm height, 2.5-cm many movements as possible had to be made be- width) in their spastic hand and could directly see tween the presented targets in the same four con- their movements. An LCD screen, which could ditions. record the x and y coordinates of the moving After putting the puppet on the left side of the object, was placed in horizontal position in front of digitizer in the starting target area, the investigator the child. Oasis software29 was used to program pressed the start button. After a random period four conditions, which were expected to be easy (0.5–1.5 secs), a tone sounded, and the right target enough for children with spastic hemiplegia to appeared on the digitizer. This was the go signal for perform. The conditions were (a) movement for 10 a series of back-and-forth movements between two cm to a 5-cm target (condition 1); (b) movement positions. Because the starting position was still for 10 cm to a 2.5-cm target (condition 2); (c) visible, it was possible to come back to the start movement for 20 cm to a 5-cm target (condition 3); position after reaching the target. After 20 secs, an and (d) movement for 20 cm to a 2.5-cm target automated auditive stop signal was given. Condi- (condition 4). tions were presented in the same order as in the The goal for the children was to get the puppet discrete task. Discrete and continuous tasks were (2.5-cm diameter) exactly in the target circle (2.5- randomly performed, and after each task, a rest of or 5-cm diameter) while performing a substantial at least 2 mins was included. arm movement (10 or 20 cm) on a digitizer Signal Analysis (Wacom, type Cintiq 18sx, sample rate 206 Hz). In the KAT, both movement speed and accu- This digitizer served as an SXGA full 24-bit color racy are combined. The relationship between LCD monitor. The surface of the digitizer was movement speed and accuracy in goal-directed made of glass, which made the sliding movement movements is expressed in the speed–accuracy very easy to perform. This is illustrated in Figure 1. tradeoff.30 This can be used to equate the process- After putting the puppet on the digitizer in the ing ability or performance of the motor system by starting circle, the investigator pressed the start the Index of Performance Effective (IP-E). In for-

July 2007 Kinematic Aiming Task in Spastic Hemiplegia 541 lation (ICC) for the nontreatment group within TABLE 2 Intraclass correlation of percentage 1mo. of successful movements, end point spread, movement time, and index The dependent variables PSM, END, MT, and of performance effective for IP-E were evaluated by means of the general linear nontreatment group in discrete and model (polynomial), repeated-measures design, continuous tasks, and of maximum with group (2) and session (2) as between-subjects voluntary contraction (MVC) variables and task (2), amplitude (2), and target measurement in the MVC task size (2) as within-subject variables. Alpha level was Intraclass Correlation set at 0.05. Post hoc analysis was done when ap- (ICC) for Nontreatment Discrete Continuous propriate. Group Task Task The changes between the two measurements of the baseline values of MVC, SRA, AROM, and KAT task ICC ICC PSM 0.89 0.95 PROM of the elbow and wrist also were calculated. END 0.87 0.78 Because of the non-Gaussian distribution of the MT 0.81 0.90 data, the Mann–Whitney U test was used to assess IP-E 0.95 0.89 the differences in the changes from baseline to the MVC task ICC second measurement between the two groups. 0.99 Spearman rank correlation was used to examine whether clinical measures (AS, SRA, AROM, PROM, and MVC) at baseline correlated with the PSM, END, MT, and IP-E in both KATs. Statistical mula format, the definition is: significance was set at 0.01 to account for the use of multiple comparisons. ϭ ϩ Ip-E a b * log2 (2A/ETW)/MT where a and b are empirical constants, A ϭ dis- RESULTS tance between targets, ETW ϭ effective target No significant differences between groups width, and MT ϭ movement time. The ETW is were found for any of the baseline variables. calculated as the target size plus the distance between the center of the target and the center of the Reproducibility (Test–Retest Reliability) puppet.31 An overview of ICCs of the PSM, END, MT, and Movement time (secs) per segment was com- IP-E in both KATs and of MVC in the MVC task is puted from the moment the puppet left the start given in Table 2. High ICCs were found for all area until it entered the target area. Spatial accu- outcome measures in both KATs and in the MVC racy was calculated in two ways: first as a clear task, indicating a high reproducibility. dichotomy between the correct (hit) and incorrect (no hit) ending of the movement within the target Correlation Between Clinical Measures and area (percentage of successful movements [PSM]). Functional Outcome Measures Second, the distance of the center of the puppet to At baseline, the outcome measures of the KAT the center of the target area was calculated. This were correlated with the AS, SRA, AROM, PROM, distance was taken as a measure for end point and MVC. The only significant correlation was seen spread (END). IP-E (bits/sec) was calculated. in AROM of the wrist with PSM and END in both tasks (for P values, see Table 3). No significant cor- Statistics relation was found with any of the tests to score Reproducibility of both KAT and MVC mea- spasticity, illustrating the lack of correlation between surement were measured with intraclass corre- KAT outcome measures and clinical measures.

TABLE 3 Correlation between Ashworth scale, stretch restricted angle, active range of motion (AROM), and passive range of motion of wrist and elbow, maximal voluntary contraction, and percentage of successful movements (PSM), movement time, end point spread (END), and index of performance effective in discrete and continuous tasks; only significant correlations are presented Spearman Correlation PSM Discrete PSM Continuous END Discrete END Continuous

AROM 0.78 0.649 Ϫ0.616 Ϫ0.77 Wrist 0.000 0.002 0.004 0.000

542 Rameckers et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 TABLE 4 Changes in median scores of stretch restricted angle (SRA), active range of motion (AROM), and passive range of motion (PROM) in wrist and elbow, and maximal voluntary contraction (MVC) in the treatment and nontreatment groups P Value (11 ؍ Nontreatment Group (n (11 ؍ Change in Median Score Treatment Group (n

SRA wrist, deg 0 (Ϫ10 to 30) 0 (Ϫ10 to 30) 0.9 SRA elbow, deg 5 (Ϫ30 to 70) 0 (Ϫ35 to 50) 0.9 AROM dorsal ﬂexion wrist, deg 12.5 (Ϫ4to82) 3(Ϫ31 to 71) 0.2 AROM extension elbow, deg 5 (Ϫ20 to 20) 0 (Ϫ20 to 30) 0.4 PROM dorsal ﬂexion wrist, deg 0 (Ϫ10 to 10) 0 (Ϫ25 to 15) 0.9 PROM extension elbow, deg 0 (Ϫ10 to 20) 0 (Ϫ5 to 10) 0.6 MVC, N 4 (Ϫ0.9 to 8.5) 0 (Ϫ0.6 to 0.6) 0.005 P values are based on comparisons between the treatment and the nontreatment groups for change from baseline. Between brackets, minimum and maximum scores are represented.22

ϭ Ͻ Effect of BTX-A main effect for task was found (F1,36 40.69, P As seen in Table 4, no significant changes in 0.001), indicating that the tasks differed in accu- differences were measured after BTX-A between the racy. For the treatment group, the PSM changed in treatment and nontreatment groups on SRA, discrete tasks from 82.9% (SD 24) to 79% (SD 23) AROM, and PROM. However, a significant median and in continuous tasks from 60.6% (SD 31) to decrease in MVC of 4 N (minimum: 0.9; maximum: 59.6% (SD 30). In the nontreatment group, these 8.5) (P Ͻ 0.001) occurred in the treatment group; values were 75.7% (SD 27) and 76.8% (SD 26) in no change was observed in the nontreatment discrete tasks and 63.7% (SD 28) and 62.1% (SD group. 32) in the continuous tasks. When examining END, a main effect for task ϭ Ͻ PSM and END was again found (F1,36 5.15, P 0.03), along The first important property of the KAT was with an interaction with group, session, target size, ϭ Ͻ that both different tasks were easy enough for the and target width (F1,36 4.95, P 0.04), indicat- children to perform successfully with their affected ing that both groups had different spreads of the hand. Generally, children in both tasks reached the end points in different conditions after BTX-A, but target in 70% (SD 29) of the movements. The approximately equal END values in the discrete treatment group succeeded at baseline in 71.7% task (1.5 cm [SD 0.8]) and in the continuous task (SD 29) of the movements and post–BTX-A 69.3% (1.8 cm [SD 0.9]). (SD 28), the nontreatment group (69.7% (SD 28) Post hoc analysis of combined tasks did show a and 69.4% (SD 29). The treatment group showed a significant interaction with group, session, target ϭ Ͻ nonsignificant decrease of 2.4% after BTX-A. This size, and target width (F1,76 5.334, P 0.03). As was mainly caused by the most difficult condition shown in Figure 2, an increase in END occurred (2.5-cm target, 20-cm distance) in both tasks. A after BTX-A, especially for the large movements,

FIGURE 2 End point spread for both treatment groups (A) and nontreatment groups (B) for both KAT tasks combined (discrete and continuous), pre (baseline) and post (post–BTX-A or second measurement). Condition 1 ϭ (target width 5 cm, target distance 10 cm); condition 2 ϭ (target width 5 cm, target distance 20 cm); condition 3 ϭ (target width 2.5 cm, target distance 10 cm); and condition 4 ϭ (target width 2.5 cm, target distance 20 cm). Bars represent 1 SE. *Signiﬁcant at 0.05 level.

July 2007 Kinematic Aiming Task in Spastic Hemiplegia 543 FIGURE 3 Movement time for both treatment groups (A and C) and nontreatment groups (B and D) in discrete and continuous tasks, pre (baseline) and post (post–BTX-A or second measurement). Condition 1 ϭ (target width 5 cm, target distance 10 cm); condition 2 ϭ (target width 5 cm, target distance 20 cm); condition 3 ϭ (target width 2.5 cm, target distance 10 cm); and condition 4 ϭ (target width 2.5 cm, target distance 20 cm). Bars represent 1 SE. As can seen in panels B and D, the reproducibility of this task is high.

ϭ Ͻ indicating that larger amplitudes led to poorer and target size (F1,36 22.38, P 0.001) were accuracy after BTX-A. found for this parameter. A trend for differences with session, group, target size, and width and Movement Time ϭ ϭ task occurred (F1,36 3.01, P 0.09). As seen in ϭ Ͻ Main effects for task (F1,36 80.49, P Figure 3 after BTX-A, the treatment group ϭ Ͻ 0.001), target width (F1,36 68.95, P 0.001), moved slightly more slowly in the discrete task

FIGURE 4 Index of performance effective (IPE) for both treatment groups (A and C) and nontreatment groups (B and D) in discrete and continuous tasks, pre (baseline) and post (post–BTX-A or second measurement). Condition 1 ϭ (target width 5 cm, target distance 10 cm); condition 2 ϭ (target width 5 cm, target distance 20 cm); condition 3 ϭ (target width 2.5 cm, target distance 10 cm); and condition 4 ϭ (target width 2.5 cm, target distance 20 cm). Bars represent 1 SE. As can be seen for controls, the reproducibility of this task is high. *Significant at 0.05 level. 544 Rameckers et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 and more quickly in the continuous task, except indicates that wrist AROM might be more impor- for the most difficult condition (target 25 cm, tant for selecting patients for treatment than AS or distance 20 cm). SRA, and it might indicate good selective motor control of the wrist. IP-E With respect to the questions about the direct effects of BTX-A, a significant reducing effect was A significant interaction for IP-E between task, confirmed by the significant decrease in MVC of the target size, distance, group, and session was found forearm flexors within 2 wks after BTX-A injec- (F ϭ 6.09, P Ͻ 0.03), indicating that perfor- 1,36 tions. This decrease in MVC is caused by decreased mance was different for both groups, tasks, and M-reflex and the blocking of a large number of sessions. motor units after BTX-A in the injected muscles, As can be seen in Figure 4, controls performed thus introducing “paresis.”18,32 similarly in both tasks in all conditions. The treat- A nonsignificant reduction in muscle tone and ment group, however, showed a slight decrease in increase in AROM has already been shown in the performance in the discrete task, which focused on study of Speth et al.21 One possible reason for the accuracy, in all conditions. In the continuous task, lack of significance could be that the treatment which is targeted on speed, a slight increase in per- group was heterogeneous and too small. In other formance was shown in the treatment group after placebo-controlled studies, significant short- BTX-A, except for the most difficult (target 2.5 cm, term effects on the muscle tone and AROM of the distance 20 cm) condition. After BTX-A, the treat- wrist after BTX-A (2–6 wks) have been shown for ment group showed a specific decrease in that most spasticity.33,34 difficult condition of IP-E (0.5 bits/sec). Post hoc The most important question was whether analysis showed that only in the continuous task, an there is a detectable effect of BTX-A on the KAT interaction with session, group, target width, and size outcome measures. The PSM did change slightly in was reliable (F ϭ 9.58, P Ͻ 0.005). 1,36 the treatment group (small decrease of 2.9% after injections) in both tasks, and the nontreatment DISCUSSION group maintained the same percentage of PSM. Our study describes two different tasks of the Overall, the treatment group showed increases in KAT to assess functional changes in the hand and END after BTX-A injections, and this effect was arm movements in a quantitative way within 2 wks significant in the most difficult condition. No after BTX-A in children with spastic hemiplegia. changes were seen in the nontreatment group. If The first aim was to determine whether the more accuracy was asked and more distance had to outcomes of the KAT were stable over time. As seen be covered, END enlarged after BTX-A. Overall, MT in Table 4, the ICCs for the KAT are high for the showed no significant change. MT increased in the nontreatment group when tested within 1 mo. The discrete task, but in the continuous task a decrease high reproducibility indicates that the KAT is was seen, except for the most difficult condition. an adequate instrument to measure quantitative This difference in MT between the two tasks leveled changes in manual ability of the spastic arm. Fur- out the effect over all conditions. Because of thermore, the high percentage of correct trials BTX-A, a lower speed induced muscle hypertonia, showed that neither task is too complex for chil- and increases in AROM and PROM were expected, dren with spastic hemiparesis. creating the possibility for faster movements, espe- No significant correlations were found be- cially in an ongoing movement task such as the tween spasticity scores at baseline (AS and SRA) continuous task. Indeed, this was found in all con- and kinematic outcomes in the KAT. One possible ditions except for the most difficult condition, in reason for this lack of correlation is that the mea- which the opposite occurred. Probably, the specific sured resistance is composed of several factors such demand of accuracy in the most difficult condition as spasticity, muscle stiffness, and contractures.15 was so high that MT had to increase.30 In the Furthermore, one may expect that scores obtained by discrete task, the movement time increased in all passive tests, such as AS and PRA, are not comparable conditions, indicating that the accuracy demand and, therefore, are not fully predictive of measures and the exact braking within the target area slowed obtained during active motor control tasks,2,17 or that down the movement. The increase of movement AS and SRA are not usable tests. time in the discrete task could also be explained by However, significant correlations were found a decrease of muscle power after BTX-A. When less between wrist AROM with PSM and END in the muscle activity can be generated and discrete KAT, suggesting that wrist AROM is probably more movements are required, MT can increase if the predictive of manual task performance than is accuracy demand is high and if more force is muscle tone. If a greater wrist AROM was present, needed to control the movement.35,36 In the con- a higher PSM and a lower END were seen. This tinuous task, in contrast, accuracy demands are

July 2007 Kinematic Aiming Task in Spastic Hemiplegia 545 not high, and the effects of a decrease of muscle children with spastic hemiplegia during isometric tasks. tone seem to be more beneficial than the negative Dev Med Child Neurol 2005;47:337–42 impact caused by loss of force. 4. Ponten E, Friden J, Thornell LE, Lieber RL: Spastic wrist flexors are more severely affected than wrist extensors in With respect to the IP-E in the continuous children with cerebral palsy. Dev Med Child Neurol 2005; task, an expected increase occurred after BTX-A, 47:384–9 because the children were able to move more 5. Kuhtz-Buschbeck JP, Sundholm LK, Eliasson AC, Forss- quickly. A decrease in IP-E occurred only in the berg H: Quantitative assessment of mirror movements in children and adolescents with hemiplegic cerebral palsy. most complex condition, when the accuracy de- Dev Med Child Neurol 2000;42:728–36 mand was very high. In general, performance after 6. Steenbergen B, Meulenbroek RG, Rosenbaum DA: Con- BTX-A decreased in the discrete task. Again, an straints on grip selection in hemiparetic cerebral palsy: increased accuracy demand results in decreased effects of lesional side, end-point accuracy, and context. performance after BTX-A. Brain Res Cogn Brain Res 2004;19:145–59 7. Ashworth B: Preliminary trial of carisoprodol in multiple Limitations of the KAT sclerosis. Practitioner 1964;192:540–2 8. Held JP, Pierrot-Desseilligny E: Le bilan moteur central. The presentation of the tasks in the same order In: Reeducation Motrice des Affections Neurologiques. with increasing difficulty was chosen because ran- Paris, JB Bailiere et fils, 1969, pp 31–42 domization made the test far too confusing for the 9. Wasiak J, Hoare B, Wallen M: Botulinum toxin A as an children, and they had difficulty complying with adjunct to treatment in the management of the upper limb in children with spastic cerebral palsy. Cochrane Database the demands. Possibly, fatigue at the moment of Syst Rev 2004;(4):CD003469 the last, most complex condition could occur. 10. Ostensjo S, Carlberg EB, Vollestad NK: Motor impairments However, the results on END and MT showed no in young children with cerebral palsy: relationship to gross clear indication for fatigue. If fatigue had occurred, motor function and everyday activities. Dev Med Child Neu- rol 2004;46:580–9 more END would have been expected in the last 11. Desloovere K, Molenaers G, Feys H, Huenaerts C, test condition, and MT would have increased. This Callewaert B, Walle PV: Do dynamic and static clinical was not observed. In fact, END was lower for the measurements correlate with gait analysis parameters in most difficult condition 1.3 (SD 0.6) than for the children with cerebral palsy? Gait Posture 2006;24: 302–13 most easy condition 1.5 (SD 0.9). MT did increase 12. Scholtes VA, Becher JG, Beelen A, Lankhorst GJ: Clinical slightly from 0.7 (SD 0.2) to 1 (SD 0.3) in the most assessment of spasticity in children with cerebral palsy: a difficult condition. critical review of available instruments. Dev Med Child Neurol 2006;48:64–73 CONCLUSION 13. Boyd RN, Graham JEA, Nattrass GR, Graham HK: Medium- term response characterisation and risk factor analysis of The KAT seems to be an adequate, reproduc- botulinum toxin type A in the management of spasticity in ible way to quantify functional changes after BTX-A children with cerebral palsy. Eur J Neurol 1999;6:S37–45 in the upper limb, especially in the most complex 14. Damiano DL, Quinlivan JM, Owen BF, Payne P, Nelson KC, condition and in the continuous task. The KAT is Abel MF: What does the Ashworth scale really measure and are instrumented measures more valid and precise? Dev very easy to perform for the children and applicable Med Child Neurol 2002;44:112–8 in clinical settings. 15. Smeulders MJ, Kreulen M, Hage JJ, Huijing PA, van der Only AROM of the wrist showed a positive Horst CM: Spastic muscle properties are affected by length correlation with the KAT outcome measures. changes of adjacent structures. Muscle Nerve 2005;32: BTX-A seems to have a direct, inverse effect on 208–15 MT, END, and IP-E in the discrete tasks, when a 16. Damiano DL, Laws E, Carmines DV, Abel MF: Relationship of spasticity to knee angular velocity and motion during high appeal is made on the accuracy and control of gait in cerebral palsy. Gait Posture 2006;23:1–8 braking movements. On the other hand, there is a 17. Boyd RN, Hays RM: Current evidence for the use of botu- positive effect on MT and IP-E in continuous tasks, linum toxin type A in the management of children with when speed is important and less accuracy is de- cerebral palsy: a systematic review. Eur J Neurol 2001; 8(suppl 5):1–20 manded. 18. Rose J, McGill KC: Neuromuscular activation and motor- Muscle force decreased immediately after unit firing characteristics in cerebral palsy. Dev Med Child BTX-A, showing the direct effect of BTX-A. Neurol 2005;47:329–36 19. van Roon D, Steenbergen B, Hulstijn W: Reciprocal tapping REFERENCES in spastic hemiparesis. Clin Rehabil 2000;14:592–600 1. Hanna SE, Law MC, Rosenbaum PL, et al: Development of 20. Wong V: Evidence-based approach of the use of botulinum hand function among children with cerebral palsy: growth toxin type A (BTX) in cerebral palsy. Pediatr Rehabil 2003; curve analysis for ages 16 to 70 months. Dev Med Child 6:85–96 Neurol 2003;45:448–55 21. Speth LA, Leffers P, Janssen-Potten YJ, Vles JS: Botulinum 2. Gordon AM, Duff SV: Relation between clinical measures toxin A and upper limb functional skills in hemiparetic and fine manipulative control in children with hemiplegic cerebral palsy: a randomized trial in children receiving cerebral palsy. Dev Med Child Neurol 1999;41:586–91 intensive therapy. Dev Med Child Neurol 2005;47:468–73 3. Smits-Engelsman BC, Rameckers EA, Duysens J: Muscle 22. Hagberg B, Hagberg G, Olow I: The changing panorama of force generation and force control of finger movements in cerebral palsy in Sweden. VI. Prevalence and origin during

546 Rameckers et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 the birth year period 1983–1986. Acta Paediatr 1993;82: 30. Fitts PM: The information capacity of the human motor 387–93 system in controlling the amplitude of movement. J Exp 23. Zancolli EA, Zancolli Ej: Surgical rehabilitation of the spas- Psychol 1954;47:381–91 tic upper limb in cerebral palsy. In: Lamb DW, ed. The 31. Smits-Engelsman BC, Swinnen SP, Duysens J: The advan- Paralyzed Hand. Edinburgh, Churchill Livingstone, 1987, tage of cyclic over discrete movements remains evident pp 153–60 following changes in load and amplitude. Neurosci Lett 2006;396:28–32 24. Graham HK, Aoki KR, Autti-Ramo I, et al: Recommenda- tions for the use of botulinum toxin type A in the manage- 32. Koman LA, Mooney JF 3rd, Smith BP, Walker F, Leon JM: ment of cerebral palsy. Gait Posture 2000;11:67–79 Botulinum toxin type A neuromuscular blockade in the treatment of lower extremity spasticity in cerebral palsy: a 25. World Health Organisation: International classiﬁcation of randomized, double-blind, placebo-controlled trial. BOTOX functioning, disability and health: ICF. Geneva, WHO, 2001 Study Group. J Pediatr Orthop 2000;20:108–15 26. Scholtes VA, Dallmeyer AJ, Harlaar J, Becher JG: The SPAT: 33. Corry IS, Cosgrove AP, Walsh EG, McClean D, Graham HK: A Clinical Spasticity Assessment for Children with a Spastic Botulinum toxin A in the hemiplegic upper limb: a double- Hemiparesis. Monaco, EACD, 2005 blind trial. Dev Med Child Neurol 1997;39:185–93 27. Horger MM: The reliability of goniometric measurements of 34. Fehlings D, Rang M, Glazier J, Steele C: Botulinum toxin active and passive wrist motions. Am J Occup Ther 1990; type A injections in the spastic upper extremity of children 44:342–8 with hemiplegia: child characteristics that predict a positive 28. Smits-Engelsman BC, Rameckers EA, Duysens J: Late de- outcome. Eur J Neurol 2001;8(suppl 5):145–9 velopmental deﬁcits in force control in children with hemi- 35. Gordon AM, Charles J, Duff SV: Fingertip forces during plegia. Neuroreport 2004;15:1931–5 object manipulation in children with hemiplegic cerebral 29. Jong de WP, Hulstijn W, Kosterman BJM, Smits-Engelsman palsy. II: bilateral coordination. Dev Med Child Neurol 1999; BCM: Oasis: a new macro language for the experimental 41:176–85 research of handwriting. In: Proceedings of the Seventh 36. Gordon AM, Duff SV: Fingertip forces during object manip- Handwriting Conference of the International Graphonom- ulation in children with hemiplegic cerebral palsy. I: antic- ics Society, London, Ontario, 1995 ipatory scaling. Dev Med Child Neurol 1999;41:166–75

July 2007 Kinematic Aiming Task in Spastic Hemiplegia 547 Authors: Rong-Ju Cherng, PhD, PT Chuan-Fei Liu, MS, PT Cerebral Palsy Tin-Wai Lau, MS, PT Rong-Bin Hong, MD

Afﬁliations: From the Department of Physical RESEARCH ARTICLE Therapy, College of Medicine, National Cheng Kung University, Tainan, Taiwan (R-JC, C-FL); Institute of Allied Health Sciences, College of Medicine, National Cheng Effect of Treadmill Training with Kung University, Tainan, Taiwan (R-JC); and Department of Physical Body Weight Support on Gait and Medicine and Rehabilitation, Chi Mei Medical Center, Tainan, Taiwan Gross Motor Function in Children (T-WL, R-BH). with Spastic Cerebral Palsy Correspondence: All correspondence and requests for reprints should be addressed to Rong- ABSTRACT Ju Cherng, Department of Physical Therapy, College of Medicine, Cherng R-J, Liu C-F, Lau T-W, Hong R-B: Effect of treadmill training with body National Cheng Kung University. No. weight support on gait and gross motor function in children with spastic cerebral 1 Ta-Hsueh Road, Tainan, Taiwan palsy. Am J Phys Med Rehabil 2007;86:548–555. 701. Objective: To examine the effect of treadmill training with body weight Disclosures: support (TBWS) on gait and gross motor function in children with spastic This study was supported by NSC cerebral palsy (CP). 92-2218-E-006-003 and through a collaboration of National Cheng Kung Design: Eight children with spastic CP participated in the study. Their University and Chi Mei Medical Center. temporal-distance gait parameters, Gross Motor Function Measure, muscle tone, and selective motor control were assessed three times: two times 0894-9115/07/8607-0548/0 under their regular therapeutic treatment (condition A), and one time after American Journal of Physical Medicine & Rehabilitation receiving the TBWS treatment in addition to their regular therapeutic Copyright © 2007 by Lippincott treatments (condition B). There were two treatment schedules, AAB and Williams & Wilkins ABA. Except for the first one (taken at study entry), the assessments were always taken after 12 wks of treatment. The children were equally divided DOI: 10.1097/PHM.0b013e31806dc302 into two groups and randomly assigned to the two schedules. The two groups were matched according to category of the Gross Motor Function Classification System. Results: The TBWS treatment significantly improved the children’s gait (increases in stride length and decreases in double-limb support percentage of gait cycle) and their Gross Motor Function Measure (dimension D and E scores as well as the total score). No significant improvements on muscle tone or selective motor control were noted. Conclusions: The TBWS treatment improved some gait parameters and gross motor functions in children with spastic CP.

Key Words: Gait Training, Cerebral Palsy, Gross Motor Function, Total Body Weight Support

548 Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 gait training in six nonambulatory children with Independent walking represents a hallmark of mo- CP and in four children with CP who were ambu- tor development and independence in functional latory but who needed varying degrees of support. mobility. It also provides children with the chances Their results show that after a 3-mo program (30 to explore the environment, broaden the minds, mins per session, three sessions per week, 36 ses- and participate in social activities.1 Therefore, en- sions overall), the children demonstrated signifi- deavoring to improve the motor function, espe- cant improvements of motor function.16 Day and cially the independent ambulation of children with colleagues17 also have presented a case of nonam- cerebral palsy (CP), is one of the important mis- bulatory child with spastic tetraplegic CP who ben- sions for clinicians. efited from locomotor training with TBWS.17 In Many different rehabilitation approaches for the present study, we examined the effect of a children with CP were adopted in clinics, each with 12-wk TBWS program on the gait, gross motor a different philosophy.2 For example, according to function, muscle tone, and selective motor control the neurodevelopmental treatment (NDT) or the in children with spastic CP. Bobath concept,3 gait training for walking should begin with the preparation of the motor compo- METHOD nents that can be found in an easier task (e.g., Experimental Design balance in standing). Once the components are trained, the effect is expected to transfer from the The study adopted a within-participant design easier task to the harder one. However, studies with just one factor—that is, the type of treatment. have challenged the philosophy of NDT and its All the participants received their regular thera- treatment effect. For example, it has been shown peutic treatment before receiving the experimental that weight shifting in standing is qualitatively treatment, which was the TBWS gait training plus different from weight shifting in walking, and their regular treatment. They were assessed three training on the former did not necessarily transfer times. There were two kinds of assessment sched- to the latter.4,5 In general, training on standing ules, each assigned to one group of participants. balance had an effect on standing balance only— For the first schedule (AAB), the participants were not on walking or on the symmetry of gait.5 assessed once at study entry and a second time 12 A different philosophy is that of the task-ori- wks later, before receiving the experimental treat- ented approach.2,6 The approach stresses the im- ment. The experimental treatment lasted for an- portance of matching the training task with the other 12 wks. The third assessment took place functional goal of the target task. According to this right after the experimental treatment. For the approach, the best way of training walking is to second schedule (ABA), the participants were also practice walking itself. It has been demonstrated assessed once at study entry. Then, they received that the effect of training was specific to the partic- 12 wks of experimental treatment before being ular characteristics of the task being trained on.5,7,8 assessed a second time. Another 12 wks of regular One illustration of the task-oriented approach treatment elapsed before the third assessment was is treadmill training with body weight support administered. The second assessment in the AAB (TBWS). It consists of a motor-driven treadmill schedule allowed us to examine whether the par- with a harness that suspends the patient’s body ticipants’ performance might change simply be- weight. Because the needs for body weight support cause they were tested twice. The second assess- and balance control are released with such a sys- ment in the ABA schedule allowed us to determine tem, training with repetitive gait cycles can be whether the experimental treatment might pro- provided for patients with spinal cord injuries or duce a long-lasting effect. strokes, even at the early stages of recovery. Prom- The two kinds of assessment schedule were ising results have been reported in the literature in assigned to two groups of participants of equal size, these patients.9–12 TBWS has also been shown to matched on their disability level as determined by facilitate early ambulation in young children13 and the Gross Motor Function Classification System 19 children with Down syndrome.14 However, studies (GMFCS). that have applied TBWS for gait training in children with CP are limited in the literature.15–18 Participants Richards et al.15 first investigated the feasibil- Inclusion criteria for participants in the study ity of applying treadmill gait training to four young were (a) a diagnosis of spastic CP, (b) age between children with spastic CP (1.7–2.3 yr old) four times 3 and 7 yrs old, (c) ability to follow instructions, (d) per week for four months. They showed that tread- Gross Motor Function Classification System rating mill training was feasible even before the children of I–III, and (e) no surgical treatment during the had developed the ability of independent walking.15 preceding 6 mos before study onset. Twenty chil- Schindl et al.16 conducted a similar study of TBWS dren were screened and 12 children met the inclu-

July 2007 Treadmill Training in Cerebral Palsy 549 TABLE 1 Demographic and clinical characteristics of subjects Age, Weight, Height, Gross Motor Function Subject mos Gender kg cm Diagnosis Classiﬁcation System

AAB1a 65 Female 17.5 111 Spastic diplegia III AAB2 47 Female 14 103 Spastic diplegia II AAB3 75 Male 18.5 110 Spastic diplegia III AAB4 41 Male 14 92 Spastic diplegia III ABA1 41 Male 12 89 Spastic diplegia III ABA2 44 Male 13.8 98 Spastic diplegia II ABA3 61 Male 16.2 100 Spastic diplegia III ABA4 54 Male 15 96 Spastic diplegia III a The child dropped out of the program before the third assessment of the program.

sion criteria, but only eight children joined the amount of body weight of suspension was deter- study program. All of the participants were diag- mined by clinical judgment. The weight was mon- nosed with spastic diplegic CP, with ages ranging itored to be sufficient to avoid knee collapse during from 3.5 to 6.3 yrs old. Two children were at level the single-limb support phase and to not hinder II motor function according to the Gross Motor the swing leg from contacting the floor with the Function Classification System, which indicated heel first.18 For the children with Gross Motor that they were able to walk without devices. Six Function class II, the suspension weight needed was children were at level III. They were moderately minimal, just for the purpose of safety. Treadmill impaired and needed devices to ambulate (Table 1). speed was adjusted to a comfortable level for each child and was gradually increased with the improve- Equipment ment of child’s control. Children were encouraged Treadmill and Suspension System not to hold the rail, and they freely moved their arms A commercial treadmill (Trackmaster TM210AC) during gait training. One independent therapist was used for gait training in this study. The tread- (T.H.) facilitated and corrected the gait pattern of the mill started at 0.0 mph and gradually increased child while the child was walking on the treadmill. speed in increments of 0.1 mph. Suspension was The treatment time was 20 min/session, 2–3 sessions/ achieved with LiteGait (LiteGait, Scottsdale, AZ). wk, for a total of 12 wks, in addition to their regular This system consisted of several parts, including a therapeutic exercise program. yoke, overhead straps, an adjustable harness, a base, and an actuator. A harness was provided to Regular Therapeutic Treatment subjects for weight suspension and safety during The regular therapeutic treatment was individ- gait training. ually planned according to the child’s needs, according to the philosophy of NDT. The treatment GAITRite Electronic Walkway System program was set to meet each child’s motor func- The system (GAITRite, CIR. Systems, Inc. tion status before entering the study program. The Clifton, NJ)contains an electronic walkway, a net- goals of the program were to normalize muscle work controller, and software. The walkway is a tone, maintain or increase the joint range of mo- 4.6-m-long, 0.9-m-wide, commercially available tion, increase muscle strength, and improve motor electronic walkway; it contains 13,824 sensors dis- function. The program was 2–3 times/wk, 30 min/ tributed in a 3.6-m-long, 0.6-m-wide, active area. session, and comprised mat exercises of range of The system outputs are temporal-distance gait pa- motion, stretching, strengthening, and motor rameters, such as velocity, cadence, stride length, function activities. Gross motor activities included and others. Excellent reliability of the quantification changing positions, lie to sit, sit to stand, and of temporal-distance gait parameters (intraclass cor- standing. Movement patterns were of concern, and relation coefficient between 0.82 and 0.92) was re- exercise was not to induce or exaggerate the ab- ported.20,21 A high concurrent validity has also been normal movement pattern. demonstrated in reference to a clinical stride analyzer and the Vicon Motion Analysis System.20,22 Outcomes Measurement The outcome measures included muscle tone, Treatment Program selective motor control, gross motor function, and TBWS temporal-distance parameters of gait. Muscle tone TBWS was administered on a treadmill and was measured with the modified Ashworth scale.23 was supported in a LiteGait suspension system. The Selective motor control was measured with the

550 Cherng et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 subject sitting on the floor, with hips flexed and dependent variable was a percentage score, the knees comfortably extended, and with the subject multivariate analysis of variance was not appropri- able to see his or her feet. The subject was asked to ate. Therefore, we adopted the following analytic dorsiflex each foot individually to a target. If the strategy, as reported in our previous paper on the subject was able to dorsiflex his or her ankle with- effect of horse riding on children with CP.30 First, out hip and knee flexion, a grade of 4 would be we decided that the primary analysis should be the given; if ankle dorsiflexion was achieved mainly one that compared the children’s performances using tibialis anterior, but accompanied by hip under the regular treatment and under the exper- and/or knee flexion, a grade of 3 was scored; if imental treatment. To increase power, we needed dorsiflexion was achieved using toe extensor mus- to pool the data from the group that received the cles and some tibialis anterior, a grade of 2 was AAB schedule and the group that received the ABA scored; if limited dorsiflexion was achieved mainly schedule. This means the second and the third using toe extensor muscles, a grade of 1 was given; assessments of the AAB schedule and the first and and a grade of 0 was given when there was no the second assessments of the ABA schedule. But, movement of ankle dorsiflexion.24 Gross motor before we did that, we had to establish that there function was measured with the Gross Motor Func- was no effect of simply taking a test twice. There- tion Measure (GMFM).25 fore, the first analysis that we did was a comparison The GMFM is a criterion-referenced evaluation of the performance between the first and second tool designed specifically for children with CP. The assessments for the children who received the AAB GMFM is composed of 88 test items, categorized schedule. Because the results of the first analysis into five developmental dimensions: dimensions A showed no signs of a test-taking effect (see the (lie/roll), B (sit), C (crawl/kneel), D (stand), and E Results section), we conducted the primary analy- (walk/run/jump). Each item is scored on a four- sis by pooling the data as described above. point rating scale. Item scores for each dimension We also added a third analysis that examined are summed together and converted, yielding a the difference between the second and third assess- percentage score for that dimension. The average ments in the ABA schedule. This analysis would of the percentage scores for all five dimensions inform us whether a potential experimental treat- yields a total score.25 Results of studies have pro- ment effect could be sustained after the treatment vided support for the high internal reliability and had discontinued. The level of statistical signifi- construct validity of measurement of changes in mo- cance was set at 0.05 for all the analyses. tor function.26,27 The test–retest reliability and interrater reliability of the GMFM was also established RESULTS with intraclass correlation coefficients between 0.7 28,29 Effect on Temporal Distance of Gait and 1.0. Therefore, GMFM was chosen as the Parameters outcome measure assessment tool for the study. The temporal-distance gait parameters measured Table 2 presents the group means, standard with the GAITRite electronic walkway included gait deviations, and ranges of gait velocity, stride velocity, stride length, cadence, and double-limb sup- length, cadence, and double-limb support percent- port time as a percentage of gait cycle. According to age of gait cycle for the two groups at three assess- the GAITRite operating manual, the velocity was ob- ments. Our first step analysis revealed no signifi- tained by dividing the distance by the ambulation cant difference in any of the gait parameters time; it was expressed in centimeters per second. The between the first two assessments for children re- stride length was defined and measured on the line ceiving the AAB schedule. These results allowed us of progression between the heel points of two con- to proceed with our primary analysis. The results of secutive footfalls of the same foot; it was expressed the primary analysis revealed a significant effect of in centimeters. The cadence was the numbers of the experimental treatment (i.e., TBWS plus regu- ϭ ϭ footfalls in a minute. One independent therapist lar therapy) on the stride length (F 10.34, P took all the measurements and was not involved in 0.0236) and a marginal, significant effect on dou- ϭ therapy; this therapist was not aware of any child’s ble-limb support percentage of gait cycle (F 6, ϭ grouping or stage within the study. The study was P 0.058) (Fig. 1). No significant change of veloc- approved by and followed the guidelines of the ity or cadence was noted. Finally, our third analysis institutional review board of National Cheng Kung revealed no significant difference in any of the gait University Hospital. parameters between the second and the third assessments for children receiving the ABA schedule. Data Analysis Statistical Analytic System (SAS) version 9.1 Effect on Dimension Score of GMFM for Windows was used for data analysis. Because Table 3 displays the means, standard devia- the sample size of the study was small, and the tions, and ranges of the dimension scores, and the

July 2007 Treadmill Training in Cerebral Palsy 551 TABLE 2 The group means, SD, and ranges of temporal distance gait parameters of two groups at three measurement times Group AAB Group ABA

Mean ؎ SD Range Mean ؎ SD Range

Velocity, cm/s T1 23.49 Ϯ 26.31 3.80–61.10 24.79 Ϯ 34.75 3.37–76.73 T2 24.64 Ϯ 25.09 3.80–60.57 28.42 Ϯ 39.40 4.90–87.30 T3 38.51 Ϯ 47.10 9.77–92.87 29.39 Ϯ 31.67 10.73–76.60 Cadence, steps/min T1 62.85 Ϯ 40.88 32.10–67.27 66.89 Ϯ 64.76 29.87–37.80 T2 63.63 Ϯ 44.65 30.33–55.93 63.56 Ϯ 60.36 28.77–40.17 T3 71.59 Ϯ 61.95 30.77–41.13 63.91 Ϯ 43.57 38.73–46.30 Stride length, cm T1 36.21 Ϯ 19.14 18.46–61.59 34.27 Ϯ 16.79 16.27–56.85 T2 40.26 Ϯ 15.46 20.26–56.18 40.63 Ϯ 20.82 21.01–69.61 T3 51.48 Ϯ 23.04 37.70–78.08 47.05 Ϯ 17.49 33.97–71.28 DLS (%) T1 58.82 Ϯ 26.09 31.77–94.18 60.84 Ϯ 31.39 15.42–87.65 T2 40.05 Ϯ 15.77 20.22–57.72 43.85 Ϯ 20.47 13.53–60.38 T3 26.38 Ϯ 6.47 20.35–33.22 43.95 Ϯ 23.08 13.95–70.22 DLS, double-limb-support time as a percentage of gait cycle.

total score of the GMFM measurement for the two the experimental treatment on the GMFM total score groups at each assessment. At study entry, the (F ϭ 52.74, P ϭ 0.0008) as well as on dimension D children had an average of score 86.8 or 90.5 on score (F ϭ 8.4, P ϭ 0.0338) and on dimension E dimension A, and 84.8 or 86.3 on dimension B. The scores (F ϭ 10.62, P ϭ 0.0225) (Fig. 2). Our third scores were nearly full scores. Therefore, a ceiling analysis found no significant differences in any of the effect of dimensions A and B could be expected. GMFM scores between the second and the third as- Our first step analysis revealed no significant sessments for children receiving the ABA schedule. difference in any of the dimension scores or the total score of GMFM between the first two assess- Effect on Muscle Tone, and Selective ments for children receiving the AAB schedule. Motor Control Therefore, we proceeded with our primary analysis. The results of the three-step analysis of muscle The primary analysis revealed a significant effect of tone and selective motor control showed that there

FIGURE 1 Comparison of the treatment effect of TBWS and regular therapeutic treatment on the gait parameters. TBWS, treadmill training with body weight support; regular, regular therapeutic treatment; DLS, double-limb support percentage of gait cycle. * P Ͻ 0.05; § P ϭ 0.058.

552 Cherng et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 TABLE 3 The group means and SD of Gross Motor Function Measure dimension scores, and total scores for both groups at three measurement times Group AAB Group ABA

Mean ؎ SD Range Mean ؎ SD Range

Dimension A T1 86.8 Ϯ 16.11 62.8–96.1 90.5 Ϯ 4.84 84.3–96.1 T2 88.3 Ϯ 16.02 64.7–100.0 95.1 Ϯ 5.89 88.2–100.0 T3 89.6 Ϯ 11.97 76.5–100.0 97.1 Ϯ 3.73 92.2–100.0 Dimension B T1 84.8 Ϯ 21.98 52.4–100.0 86.3 Ϯ 7.97 75.0–93.3 T2 86.1 Ϯ 19.11 59.5–100.0 90.4 Ϯ 4.36 86.7–95.0 T3 84.0 Ϯ 23.46 57.1–100.0 87.9 Ϯ 5.82 81.7–95.0 Dimension C T1 70.3 Ϯ 27.14 33.3–92.9 80.4 Ϯ 9.00 71.4–88.1 T2 71.5 Ϯ 26.49 38.3–100.0 84.0 Ϯ 13.92 66.7–100.0 T3 74.1 Ϯ 23.75 53.3–100.0 77.4 Ϯ 15.60 54.8–90.5 Dimension D T1 46.2 Ϯ 35.36 5.1–76.9 44.9 Ϯ 22.77 23.1–76.9 T2 43.6 Ϯ 41.98 5.1–87.2 47.4 Ϯ 25.86 25.6–84.6 T3 37.4 Ϯ 41.31 7.7–84.6 40.4 Ϯ 34.05 7.7–87.2 Dimension E T1 19.5 Ϯ 25.71 0–54.2 26.1 Ϯ 27.01 0–63.9 T2 20.5 Ϯ 24.28 1.4–54.2 30.9 Ϯ 25.89 13.9–69.4 T3 34.3 Ϯ 37.80 9.7–77.8 32.6 Ϯ 27.60 13.9–73.6 Total T1 61.5 Ϯ 23.78 30.7–83.6 65.6 Ϯ 12.54 54.4–83.1 T2 62.0 Ϯ 23.79 33.8–85.2 69.6 Ϯ 14.01 57.6–89.8 T3 63.9 Ϯ 26.24 40.9–92.5 67.1 Ϯ 16.26 50.7–89.3

was no effect of testing taking in children receiving DISCUSSION the AAB schedule, no effect of the experimental Using a task-oriented approach, this study ex- treatment, and no difference between the second amined the effect of TBWS gait training on gait and the third assessments in children receiving the performance and gross motor function in children ABA schedule. In addition, there was no signiﬁcant correlation between the change in GMFM or gait with spastic CP. The results show that TBWS gait parameter and the change in muscle tone or selec- training for 12 wks helped to increase stride length tive motor control. and decrease the double-limb support percentage

FIGURE 2 Comparison of the treatment effect of TBWS and regular therapeutic treatment on the dimension scores and total score of GMFM. TBWS, treadmill training with body weight support; regular, regular therapeutic treatment. * P Ͻ 0.05.

July 2007 Treadmill Training in Cerebral Palsy 553 of gait cycle. The training also improved the chil- should be limited. The lack of significant changes dren’s gross motor function, as manifested in their of the scores of dimension A, B, and C were prob- GMFM dimension D and E scores and their total ably attributable to the subjects’ nearly full scores scores. on those dimensions at the baseline. Deficiencies in gait of children with spastic CP The task-oriented approach emphasized that is one of their parents’ major concerns. Gait of the training program should be specific and func- children with CP is characterized with slow veloc- tional (meaningful) to the individual. Results of the ity, short stride length, and poor balance (increase previous studies have shown that both the content of double-limb support percentage of gait cycle).31 and the amount of therapy were important factors At the baseline evaluation, our subjects showed for improvement in the functional outcomes; the slow walking velocity (average, 27.05 cm/sec) and a training effect was larger as the intensity of the short stride length (average, 37.64 cm). These program was increased.15,33,34 Our 12-wk study of numbers were much smaller than those reported TBWS gait training, with the assistance of a ther- in normal children of a comparable age,32 but they apist’s sensory guidance for correct foot placement, were similar to those for children with spastic CP allowed the children to perform a meaningful, at a comparable age level and functional level.31 functional task (walking) with multiple repetitions. After 12 wks of TBWS gait training, children with The TBWS program in our study can be said to be spastic CP showed a significant increase in stride quite intensive and lengthy, and the children, length and decrease in double-limb support per- therefore, showed improvement. centage of gait cycle. Although the change in gait The results of no significant effect of TBWS on velocity did not reach the conventional level of muscle tone and selective motor control—neither statistical significance, it was in the direction of a significant correlation of the change in gait per- improvement. formance or GMFM scores with muscle tone or Children with spastic CP also displayed im- selective motor control—were not surprising. This provement in GMFM dimension D (standing) and E is probably attributable to the fact that muscle tone (walk) scores and total score. Further, the effect of and selective motor control were measured under a TBWS gait training on GMFM dimension D and E static condition, and that gait and gross motor scores and total score seemed to be sustained for at function are dynamic functions. Therefore, the least 12 wks, as demonstrated by a lack of statisti- muscle tone and the selective motor control may cally significant difference in group ABA partici- not be associated with walking performance or pants between the second and the third measure- gross motor function. ments, when the participants had returned to their Several limitations of the present study must regular therapy programs. The results of our study be acknowledged. First, the sample size was small. are partly consistent with the findings of two pre- Therefore, the power of the study to detect some vious studies.15,16 In Richards et al.’s15 study, smaller beneficial effects of training could be lim- young children with spastic CP (age range: 1.7–2.3 ited. The difficulty of recruiting children with CP yrs) showed improvements in GMFM dimension to participate in such a lengthy study was the scores of D and E after 4 mos of a combination of primary reason for the small sample. Most parents the conventional therapy and treadmill training. of children with CP were hesitant to let their chil- Schindl and colleagues16 also noted an improve- dren join a study that lasted 6 mos (24 wks). ment in the GMFM dimension D and E scores in Second, the sample was a convenient one, like that children with either severely involved nonambula- of many other studies. Therefore, the generaliza- tory CP or mildly involved independently ambulation of the study must be limited. Third, the tory CP after 3 mos of TBWS gait training.16 amount of body weight support and treadmill speed GMFM scores reflect a subject’s complex needed for training was determined individually movement patterns that incorporate trunk according to the therapist’s clinical decision. Un- strength and mobility as well as coordination and fortunately, we did not keep a log of these data, balance. Test items of dimensions A, B, and C are and, therefore, we were unable to examine how typically items involving mat activities, such as they might affect the training effect. Nevertheless, rolling, crawling, and transfer activities of sitting. this study, in accordance with previous ones, dem- Test items of dimensions D and E typically com- onstrates that it is feasible to apply TBWS for gait prise upright activities such as standing and walk- training in children with spastic CP, and the re- ing. Our regular therapeutic exercises contained sults are encouraging. some training activities in upright posture; however, they mostly consisted of therapeutic exercises CONCLUSIONS on a mat. Therefore, from a task-oriented point of Our study has demonstrated the effects of view, the effects of the regular therapeutic treat- TBWS gait training on some gait parameters and ment on the GMFM dimensions D and E, if any, gross motor function in children with spastic CP.

554 Cherng et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 However, because of the small sample size of the with partial body weight support in nonambulatory patients present study, and the limited number of studies of with cerebral palsy. Arch Phys Med Rehabil 2000;81:301–6 this kind, a more definitive conclusion cannot be 17. Day JA, Fox E, Lowe J, et al: Locomotor training with partial body weight support on a treadmill in a nonambulatory made until more findings are available. child with spastic tetraplegic cerebral palsy: a case report. Pediatr Phys Ther 2004;16:106–13 ACKNOWLEDGMENTS 18. McNevin NH, Coraci L, Schafer J: Gait in adolescent cerebral palsy: the effect of partial unweighting. Arch Phys Med We would like to extend our special acknowl- Rehabil 2000;81:525–8 edgment to all the participants of this study and 19. Palisano R, Rosenbaum P, Walter S, et al: Development and their families. reliability of a system to classify gross motor function in children with cerebral palsy. Dev Med Child Neurol 1997; REFERENCES 39:214–23 20. Bilney B, Morris M, Webster K: Concurrent related validity 1. Campos JJ, Anderson DI, Barbu-Roth MA, et al: Travel of the GAITRite walkway system for quantification of the broadens the mind. Infancy 2000;1:149–219 spatial and temporal parameters of gait. Gait Posture 2003; 2. Stranger M, Oresic S: Rehabilitation approaches for chil- 17:68–74 dren with cerebral palsy: overview. J Child Neurol 2003;18: 21. Menz HB, Latt MD, Tiedemann A, et al: Reliability of the S79–88 GAITRite walkway system for the quantification of temporo- 3. Bobath K: A Neurophysiological Basis for the Treatment of spatial parameters of gait in young and older people. Gait Cerebral Palsy. London, Heinemann Medical, 1980 Posture 2004;20:20–5 4. Liao HF, Jeng SF, Lai JS, et al: The relation between 22. Webster KE, Wittwer JE, Feller JA: Validity of GAITRite standing balance and walking function in children with walkway system for the measurement of averaged and in- cerebral palsy. Dev Med Child Neurol 1997;39:106–12 dividual step parameters of gait. Gait Posture 2005;22: 5. Winstein CJ, Gardner ER, McNeal DR, et al: Standing bal- 317–21 ance training: effect on balance and locomotion in hemi- 23. Bohannon R, Smith M: Interrater reliability of a modified paretic adults. Arch Phys Med Rehabil 1989;70:755–62 Ashworth scale of muscle spasticity. Phys Ther 1987;67: 206–7 6. Campbell SK: Decision Making in Pediatric Neurological Physical Therapy. Philadelphia, Pa, Churchill Livingstone, 24. Boyd RN, Graham HK: Objective measurement of clinical 1999 findings in the use of botulinum toxin type A for the management of children with cerebral palsy. Eur J Neurol 7. Schmidt RA: Motor Control and Learning: A Behavioral 1999;6:S23–36 Emphasis. Champaign, Ill, Human Kinetics Publishers, 1988 25. Russell D, Rosenbaum P, Cadman D, et al: The gross motor function measure: a means to evaluate the effects of phys- 8. Winstein CJ: Designing practice for motor learning: clinical ical therapy. Dev Med Child Neurol 1989;31:341–51 implications. In: Lister MJ, ed. Step II: Contemporary Man- agement of Motor Control Problems. Washington, DC, 26. Bjornson KF, Graubert C, Buford V, et al: Validity of the Foundation for Physical Therapy, 1991, pp 65–76 Gross Motor Function Measure. Pediatr Phys Ther 1998;10: 43–7 9. Wernig A, Mu¨ ller S: Laufband locomotion with body weight support improved walking in persons with severe spinal 27. McCarthy ML, Silberstein CE, Atkins EA, et al: Comparing cord injuries. Paraplegia 1992;30:229–38 reliability and validity of pediatric instruments for measuring health and well-being of children with spastic cerebral 10. Visintin M, Barbeau H: The effects of parallel bars, body palsy. Dev Med Child Neurol 2002;44:468–76 weight support and speed on the modulation of the loco- 28. Bjornson KF, Graubert C, McLaughlin JF, et al: Test-retest motor pattern of spastic paretic gait. A preliminary com- reliability of the Gross Motor Function Measure in children munication. Paraplegia 1994;32:540–53 with cerebral palsy. Phys Occup Ther Pediatr 1998;18:25–8 11. Hesse S, Bertelt C, Schaffrin A, et al: Restoration of gait in 29. Bjornson KF, Graubert C, McLaughlin JF, et al: Inter-rater nonambulatory hemiparetic patients by treadmill training reliability of the Gross Motor Function Measure. Dev Med with partial body-weight support. Arch Phys Med Rehabil Child Neurol 1998;40:693–701 1994;75:1087–93 30. Cherng RJ, Liao HF, Leung HWC, et al: The effects of 12. Barbeau H, Visintin M: Optimal outcomes obtained with therapeutic horseback riding on children with spastic cere- body weight support combined with treadmill training in bral palsy. Adapted Phys Act Q 2004;21:101–19 stroke subjects. Arch Phys Med Rehabil 2003;84:1458–65 31. Abel MF, Damiano DL: Strategies for increasing walking 13. Thelen E, Ulrich BD: Hidden skills. Monogr Soc Res Child speed in diplegic cerebral palsy. J Pediatr Orthop 1996;16: Dev 1991;56:59–89 753–8 14. Ulrich DA, Ulrich BD, Angulo-Kinzler RM, et al: Treadmill 32. Sutherland DH, Olshen RA, Cooper L, et al: The develop- training of infants with Down syndrome: evidence-based ment of mature gait. J Bone Joint Surg 1980;62:336–53 developmental outcomes. Pediatrics 2001;108:E84–91 33. Bode RK, Heinemann AW, Semik P, et al: Relative impor- 15. Richards CL, Malouin F, Dumas F, et al: Early and intensive tance of rehabilitative therapy characteristics on functional treadmill locomotor training for young children with cere- outcomes for persons with stroke. Stroke 2004;35:2537–42 bral palsy: a feasibility study. Pediatr Phys Ther 1997;9: 34. Malouin F, Potvin M, Prevost J, et al: Use of an intensive 158–65 task-oriented gait training program in a series of patients 16. Schindl MR, Forstner C, Kern H, et al: Treadmill training with acute cerebral accidents. Phys Ther 1992;72:781–93

July 2007 Treadmill Training in Cerebral Palsy 555 Authors: Ray-Yau Wang, PT, PhD Pei-Yi Lin, PT Hemiparesis Chao-Chung Lee, PT, MS Yea-Ru Yang, PT, PhD

Afﬁliations: From the Institute and Faculty of RESEARCH ARTICLE Physical Therapy, National Yang-Ming University, Taipei, Taiwan (R-YW, P-YL, Y-RY); Department of Education and Research (R-YW) and Section of Physical Therapy (Y-RY), Gait and Balance Performance Taipei City Hospital, Taipei, Taiwan; and Department of Physical Medicine Improvements Attributable to & Rehabilitation, Taipei Veterans General Hospital, Taipei, Taiwan Ankle–Foot Orthosis in Subjects (C-CL). with Hemiparesis Correspondence: All correspondence and requests for reprints should be addressed to ABSTRACT Yea-Ru Yang, Associate Professor, Institute & Faculty of Physical Wang R-Y, Lin P-Y, Lee C-C, Yang Y-R: Gait and balance performance improve- Therapy, National Yang-Ming ments attributable to ankle–foot orthosis in subjects with hemiparesis. Am J Phys University, 155, Sec 2, Li Nong St., Med Rehabil 2007;86:556–562. Shih-Pai, Taipei, Taiwan. Objective: To assess the change in the balance performance and the Disclosures: improvement in the gait performance of subjects with hemiparesis, as a This study was supported by grants result of their wearing an ankle–foot orthosis. NHRI-EX92-9129EI and NHRI-EX93- 9129EI from the National Health Design: This was a cross-sectional control trial. Fifty-eight subjects with Research Institute of the Republic of China. hemiparesis of a duration of less than 6 mos participated in this study. Each subject was evaluated for the balance and gait performance with and 0894-9115/07/8607-0556/0 without an ankle–foot orthosis on the affected side. The balance activities American Journal of Physical Medicine & Rehabilitation were evaluated by the Balance Master System, and the gait performance Copyright © 2007 by Lippincott was measured using GAITRite. Williams & Wilkins Results: The increase in movement velocity and the change in maximal DOI: 10.1097/PHM.0b013e31806dd0d3 excursion toward the affected side during the balance testing were found to be correlated significantly with the change in walking speed as a result of wearing an ankle–foot orthosis (r ϭ 0.274, P ϭ 0.039; r ϭ 0.325, P ϭ 0.020; respectively). Only the change in maximal excursion toward the affected side was found to be significantly correlated with the change in nonaffected step length (r ϭ 0.381, P ϭ 0.010). Conclusion: The maximal excursion toward the affected side improved as a result of wearing an ankle–foot orthosis. This correlated with an increase in step length on the nonaffected side and, hence, an improvement in the walking speed of the subjects with hemiparesis.

Key Words: Ankle–Foot Orthosis, Balance, Gait, Hemiparesis

556 Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 experienced an increase in their gait speed and Attaining gait ability remains a challenge for cadence, and an improvement in their dynamic hemiparetic patients. Considerable time and effort standing balance. However, the effectiveness of are required for ambulation training and orthotic wearing an AFO has been found to be minimal for management.1 The spatiotemporal gait parameters patients with hemiparesis of long duration.17 of hemiparetic patients are significantly different Mojica et al.18 found that wearing an AFO signifi- from those of healthy subjects. That is, the walking cantly reduced body sway and increased the maxi- speed of hemiparetic patients is slower, the ca- mum walking speed of eight poststroke hemipa- dence is lower, their step length is shorter, and the retic patients. But a clear correlation between the stance phase of the uninvolved leg is longer.2 improvements in body sway and gait variables has Ankle–foot orthosis (AFO) is typically recom- not yet been established, because of small sample mended to stroke patients to compensate for the size and the heterogeneity of the patients.18 There- effects of impairments to their walking and thus fore, the present study was aimed at assessing a assist their rehabilitation following stroke. In par- large group of hemiparetic individuals within 6 ticular, it is used in cases where there is inadequate mos poststroke for changes in the balance perfor- dorsiflexion in swing and mediolateral subtalar in- mance and improvement in the gait performance stability during stance.3,4 Patients with hemipare- as a result of wearing AFOs. sis are generally inefficient in the heel-strike and pushoff phases of their gait.5 Major AFO studies in METHODS the past on hemiparetic subjects emphasized the All subjects who participated in this study were gait spatiotemporal, kinematic, and kinetic param- referred from medical centers and district hospitals eters. Gok et al.6 found that AFO changed the near the Taipei area in Taiwan. The diagnosis, age, plateau pattern of curve to the usual double peak sex, affected side, and onset time of hemiparesis pattern in their six subjects, indicating better heel were obtained from patient interviews and medical strike and more effective pushoff. They also ob- charts. Ankle muscle strength was evaluated using served an increase in ankle dorsiflexion at heel a handheld dynamometer (PowerTrack II; JTech strike and midswing.6 Improvements in walking Medical). All tests undertaken were isometric tests speed,7,8 gait pattern,9 and stride length10 have in which the dynamometer was held stationary by the been documented. Franceschini et al.11 found that examiner while each subject exerted a maximum AFO significantly improved self-selected speed, force against it. The ankle dorsiflexor strength and stride cycle time, stance, and double-support time ankle plantarflexor strength were obtained in the in their nine studied hemiparetic subjects, with supine position. The lower-extremity function of postonset duration ranging from 2 to 244 mos. all the subjects was evaluated using the Fugl– After a systematic literature review, Leung and Meyer leg subscale, which included 17 items (the Moseley12 suggest that AFO might improve the highest possible score was 34 points).19 The criteria velocity, stride length, gait pattern, and walking for subject selection were as follows: 1) a diagnosis efficiency of people with hemiplegia who could of unilateral hemiparesis secondary to a cerebro- walk without an AFO. vascular accident, with symptoms having lasted It is also common for hemiparetic patients to less than 6 mos; 2) the ability to walk for 10 m suffer alterations in their postural stability or bal- without an assistive device; 3) never having worn ance.13 Turnbull and associates14 found that even an AFO before this study; 4) the ability to follow functionally ambulant hemiparetic patients dem- simple verbal commands or instructions; and 5) onstrated marked limitations in their capacity to having no history of significant orthopedic prob- shift weight, and possessed a reduced range of lems that would interfere with gait and balance weight shift. The greater the weight the paretic performance tested in our study. Fifty-eight sub- limb was able to bear, the greater the distance the jects participated in this study, and all gave their patient could shift the weight. Dettman and col- informed consent before participation. All proce- leagues15 found a significant relationship between dures performed in this study were approved by the postural instability and walking performance. human subject review board of Taipei Veterans However, there have been few studies that have General Hospital. specifically addressed the effects of AFO on the Each subject performed all measurement tests balance dynamics of hemiparetic patients. Chen et within 2 hrs. The tests were carried out both with al.16 found that AFO had significant effects on and without an AFO on the affected foot. There was long-term hemiparetic patients with respect to lat- a 5-min rest period between each test, and the eral weight shifting and weight bearing through testing sequences were randomized. The AFO used their affected side. In an earlier study, the present in this study was a standard, posterior leaf type authors noted that after being provided with an weighing 125 g, with a setting in neutral position. AFO, patients with hemiparesis of recent onset The orthosis came in three different sizes, and each

July 2007 Gait and Balance with Ankle–Foot Orthosis 557 subject wore the size that fitted them best. Before The paired t testing was used to compare the measurement, the subjects took a short time to balance and gait performance between wearing and familiarize themselves with their newly prescribed not wearing the AFO. The correlations between the AFO. changes in balance ability and in gait performance with the AFO were assessed by Pearson correlation Measurement of Standing Balance coefficients. A level of P Ͻ 0.05 was considered The Balance Master System was used in the statistically significant. All statistical analyses were study to test standing balance.17,20,21 The differ- carried out with SPSS version 10.0 for Windows. ence of the weight-bearing distribution (%) between each leg was recorded while the subjects RESULTS stood as still as possible. Limit of stability (LOS) Fifty-eight subjects who met our selection cri- testing was used to record the subjects’ dynamic teria—44 males and 14 females—participated in balance—that is, their ability to control the move- this study. Among them, 34 had right hemiparesis, ment (maximum excursion) and speed (movement and 24 had left hemiparesis; 19 had sustained hem- velocity) of their center of gravity (COG) during orrhagic strokes, and 39 had sustained infarctions. tasks that required weight shifts toward different The mean age of the subjects was 60.36 Ϯ 13.95 yrs directions. During this assessment, the location of (range: 26–84 yrs); the mean onset duration was the patient’s COG was displayed on screen as a 3.29 Ϯ 1.17 mos (range: 1–6 mos); the mean cursor. The patient controlled the cursor by weight dorsiflexor strength measurements (in pounds) on shifting. To perform the assessment task, the pa- the patients’ affected and nonaffected sides were tient needed to move quickly and precisely to make 26.72 Ϯ 11.57 (range: 12.95–45.19) and 40.02 Ϯ the cursor reach the target. Three directions— 7.24 (range: 32.71–47.39), respectively; the mean anterior, affected side, and nonaffected side—were plantarflexor strength measurements on the af- included in the test. The subjects were instructed fected and nonaffected sides were 39.64 Ϯ 16.86 to stand with their arms at their sides and to not (range: 18.33–68.50) and 53.72 Ϯ 13.53 (range: move their feet throughout the testing procedure. 28.00–82.27), respectively; and the mean Fugl– Movement velocity was the average speed in de- Meyer leg score was 25.12 Ϯ 3.97 (range: 17–32). grees per second of the rhythmic movement along Furthermore, because the testing sequences were the specified direction. Maximal excursion was randomized, we compared the demographic char- measured as the distance of the movement toward acteristics between the subjects who were first the designated target, expressed as a percentage of tested wearing the AFO and those who were first maximum LOS distance.17 tested without wearing the AFO. There were no statistically significant differences between these Assessment of Gait groups for age, stroke onset, stroke type, gender, The GAITRite system (CIR System, Inc.) was hemiparetic side, ankle muscle strength, or Fugl– used to measure the subjects’ gait performance.22,23 Meyer leg score (Table 1). The validity and reliability of GAITRite system have Patients’ weight bearing was more evenly dis- been well established.22,23 The GAITRite system pro- tributed when wearing the AFO than when not vided temporal (time) and spatial (distance) gait wearing the AFO. The weight-bearing differences parameters via an electronic walkway connected to were 12.12 Ϯ 8.25 and 8.86 Ϯ 9.31% for without the serial port of a personal computer. The stan- AFO and with AFO, respectively (P ϭ 0.044). Table dard GAITRite walkway contained six sensor pads 2 shows the LOS test results for increase in move- encapsulated in a roll-up carpet with an active area ment velocity toward the affected (3.39 Ϯ 1.62 3.66 m long and 0.61 m wide. As the subject walked deg/sec without AFO and 4.53 Ϯ 1.48 deg/sec with through the walkway, the sensors captured each AFO, P ϭ 0.040) and toward the nonaffected side footfall as a function of time and transferred the (3.93 Ϯ 2.20 deg/sec without AFO and 5.64 Ϯ 5.87 gathered information to a personal computer to deg/sec with AFO, P ϭ 0.012). Table 2 also docu- process the raw data into footfall patterns. The ments the increase in maximal excursion toward computer computed the temporal and spatial gait the affected side (68.70 Ϯ 23.61% without AFO and parameters. The gait parameters included in our 74.81 Ϯ 20.46% with AFO, P ϭ 0.046). The in- study were gait speed, cadence, cycle time, swing creases were significant as a result of wearing an time, stance time, single-support time, double-sup- AFO. The improvements in the patients’ weight port time, step length, stride length, and base distribution and dynamic standing balance were width. Subjects were asked to walk three times further confirmation of our previous study re- through a 10-m hallway, at a comfortable speed, sults.17 The mean comfortable speed increased without any assistive device. The GAITRite walkway from 62.83 Ϯ 26.71 to 66.94 Ϯ 29.47 cm/sec (mean was placed in the middle of the 10-m hallway to of change: 4.45 Ϯ 10.71 cm/sec, P ϭ 0.006) after eliminate the effect of acceleration or deceleration. wearing an AFO (Table 3). Among the spatiotem-

558 Wang et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 TABLE 1 Comparisons of demographic characteristics between subjects performing the tests without the ankle–foot orthosis (AFO) first and subjects performing the tests with the AFO first P Value (29 ؍ With AFO First (n (29 ؍ Without AFO First (n

Age, yrs 60.38 Ϯ 12.59 (30–84) 60.34 Ϯ 15.41 (26–81) 0.99 Months after stroke 3.24 Ϯ 1.35 (1–6) 3.34 Ϯ 0.97 (2–6) 0.74 Gender Male 24 (82.76%) 20 (68.97%) 0.36 Female 5 (17.24%) 9 (31.03%) Hemiplegic side Right 16 (55.17%) 18 (62.07%) 0.79 Left 13 (44.83%) 11 (37.93%) Stroke type Infarction 18 (62.07%) 21 (72.41%) 0.58 Hemorrhage 11 (37.93%) 8 (27.59%) Ankle strength, pounds Affected dorsiflexors 25.98 Ϯ 11.14 (14.10–38.84) 27.45 Ϯ 12.98 (12.95–45.19) 0.42 Nonaffected dorsiflexors 41.4 Ϯ 6.52 (36.82–47.39) 38.65 Ϯ 8.68 (32.71–46.75) 0.63 Affected plantarflexors 40.01 Ϯ 15.63 (18.33–51.50) 39.35 Ϯ 18.52 (21.33–68.50) 0.75 Nonaffected plantarflexors 55.94 Ϯ 10.62 (33.33–66.50) 51.49 Ϯ 17.42 (28.00–82.27) 0.34 Fugl–Meyer leg score 25.41 Ϯ 3.81 (18–32) 24.83 Ϯ 4.16 (17–32) 0.58 Data were expressed as mean Ϯ SD (range) or frequency (percentage).

poral parameters measured, the patients’ step in maximal excursion toward the affected side cor- length (mean of change: 2.69 Ϯ 5.87 cm, P ϭ 0.010 related significantly with the change in step length for the affected side; mean of change: 1.98 Ϯ 5.06 of the nonaffected side (r ϭ 0.381, P ϭ 0.010) and cm, P ϭ 0.008 for the nonaffected side), stride the change in stride length (r ϭ 0.360, P ϭ 0.012) length (mean of change: 4.87 Ϯ 9.55 cm, P ϭ (Table 4). Also, the change in weight distribution 0.002), and base width (mean of change: Ϫ1.55 Ϯ was found to correlate negatively with the change 3.55 cm, P ϭ 0.002) showed significant improve- in step length of the affected side (r ϭϪ0.325, P ϭ ments after wearing an AFO (Table 3). Other gait 0.015) and the change in stride length (r ϭ parameters did not change significantly (Table 3). Ϫ0.264, P ϭ 0.026) (Table 4). The change in the patients’ movement velocity and maximal excursion toward the affected side DISCUSSION were found to correlate significantly with the In the present study, we found that the increase change in walking speed as a result of wearing an in hemiparetic patients’ walking speed when wearing AFO (r ϭ 0.274, P ϭ 0.039; r ϭ 0.325, P ϭ 0.020; an AFO was correlated significantly with an improve- respectively) (Table 4). However, only the change ment in their dynamic balance control on their af-

TABLE 2 Comparisons of balance performance between wearing and not wearing ankle–foot orthosis (AFO) in subjects with hemiparesis (n ϭ 58) 95% Conﬁdence Without AFO With AFO Changes P Value Interval

Weight-bearing difference, % 12.12 Ϯ 8.25 8.86 Ϯ 9.31 Ϫ3.26 Ϯ 4.43 0.044* Ϫ3.863, Ϫ0.660 Movement velocity, deg/sec Anterior 2.53 Ϯ 1.44 2.66 Ϯ 1.19 0.13 Ϯ 1.41 0.498 Ϫ0.506, 0.249 Affected 3.39 Ϯ 1.62 4.53 Ϯ 1.48 1.14 Ϯ 1.61 0.040* 0.060, 1.919 Nonaffected 3.93 Ϯ 2.20 5.64 Ϯ 5.87 1.71 Ϯ 6.22 0.012* 0.482, 2.418 Maximal excursion (%) Anterior 64.88 Ϯ 20.72 68.20 Ϯ 18.41 3.32 Ϯ 17.19 0.180 Ϫ10.926, 1.717 Affected 68.70 Ϯ 23.61 74.81 Ϯ 20.46 6.11 Ϯ 14.20 0.046* Ϫ5.873, Ϫ1.663 Nonaffected 78.70 Ϯ 19.52 78.84 Ϯ 21.58 0.14 Ϯ 18.58 0.955 Ϫ4.790, 5.070 Data were expressed as mean Ϯ SD. Changes: with AFO Ϫ without AFO. * P Ͻ 0.05 vs. without AFO.

July 2007 Gait and Balance with Ankle–Foot Orthosis 559 TABLE 3 Comparisons of spatiotemporal gait parameters between wearing and not wearing ankle–foot orthosis (AFO) in subjects with hemiparesis (n ϭ 58) 95% Conﬁdence Without AFO With AFO Changes P Value Interval

Speed, cm/sec 62.83 Ϯ 26.71 66.94 Ϯ 29.47 4.45 Ϯ 10.71 0.006** 1.226, 6.977 Cadence 88.62 Ϯ 19.06 90.31 Ϯ 22.98 1.75 Ϯ 13.97 0.357 Ϫ1.955, 5.331 Cycle time, secs 1.45 Ϯ 0.49 1.45 Ϯ 0.48 0.00 Ϯ 0.18 0.962 Ϫ0.047, 0.045 Swing time, secs Affected 0.52 Ϯ 0.15 0.53 Ϯ 0.19 0.01 Ϯ 0.08 0.355 Ϫ0.011, 0.030 Nonaffected 0.39 Ϯ 0.08 0.40 Ϯ 0.10 0.02 Ϯ 0.09 0.276 Ϫ0.011, 0.039 Stance time, secs Affected side 0.93 Ϯ 0.38 0.92 Ϯ 0.34 Ϫ0.01 Ϯ 0.17 0.620 Ϫ0.057, 0.035 Nonaffected side 1.06 Ϯ 0.47 1.06 Ϯ 0.49 Ϫ0.00 Ϯ 0.17 0.874 Ϫ0.041, 0.048 Single-support time, secs Affected 0.39 Ϯ 0.08 0.40 Ϯ 0.10 0.02 Ϯ 0.09 0.276 Ϫ0.011, 0.039 Nonaffected 0.52 Ϯ 0.15 0.53 Ϯ 0.19 0.01 Ϯ 0.08 0.355 Ϫ0.011, 0.030 Double- support time, secs 0.54 Ϯ 0.36 0.53 Ϯ 0.35 Ϫ0.01 Ϯ 0.19 0.775 Ϫ0.057, 0.042 Step length, cm Affected 42.29 Ϯ 12.27 44.58 Ϯ 13.19 2.69 Ϯ 5.87 0.010* 0.571, 4.016 Nonaffected 39.98 Ϯ 12.08 41.82 Ϯ 14.63 1.98 Ϯ 5.06 0.008** 0.493, 3.186 Stride length, cm 82.53 Ϯ 22.95 86.86 Ϯ 26.47 4.87 Ϯ 9.55 0.002** 1.616, 7.043 Base width, cm 15.45 Ϯ 4.70 13.95 Ϯ 4.97 Ϫ1.55 Ϯ 3.55 0.002** Ϫ2.428, Ϫ0.566 Data were expressed as mean Ϯ SD. Changes: with AFO Ϫ without AFO. * P Ͻ 0.05; ** P Ͻ 0.01 vs. without AFO.

fected side. This finding concurs with that of a pre- sample size and heterogeneity of the patients, how- vious study by Mojica et al.,18 who have demonstrated ever, Mojica et al.18 could not obtain a clear correla- that the use of AFO by hemiparetic stroke patients tion between the improvements in body sway and gait resulted in a significant decrease in body sway and variables. In the present study, with a larger sample improved walking capacity.18 Because of the small group, we demonstrated a clear, significant correla-

TABLE 4 Correlations (r) between changes in balance performance and gait parameters, resulting from wearing the AFO in subjects with hemiparesis (n ϭ 58) Movement Velocity, deg/sec Maximal Excursion, % Weight-Bearing CorrelationsDifference, % Anterior Affected Nonaffected Anterior Affected Nonaffected

Speed, cm/sec 0.111 0.109 0.274* 0.142 0.069 0.325* Ϫ0.054 Cadence 0.061 0.107 0.062 0.196 0.014 0.238 Ϫ0.086 Cycle time, secs Ϫ0.007 0.198 0.084 0.126 0.022 Ϫ0.106 0.137 Swing time, secs Affected Ϫ0.102 Ϫ0.050 Ϫ0.028 0.073 0.025 Ϫ0.025 0.149 Nonaffected 0.107 0.088 0.148 Ϫ0.006 0.028 0.179 0.162 Stance time, secs Affected 0.063 Ϫ0.230 Ϫ0.101 0.022 0.011 Ϫ0.102 0.068 Nonaffected Ϫ0.090 0.146 Ϫ0.063 Ϫ0.164 0.020 Ϫ0.060 0.093 Single-support time, secs Affected 0.107 0.088 0.148 Ϫ0.006 0.028 0.179 0.162 Nonaffected Ϫ0.202 Ϫ0.050 Ϫ0.028 0.073 0.025 Ϫ0.025 0.149 Double-support time, secs Ϫ0.009 Ϫ0.193 Ϫ0.056 Ϫ0.136 0.024 Ϫ0.064 Ϫ0.011 Step length, cm Affected Ϫ0.325* Ϫ0.131 Ϫ0.004 0.040 0.060 0.072 0.067 Nonaffected Ϫ0.183 Ϫ0.052 Ϫ0.097 Ϫ0.097 0.037 0.381* 0.021 Stride length, cm Ϫ0.264* 0.076 Ϫ0.022 0.003 0.075 0.360* 0.057 Base width, cm 0.189 0.006 0.055 Ϫ0.084 0.189 0.035 Ϫ0.008 * P Ͻ 0.05.

560 Wang et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 tion between the improvement in hemiparetic pain the anterior–posterior movement of the ankle tients’ dynamic balance control and an increase in joint, because of the wearing of an AFO, may have their walking speed when using AFO. resulted in it having a minimal effect on the max- Previous studies have indicated that stroke pa- imal balance range in the forward direction. How- tients showed excessive postural sway and inade- ever, the patients’ ability for forward weight shift- quate weight-shifting capacity in the frontal ing, as indicated by the forward maximal excursion plane.24,25 It has been suggested that patients’ fron- in our study group, was 64.88 Ϯ 20.72% without tal plane balance is particularly responsive to bal- an AFO and 68.20 Ϯ 18.41% with one; this was ance training and recovery.24,25 Chen et al.16 found significantly less than the forward maximal excur- that the use of AFOs by hemiparetic patients had sion for normal, similar-aged adults (88.79 Ϯ significant effects on lateral weight shifting and 17.91; mean age, 71.9 Ϯ 6.7 yr old) that we re- weight bearing through their affected side. The ported in our other related study (Wang et al., greater the weight borne by the affected leg, the unpublished data, 2005). It is important for clini- greater the range the subject could shift the weight cians to take note of the insufficient control over to allow the opposite leg to move forward and, anterior weight shifting and the limitations of consequently, to take a step. According to present AFOs on their forward movement control in sub- 17 and our previous studies, the use of AFO has jects with hemiparesis. shown significant effects on weight bearing during The results of this study reveal a statistically quiet standing, and in movement velocity during significant improvement in the hemiparetic pa- the LOS test toward both the affected and the tients’ dynamic balance control and walking speed nonaffected side, and maximal excursion toward when wearing an AFO. However, given the small the affected side. In the present study, we further percentage of the increase in walking speed, it is documented that the change in maximal excursion noteworthy that, despite the clear statistical signif- toward the affected side correlated significantly icance, the clinical significance of AFOs in gait with the change in step length of the nonaffected performance might remain limited. side (r ϭ 0.381, P ϭ 0.010) and walking speed (r ϭ We also have noted that the step length of the 0.325, P ϭ 0.020) when wearing an AFO. Sackley patients’ affected side increased and the base width and Lincoln26 demonstrated that improved stance decreased during walking when they were wearing symmetry was associated with superior ability to AFOs, even though neither factors significantly af- perform functional tasks. Walker et al.27 further fected walking speed. Ankle dorsiflexion at heel found that, in addition to balance improvement, strike and at midswing has been shown to increase hemiparetic patients’ walking speed was faster after when patients have donned the plastic AFO.6 The balance training. It is thus suggested that improve- increased ankle dorsiflexion at midswing may re- ment in weight bearing and in maximal excursion on the affected side, as a consequence of wearing sult in an increased step length. The decreased base AFO, contributes to the increase in step length on width during walking may further support the im- the nonaffected side and, hence, an improvement provement of dynamic balance control during the in walking speed. In the present study, in addition wearing of an AFO. to improved maximal excursion toward the affected In our study, AFOs affected spatial parameters side, we also have noted that patients’ movement (step length and stride length) but not temporal pa- 11 28 velocity increased toward the affected and nonaf- rameters. Franceschini et al. and Churchill et al. fected side when wearing the AFO. The effect of have already documented the improvement in tem- improved movement velocity with respect to gait poral parameters during walking while wearing an velocity was more significant toward the affected AFO. On examining the characteristics of our sub- side (r ϭ 0.274, P Ͻ 0.05) than toward the nonaf- jects and the characteristics of these previous study fected side (r ϭ 0.142, NS). According to these subjects, we noted that the walking speed of our findings, we speculate that the improvement in subjects (66.9 Ϯ 29.4 cm/sec) was faster than that of 28 walking speed is attributable to the improved bal- their subjects (31 Ϯ 2.0 cm/sec for Churchill et al.’s ance control in the affected side that occurs as a subjects, and 25.8 Ϯ 11.5 cm/sec for Franceschini et result of wearing an AFO. al.’s11 subjects). According to the study by Perry The improvement in our study in patients’ et al.29 on the classification of walking handicaps in walking speed was not, as expected, attributable to stroke populations, the walking ability of our subjects the improved maximal excursion range in the an- fell into the functional walking category of least- terior direction. The maximal excursion in the for- limited community (58 Ϯ 18 cm/sec), whereas ward direction was not changed significantly after Churchill et al.’s28 and Franceschini et al.’s11 subjects wearing an AFO. The test of maximal excursion in were categorized as limited household (23 Ϯ 16 cm/ the forward direction was within the inverted pen- sec) or unlimited household (26 Ϯ 11 cm/sec) am- dulum model of the upright stance. The limitations bulators. Further research is needed to investigate

July 2007 Gait and Balance with Ankle–Foot Orthosis 561 the effect of the AFO on subjects with different walk- ankle-foot orthosis on spatiotemporal parameters and en- ing abilities or speed. ergy cost of hemiparetic gait. Clin Rehabil 2002;17:368–72 In summary, for subjects with hemiparesis of 12. Leung J, Moseley A: Impact of ankle-foot orthoses on gait and leg muscle activity in adults with hemiplegia: system- less than 6-mo duration and with a relatively fast atic literature review. Physiotherapy 2003;89:39–55 walking speed, wearing an AFO can still improve 13. Carr JH, Shepherd RB: Neurological Rehabilitation: Opti- gait speed. We found that this improvement corre- mizing Motor Performance. Oxford, UK, Butterworth Hei- lated significantly with the dynamic balance con- nemann, 1998 trol of the patients’ affected side. The results of our 14. Turnbull GI, Charteris J, Wall JC: Deficiencies in standing weight shifts by ambulant hemiplegic subjects. Arch Phys previous study have shown that wearing an AFO Med Rehabil 1996;77:356–62 had a minimal effect on subjects with long-dura- 15. Dettman MA, Linder MT, Sepic SB: Relationships among 17 tion (Ͼ12 mos) hemiparesis. Therefore, it may walking performance, posture stability, and functional as- be beneficial to prescribe an AFO to patients in the sessments of the hemiplegic patient. Am J Phys Med 1984; early stage, rather than the late stage, of stroke 66:77–90 16. Chen CL, Yeung KT, Wang CH, et al: Anterior ankle-foot recovery. Patients wearing an AFO could improve orthosis effects on postural stability in hemiplegic patients. their dynamic balance control and, thus, increase Arch Phys Med Rehabil 1999;80:1587–92 their walking speed. Furthermore, it is suggested that 17. Wang RY, Yen LL, Lee CC, et al: Ankle foot orthosis on AFOs may play an important role in the functional balance performance of subjects with hemiparesis of differ- improvement of hemiparetic stroke patients. Use of ent durations. Clin Rehabil 2005;9:37–44 AFOs at discharge has been associated with the 18. Mojica JA, Nakamura R, Kobayashi T, et al: Effect of ankle- foot orthosis on body sway and walking capacity of hemi- walking and stairs component of the FIM instru- paretic stroke patients. Tohoku J Exp Med 1988;156:395– ment and the Berg balance scale score.30 The 401 present study has only investigated the immediate 19. Sanford J, Moreland J, Swanson LR, et al: Reliability of the effects of using an AFO. It is possible that more Fugl-Meyer Assessment for testing motor performance in patients following stroke. Phys Ther 1993;73:447–54 sustained use of an AFO would affect these param- 20. Kauffman TL, Nashner LM, Allison LK: Balance is a critical eters as well; further studies are needed to examine parameter in orthopedic rehabilitation. Orthop Phys Ther this possibility. Clinics North Am 1997;6:43–78 21. Lee CC, Wang RY, Yang YR: Correlations among balance REFERENCES and mobility measures for patients with stroke. Formos J 1. Bonita R, Beaglehole R: Recovery of motor function after Phys Ther 2003;28:139–46 stroke. Stroke 1988;19:1497–500 22. Nelson AJ, Zwick D, Brody S: The validity of the GaitRite 2. Hill KD, Goldie PA, Baker PA, et al: Retest reliability of the and the Functional Ambulation Performance scoring sys- temporal and distance characteristics of hemiplegic gait tem in the analysis of Parkinson gait. Neurorehabilitation using a footswitch system. Arch Phys Med Rehabil 1994;75: 2002;17:255–62 577–83 23. Bilney B, Morris M, Webster K: Concurrent related validity 3. Duncan PW, Badke MB: Stroke Rehabilitation: The Recov- of the GAITRite walkway system for quantification of the ery of Motor Control. Chicago, Ill, Yearbook Medical Pub- spatial and temporal parameters of gait. Gait Posture 2003; lishers, 1987 17:68–74 4. Umphred DA: Neurological Rehabilitation. St. Louis, Mo, 24. de Haart M, Geurts AC, Dault MC, et al: Restoration of Mosby, 2001 weight-shifting capacity in patients with postacute stroke: a 5. Lehmann JF, Condon SM, Price R, et al: Gait abnormalities rehabilitation cohort study. Arch Phys Med Rehabil 2005; in hemiplegia: their correction by ankle-foot orthoses. Arch 86:755–62 Phys Med Rehabil 1987;68:763–71 25. de Haart M, Geurts AC, Huidekoper SC, et al: Recovery of 6. Gok H, Kucukdeveci A, Altinkaynak H, et al: Effects of standing balance in postacute stroke patients: a rehabilita- ankle-foot orthoses on hemiparetic gait. Clin Rehabil 2003; tion cohort study. Arch Phys Med Rehabil 2004;85:886–95 17:137–9 26. Sackley CM, Lincoln NB: Single blind randomized con- 7. Tyson S, Thornton H: The effect of a hinged ankle-foot trolled trial of visual feedback after stroke: effects on stance orthosis on hemiplegic gait: objective measures and users’ symmetry and function. Disabil Rehabil 1997;19:536–46 opinions. Clin Rehabil 2001;15:53–8 27. Walker C, Brouwer BJ, Culham EG: Use of visual feedback 8. de Wit DCM, Buurke JH, Nijlant JMM, et al: The effect of an in retraining balance following acute stroke. Phys Ther ankle-foot orthosis on walking ability in chronic stroke 2000;80:886–95 patients: a randomized controlled trial. Clin Rehabil 2004; 28. Churchill AJG, Halligan PW, Wade DT: Relative contribu- 18:550–7 tion of footwear to the efficacy of ankle-foot orthosis. Clin 9. Wong AM, Tang FT, Wu SH, et al: Clinical trial of a low- Rehabil 2003;17:553–7 temperature plastic anterior ankle foot orthosis. Am J Phys 29. Perry J, Garrett M, Gronley JK, et al: Classification of Med Rehabil 1992;71:41–3 walking handicap in the stroke population. Stroke 1995;26: 10. Hesse S, Luecke D, Jahnke MT, et al: Gait function in 982–9 spastic hemiparetic patients walking farefoot, with firm 30. Teasell RW, McRae MP, Foley N, et al: Physical and func- shoes, and with ankle-foot orthosis. Int J Rehabil Res 1996; tional correlations of ankle-foot-orthosis use in the reha- 19:133–41 bilitation of stroke patients. Arch Phys Med Rehabil 2001; 11. Franceschini M, Massucci M, Ferrari L, et al: Effects of an 82:1047–9

562 Wang et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 Authors: Marie-Jose´e Sirois, PhD Andre´ Lavoie, PhD Trauma Clermont E. Dionne, PhD

Affiliations: From the Research Center of the Centre hospitalier affilie´ universitaire RESEARCH ARTICLE de Que´bec, Quebec City, Canada (M-JS, AL, CED); and Department of Social and Preventive Medicine (M-JS, AL) and the Department of Rehabilitation (CED), Laval Predicting Discharge of Trauma University, Quebec City, Canada. Survivors to Rehabilitation Correspondence: A Sampling Frame Solution for a Population-Based All correspondence and requests for reprints should be addressed to Trauma-Rehabilitation Survey Marie-Joseé Sirois, PhD, CHAUQ, Pavillon Enfant-Je´sus, Unite´de recherche en Traumatologie, ABSTRACT 1401-18ie`me Rue, Que´bec, QUE G1J 1Z4, Canada. Sirois M-J, Lavoie A, Dionne CE: Predicting discharge of trauma survivors to rehabilitation: a sampling frame solution for a population-based trauma-rehabili- Disclosures: tation survey. Am J Phys Med Rehabil 2007;86:563–573. Preliminary results of this study were Objectives: presented in part at the 2005 ACRM– To conduct a population-based survey among trauma ASNR Joint Conference, Chicago, IL, survivors on accessibility to rehabilitation services in metropolitan, urban, September 28 through October 2, and rural areas in Quebec (Canada), we attempted to use trauma regis- 2005, and at the Congre`s que´bećois de reádaptation en traumatologie, tries as a sampling frame of subjects discharged to rehabilitation. Dis- Montreal, Canada, November 16–17, charge destinations were inaccurate in many registries, preventing 2005. straightforward identification of the survey subjects. Using the best registry This work was supported by the data, we aimed to identify predictors of rehabilitation discharge and to use Canadian Institutes for Health them to specify a reliable sampling frame for the survey. Research (CIHR, grant no. MOP- 62873) and the Quebec Health Design: A logistic predictive model of rehabilitation discharge was Research Funds (FRSQ, grant no. 3255). developed. This model was applied to data from metropolitan, urban, and rural trauma centers to identify all subjects predicted to be discharged to 0894-9115/07/8607-0563/0 a rehabilitation facility. American Journal of Physical Medicine & Rehabilitation Results: Age, acute-care length of stay, injury-severity score, lower- Copyright © 2007 by Lippincott limb injuries, and seven other predictors were included in the model that Williams & Wilkins generated an area under the ROC curve (AUC) of 0.83 and a classifi- DOI: 10.1097/PHM.0b013e31806e84d2 cation accuracy of 76.6%. The metropolitan, urban, and rural frames were slightly different. They included, respectively, 808, 798, and 929 subjects. Conclusions: The procedure helped us bypass largely inaccurate data from trauma registries. The sampling frames reflected severely injured trauma survivors who were likely to have been referred to postacute rehabilitation.

Key Words: Prediction, Rehabilitation, Trauma, Sampling Method

July 2007 Predicting Discharge to Rehabilitation 563 bilitation in 2000–2001 from level I and II trauma Various barriers to accessing appropriate and centers. Verifications with registrars from urban timely rehabilitation services for individuals with and rural level II trauma centers (there are no level disabilities have been identified recently in the I trauma center in those areas) revealed important United States, Australia, and Canada.1–4 However, miscoding of their discharge destination variable. the effects on the health status of people who Indeed, large proportions of their patients requir- require those services have rarely been studied. In ing postacute rehabilitation were actually coded as the province of Quebec, Canada, numerous diffi- discharged to acute care or home. The QTR vari- culties in accessing rehabilitation services, as well able discharge to rehabilitation was thus invalid as an undersupply of general and specialized reha- for an unquantifiable, but presumably large, num- bilitation resources, have been identified by health ber of patients treated in many urban and rural administrators and researchers in many regions.3–7 level II trauma centers. Clearly, using the list of In an attempt to measure accessibility to a discharges to rehabilitation from those centers large range of rehabilitation services, and its im- would have led to severe undersampling of trauma pact on the health status of trauma survivors, a survivors in those areas. large survey was conducted across metropolitan, To specify the sampling frame, we aimed to urban, and rural areas of the province of Quebec. identify the predictors of rehabilitation discharge This paper presents the analytical method used to in trauma survivors using data from trauma cen- specify the sampling frame for this study. ters that provided standardized, valid information It is well known that cost-effective drawing of on discharge destinations in the QTR. These pre- any unbiased sample from a target population, with dictors would then be used to identify all potential response rates high enough to answer a research subjects for the survey. question, is far from trivial.8 Sampling trauma sur- The literature is rather scarce on who, among vivors discharged to rehabilitation across the prov- trauma survivors, are discharged to rehabilitation. ince of Quebec, through various means, in a large Expert opinions2,15,16 and quantitative studies17–20 number of facilities such as rehabilitation centers converge on the identification of a few medical and and clinics, nursing homes, and community orga- nonmedical predictors of rehabilitation discharge. nizations, would have been prohibitive in cost and Above all, diagnosis and severity of injuries are the time for a probable low response rate. Beatty et al.1 most important factors.2,15,16 In their traumatic have recently illustrated their difficulties in sam- brain injury (TBI) sample, Wagner et al.19 found pling people with disabilities through a similar the injury-severity score (ISS) and the revised approach in the United States. trauma score (RTS) to be significant predictors of The Quebec Trauma Registry (QTR) was thought rehabilitation vs. home discharge. In that study, to offer an ideal, readily available pool of study sub- patients were 0.875 times as likely to be discharged jects. Indeed, the QTR is the mandatory provincial home for each unit increase in ISS and 1.271 times clinical–administrative database for each of the 59 for each unit increase in RTS. Regardless of diag- designated level I, II, and III trauma centers across noses, Emhoff et al.,18 in their study of 109 multi- the province; it contains prospectively collected trauma patients, found significant differences in information on traumatic events, injury severity, discharge FIM scores among patients who were treatments received, and discharge destination discharged home (108/126) and those who went to from acute care. Such trauma registries have been a rehabilitation facility (52/126). In multiple-re- used for more than two decades to evaluate the gression analyses conducted on a subset of TBI effectiveness of prehospital and acute-care trauma patients, Wrigley et al.17 found acute-care length of systems in the United States and Canada,9–11 but stay (ALOS), age, and the presence of complica- these registries are rarely employed in rehabilita- tions to be significantly associated with referral to tion research. The QTR list of trauma survivors rehabilitation. Among nonmedical factors, type of discharged to various rehabilitation settings in the healthcare insurance is considered by many au- province seemed to constitute an ideal sampling thors15,16,20,21 to be a potential source of disparities frame for our survey and a natural and timely exten- in discharge destinations after acute care in the sion toward rehabilitation research of the QTR. United States. Chan et al.20 have shown that in a Coding variability and missing data are com- Washington State level I trauma center, Medicaid mon in large trauma registries, mostly for prehos- and HMO TBI patients had a higher adjusted rela- pital and emergency room data.9,10,12,14 Some spe- tive risk of being discharged to a skilled nursing cific discrepancies in the coding and interpretation home vs. a rehabilitation facility compared with of discharge destinations have also been de- patients insured by fee-for-services plans. In the scribed.12 The QTR is no exception. Indeed, the Wagner et al.19 study, Medicaid and Medicare were data show important variations (1–40%) in the important predictors of discharge home vs. reha- proportion of trauma survivors discharged to rehabilitation. In the province of Quebec, concerns

564 Sirois et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 have been raised about the fact that victims of mandatory for all centers, and an extended dataset motor vehicle crashes, all automatically covered by is mandatory only for the level I trauma centers. the Quebec Automobile Insurance Society (SAAQ), Given that a predictive model was to be built on the might benefit from a privileged access to rehabili- level I trauma center data, and that this model tation services4,7 compared with other types of would later be applied to data from level II trauma trauma victims who benefit from the general, less centers, only the minimal core dataset could be rewarding provincial health-coverage plan. Other considered. nonmedical factors, such as the family’s willing- The inclusion criteria applied to each dataset ness and ability to provide care and support,2,16 the were for patients to be 1) 18 yrs or older, and 2) marital status of the patient,17 and various health discharged alive from acute care. Trauma patients system organizational elements,2 have been iden- were excluded if they 1) discharged themselves tified as influencing TBI patients’ discharge desti- against medical advice, 2) had a missing discharge nations. destination, or 3) were discharged outside the The objectives of this first phase of our pro- province. The available data cover the time period vincial survey on access to rehabilitation services from April 1, 1999 to March 31, 2001. and its impact on trauma survivors’ health status To constitute a meaningful, manageable data- were to 1) use the best registry data to identify base, all available information on demographics, predictors of rehabilitation discharge, 2) use these injury-related variables, acute medical conditions, predictors to specify the sampling frames of sub- acute interventions, and nonmedical factors iden- jects in metropolitan, urban, and rural areas, and tified in the literature as potentially associated with 3) compare the sampling frames generated by this rehabilitation discharge was kept in the datasets. method. A preliminary dataset including 136 variables and 7782 patients was, thus, constituted for model METHODS building. Table 1 shows the characteristics of the The methodology involved four steps: 1) iden- subjects included in each dataset. tifying the most valid and standardized registry data, and pruning the data; 2) building the predic- Step 2: Building the Predictive Model tive model; 3) specifying the sampling frames in Predicted Variable (Outcome) the three areas; and 4) comparing the sampling There are 10 possible postacute destinations in frames. Each step is described in detail below. the QTR: 1) home with help, 2) home without help, Step 1: Identifying the Most Valid and 3) transfer to another acute-care center, 4) long- Standardized Registry Data, and Pruning term care hospital, 5) nursing home, 6) rehabilita- the Data tion, 7) discharge against medical advice, 8) unknown destination, 9) awaiting rehabilitation The data used for model building had to fulfill services, and 10) other. Because the objective of the two conditions. First, the data had to come from current study was to identify trauma survivors dis- trauma centers located in areas where the avail- charged to rehabilitation services among all other ability of the rehabilitation resources would be post–acute care issues, a dichotomous variable was sufficient for the patients to be discharged to them. created, with the categories rehabilitation and Second, to rely with confidence on the discharge- awaiting rehabilitation services in the discharged destination information, the coding of that variable to rehabilitation group; all other categories fit into had to be valid and standardized across the trauma the no rehabilitation discharge group. It is to be centers. The research3,4,7 and administrative6 in- noted that the QTR rehabilitation category does formation available indicates that in the late 1990s, not differentiate between in- and outpatient reha- the nonmetropolitan areas were, in fact, under- bilitation services. served in general and in specialized rehabilitation resources, and the availability of rehabilitation services was best in the two largest metropolitan areas Selecting Potential Predictors: Preliminary of the province. Data from the two level I trauma Analyses centers located in Montreal and the one in Quebec Preliminary bivariate analyses were performed City were thus selected. Moreover, these centers for each variable to select the best potential pre- use full-time, seasoned QTR coders, and they offer dictors of rehabilitation discharge. Selection crite- the best homogeneity in discharge-destination in- ria for dichotomous variables were to present a 5% terpretation and coding. events (rehabilitation discharge) per variable22 and The QTR includes all trauma patients who an area under the ROC curve23 (AUC) larger then died, were transferred to another hospital, were 0.50. AUC ranges from 0 to 1; an AUC of 0.50 is admitted to the intensive care unit, or were admit- considered nondiscriminant. The closer AUC is to ted for more than 2 days. A minimal core dataset is 1, the better the variable can classify the subjects

July 2007 Predicting Discharge to Rehabilitation 565 TABLE 1 Selected characteristics of the subjects, included in the different datasets Datasets Used for Dataset Used for Sampling Model Building Three Level I Centers Two Level I Centers Eight Level II Centers (Metropolitan) (Metropolitan) (Urban and Rural) (8291 ؍ n) (5531 ؍ n) (7782 ؍ n)

Discharges to rehabilitationa 2243 (28.8) 1606 (29.0) 275 (3.3)b according to the registries, n (%) Age, n (%) 18–55 yrs 3759 (48.3) 2522 (45.6) 3936 (47.5) 56–75 yrs 1761 (22.6) 1308 (23.7) 1967 (23.7) Ն76 yrs 2262 (29.1) 1701 (30.7) 2388 (28.8) Male gender, n (%) 4304 (55.3) 2976 (53.8) 4382 (52.9) Mechanism of injury, n (%) Falls 4391 (56.4) 3205 (58.0) 5005 (60.4) MVC 2213 (28.4) 1572 (28.4) 2242 (27.0) Other 1178 (15.2) 745 (13.6) 1044 (12.6) ISS, n (%) 1–8 1941 (24.9) 1443 (26.1) 2813 (33.9)* 9–15 3793 (48.7) 2650 (47.9) 4481 (54.0)* 16–21 925 (11.9) 654 (11.8) 552 (6.7)* 22–29 747 (9.6) 507 (9.2) 346 (4.2)* 30–41 272 (3.5) 196 (3.5) 81 (1.0)* 42–75 104 (1.3) 81 (1.5) 18 (0.2)* Head injuries, n (%) 2293 (29.5) 1556 (28.1) 1185 (14.3)* Spine injuries, n (%) 1112 (14.3) 837 (15.1) 970 (11.8)* Lower-limb injuries, n (%) 4138 (53.2) 2895 (52.3) 5082 (61.3)* Thoracic injuries, n (%) 1181 (15.2) 890 (16.1) 1079 (13.0)* ALOS, days (mean Ϯ SD) 13.2 Ϯ 15.9 12.7 Ϯ 14.6 12.9 Ϯ 16.6 Assisted ventilation, n (%) 926 (11.9) 670 (12.1) 289 (3.6)* Acute-care complications, n (%) 2188 (28.1)* 1258 (22.7) 1942 (23.4) MVC, motor vehicle crash; ISS, injury-severity score; ALOS, acute-care length of stay. a The trauma registries do not differentiate between in- and outpatient rehabilitation; b invalid measure. * Different at the P ϭ 0.01 level.

who experience the outcome (rehabilitation dis- Injury-Related Variables 23 charge) and the subjects who do not. A number of Injury-related variables were derived from the bivariate analyses were performed with continuous 1990 revision27 of the Abbreviated Injury Scale (AIS) variables to identify their best functional form scoring system. The AIS describes the anatomic site (continuous vs. various categorizations), using the and the nature and severity of each injury on an highest AUC as the criterion. ordinal scale ranging from 1 (minor injury) to 6 Missing Data (lethal injury). According to the AIS codes, dichotomous variables were created for each of the following The RTS24,25 and the Glasgow coma scale,24,25 anatomic locations of injury: head, face, cervical area, which are two important measures of injury sever- abdomen and thorax, thoracic spine, lumbar spine, ity, were found to present, respectively, 25% and pelvic area, and lower and upper limbs. Also derived 50% missing data. The RTS is a physiologic severity index derived from the patient’s initial systolic from the AIS were the number of injuries (counts of AIS codes) and the number of injured body regions. blood pressure, respiratory rate, and Glasgow coma 25,28 scale. Missing RTS and Glasgow coma scale scores The ISS is a measure of anatomic severity are reported to be up to 21% in mortality studies.26 of injury. It ranges from 1 to 75 and is deﬁned as Although some imputation methods are promis- the sum of squares of the highest AIS severity ing,26 excluding patients with missing RTS and scores in the three most severely injured body Glasgow coma scale scores is still common, regard- regions. The ISS was developed to predict mortal- less of potential biases.9,26 We chose to remove ity, which ranges from 8% for an ISS Յ 16 to above these two variables rather than the patients, and to 80% for an ISS of 75. Its speciﬁc relationship with rely on other clinical variables for the prediction of rehabilitation discharge is still unknown in the rehabilitation discharge in trauma survivors. general trauma population.

566 Sirois et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 Acute Care–Related Variables the goal of the study, parsimony was not the most Acute complication variables were created us- important aspect of the analyses. The AUC was ing the two QTR coding systems that account for systematically computed for the first five best mod- medical complications during the acute-care els of each size—those containing 1 to 21 variables. phase. The first one includes 16 dichotomous (yes/ The five best models (AUC Ն 0.83) were then sub- no) variables describing the most common compli- mitted to a Monte Carlo cross-validation strategy to 30 cations, such as pneumonia, cardiac failure, and validate the models’ rates of classification accuracy. 30 urinary track infection. Others are coded with the The SAS macro CVLR was used to perform 1000 ninth revision of the International Classification of resampling iterations with a proportion of 25% of the Diseases. data in the validation sets. The most accurate model was selected as the final predictive model of rehabil- Nonmedical Factors itation discharge in trauma survivors. Interaction Among the nonmedical factors possibly related terms were then examined at the P Յ 0.01 level. to rehabilitation discharge, only the type of insur- Regarding the predictive context of the analyses, con- ance coverage was available in the QTR. founding effects were not considered. At the end of the preliminary analysis, a total For the final model, the classification table of 21 variables were identified as potential predic- that summarized the sensitivity, specificity, and tors of rehabilitation discharge in trauma survivors proportion of subjects correctly classified for each (Table 2). level of probability of being discharged to rehabilitation (cut point) was generated. The cut point Model Building: Main Analysis that both maximized the proportion of subjects The 21 preselected variables were considered correctly classified and minimized the proportion in the “best subsets” automated selection of covari- of subjects falsely classified as no rehabilitation ates methods, provided by the SAS PROC LOGIS- discharge (false-negative) was selected. The selec- TIC module, to quickly screen a large number of tion was made regardless of the proportion of false- potential models.23 Parsimonious models are gen- positive subjects generated by the logistic model. erally recommended for numeric and clinical rea- Indeed, we had the a priori clinical knowledge, from sons.23,29 However, given the size of the dataset and level I QTR coders and trauma team coordinators,

TABLE 2 Variables preselected as potential predictors of discharge of trauma survivors to rehabilitation Best Functional Form AUC

Age, yrs Three categories: Յ55, 56–75, 76ϩ 0.62 Gender n/a 0.55 Injury-related variables Lower-limb injuries Yes/no 0.62 Brain injury Yes/no 0.55 Injury to any spine level Yes/no 0.53 Cervical spine injury Yes/no 0.52 Thoracic injury Yes/no 0.52 Number of injuries Count (range 1–19) 0.58 Number of injured body regions Count (range 1–8) 0.57 Injury mechanism Categories 0.57 ISS Categories: 1–8, 9–15, 16–21, 22–29, 30–41, 42–75 0.66 Acute care–related variables ALOS, days Categories: 1–3, 4–5, 10–13, 14–16, 17ϩ 0.77 ICU stay Yes/no 0.59 Any medical complication Yes/no 0.64 Infectious complications Yes/no 0.59 Circulatory/respiratory complications Yes/no 0.55 Dermatologic/wounds complications Yes/no 0.53 Change in mental state Yes/no 0.54 Assisted ventilation Yes/no 0.58 Nonmedical factors SAAQ insurance Yes/no 0.54 CSST insurance Yes/no 0.52 AUC, area under the ROC curve; ISS, injury-severity score; ALOS, acute-care length of stay; ICU, intensive care unit; SAAQ, Quebec automobile insurance board; CSST, Quebec workmen’s compensation plan.

July 2007 Predicting Discharge to Rehabilitation 567 that an unknown number of trauma survivors admit- ries, 24.0% had SAAQ insurance coverage, 7.0% ted to level I trauma centers, who required rehabili- had a workmen insurance coverage, and 64.2% tation services, remained misclassified. Those sub- were covered by the general Quebec public health- jects were most frequently sent home as they awaited care plan. As shown in Table 1, significant differ- their admission to various rehabilitation settings. The ences at the P ϭ 0.01 level were observed between best classification cut point was determined at 0.40. level II and level I datasets. Subjects treated in level Thus, all subjects with a probability Ն0.40 of being II trauma centers were generally less severely in- discharged to rehabilitation, according to the selected jured, with a mean ISS of 9.2 Ϯ 6.1 vs. 12.1 Ϯ 8.8 predictive model, were considered discharged to re- and 11.8 Ϯ 8.9. The level II trauma survivors also habilitation services. presented fewer head and spinal injuries than their level I counterparts. The proportion of lower-limb Step 3: Specification of the Sampling injuries was almost 10% higher in level II data, and Frames in the Three Areas the proportion of assisted ventilation was almost The selected predictive model with the 0.40 8% lower. Finally, the acute-care complications cut point was then applied to two metropolitan were more frequent in the model-building dataset level I and eight urban or rural level II trauma than with the two sampling sets. centers’ data to determine the three sampling frames for the survey. Given the important unreli- Potential Predictors of Rehabilitation ability of the outcome variable discharge destina- Discharge tion from most level II trauma centers, the external Overall, 21 variables were identified as poten- validity of the predictive model could not be as- tial predictors of rehabilitation discharge in trauma sessed using data from those centers. survivors. Table 2 shows those variables, their best functional form (dichotomous, categorical, or con- Step 4: Comparing the Sampling Frames tinuous), and their associated AUC. Figure 1 illus- The characteristics of the trauma survivors trates the proportions of trauma survivors dis- included in the metropolitan, urban, and rural charged to rehabilitation according to the best sampling frames were compared to examine poten- discriminant form of the three strongest predictive tial biases. ANOVA and the Scheffe´ test were con- variables taken independently: ALOS (AUC ϭ 0.77), ducted to test for significant differences in contin- ISS (AUC ϭ 0.66), and age (AUC ϭ 0.62). These uous variables among groups as defined by the intermediate results are important because the role 2 metropolitan, urban, and rural frames. ␹ tests of these variables in the prediction of rehabilitation (exact procedure) were conducted to test for differ- discharge in general trauma populations has not ences in categorical variables. been strongly established yet. All statistical analyses were performed using version 8.2 of the SAS software. Predictive Model of Rehabilitation Discharge Ethical Considerations Table 3 shows the 11 variables included in the This study received approval from the Com- final predictive model of rehabilitation discharge, mission d’acce`s a` l’information du Que´bec (Quebec their estimated regression coefficients, and odds Information Access Board) and from the research ratios with 95% confidence intervals. The general and ethic committees of each of the 11 trauma capacity of this model to discriminate trauma sur- centers that provided data. vivors discharged to rehabilitation from those who were not was excellent, with an AUC of 0.83. At the RESULTS 0.40 cut point, the model had a specificity of 82.8% Comparing the Datasets and a sensitivity of 61.5%. At that probability level, Table 1 presents selected characteristics of the 76.6% of subjects were correctly classified when 7782 trauma survivors included in the dataset used compared with their actual QTR discharge destina- to build the predictive model, as well as the 13,882 tions. The model inappropriately classified 15.9% subjects in the datasets used to identify the sam- of subjects, for whom it predicted no rehabilitation pling frames. All subjects were adults discharged discharge, and 40.9% of subjects, for whom a dis- alive from the trauma centers between April 1, charge to rehabilitation was predicted. In this pre- 1999 and March 31, 2001. Across the data, age dictive model, each unit increase in age, ISS, and ranged from 18.0–106.1 yrs. In model building, ALOS categories (Fig. 1) increased the odds of sampling level I, and sampling level II datasets, the being discharged to rehabilitation by 1.6, 1.7, and mean age was respectively 56.3 Ϯ 23.0, 57.4 Ϯ 1.8 times, respectively. The presence of lower-limb, 23.0, and 56.8 Ϯ 22.8 yrs. Overall, a proportion of spinal, and cervical injuries increased the odds of 72.0% of subjects suffered from two or more inju- rehabilitation discharge by 3.0, 1.5, and 1.5 times,

568 Sirois et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 FIGURE 1 Proportions of trauma survivors discharged to rehabilitation, in the dataset used for model building (n ϭ 7782).

respectively. Trauma survivors with SAAQ insur- 929 (36.6%) from rural areas. Table 4 shows the ance coverage were somewhat less likely to require characteristics of the identified subjects in each rehabilitation than those who were not insured by area. The three groups were significantly different the SAAQ. in some of their characteristics. The rural group was slightly but significantly younger than the Comparing the Sampling Frames metropolitan one. The proportion of falls was The proportions of subjects for whom the higher, and the proportion of motor vehicle model predicted a discharge to rehabilitation ser- crashes was lower, among the metropolitan sub- vices were 28.8% in level I trauma centers and jects than in the two other groups. Except for 20.9% in urban or rural level II trauma centers. lower-limb injuries, which were more frequent in From the 3314 trauma survivors who were identi- the rural sample, there were no other differences fied by the model as discharged to rehabilitation, among the groups with regard to injured body 2535 were eligible for the survey (admitted be- regions. ALOS was shorter in the metropolitan tween January 1, 2000 and December 31, 2001). A sample than in the two other groups. Finally, the total of 808 (31.8%) subjects were from metropol- proportion of acute-care complications was much itan areas, 798 (31.5%) were from urban areas, and higher among the rural subjects.

TABLE 3 Final predictive logistic model of discharge of trauma survivors to rehabilitation (n ϭ 7782) Conﬁdence Regression Standard Odds Interval Coefﬁcient Error Ratio (95%)

Intercept Ϫ4.36 0.15 ALOS 0.47 0.02 1.61 1.54–1.67 ISS 0.51 0.04 1.67 1.56–1.79 Age 0.56 0.04 1.75 1.60–1.90 Spinal injury 0.41 0.11 1.51 1.21–1.88 Assisted ventilation 0.47 0.11 1.60 1.30–1.96 Gender (male) Ϫ0.23 0.07 0.80 0.70–0.91 SAAQ Ϫ0.04 0.01 0.96 0.94–0.98 Acute-care complications 0.31 0.12 1.37 1.07–1.74 Thoracic injuries Ϫ0.74 0.12 0.48 0.39–0.58 Lower-limb injuries 1.10 0.07 3.01 2.63–3.44 Cervical injuries 0.40 0.16 1.49 1.10–2.02 AUC ϭ 0.83 Speciﬁcity ϭ 82.8%; sensitivity ϭ 61.5%; correct classiﬁcation rate ϭ 76.6% ALOS, acute-care length of stay; ISS, injury-severity score; SAAQ, Quebec Automobile Insurance Society, AUC, area under the ROC curve.

July 2007 Predicting Discharge to Rehabilitation 569 TABLE 4 Selected characteristics of the subjects included in each sampling frame Metropolitan Urban Rural (929 ؍ n) (798 ؍ n) (808 ؍ n)

Age, yrs (mean Ϯ SD) 72.8 Ϯ 20.0 70.4 Ϯ 20.2 69.7 Ϯ 21.0* Male gender, n (%) 286 (35.4) 294 (36.8) 346 (37.3) Mechanism of injury, n (%) Falls 593 (73.5) 543 (68.0) 654 (70.5)* MVC 185 (22.9) 232 (29.1) 249 (26.8* Others 30 (3.6) 23 (2.9) 26 (2.7) ISS (mean Ϯ SD) 13.9 Ϯ 9.3 14.1 Ϯ 9.8 14.3 Ϯ 9.9 Injured body regions Head, n (%) 213 (26.4) 191 (23.9) 209 (22.5) Spine, n (%) 118 (14.6) 118 (14.8) 137 (14.8) Lower, n (%) 650 (80.4) 659 (83.0) 797 (86.0)* Thorax, n (%) 114 (14.1) 105 (13.2) 149 (15.7) ALOS, days (mean Ϯ SD) 24.3 Ϯ 18.8 30.5 Ϯ 21.9 32.3 Ϯ 25.6* Assisted ventilation, n (%) 142 (17.6) 132 (16.6) 183 (19.7) Acute-care complications, n (%) 360 (44.6) 276 (34.6) 601 (64.7)* ISS, injury-severity score; MVC, motor vehicle crash; ALOS, acute-care length of stay. * Variables different at the P ϭ 0.01 level.

DISCUSSION number. If we had relied solely on those data, only The goal of this study was to use trauma reg- 275 subjects out of 8291 trauma survivors would istries to specify the sampling frames of trauma have been identified from the eight urban or rural survivors discharged to postacute rehabilitation level II trauma centers (Table 1). services in metropolitan, urban, and rural areas of Eleven predictors of rehabilitation discharge the province of Quebec to further study their ac- were identified in the present study. Like Wagner et cessibility to rehabilitation services and their long- al.,19 who found rehabilitation discharge to in- term health status. Because direct use of the dis- crease 1.14 times (1/0.875 risk increase of home charge-destination information recorded in the discharge) with each unit increase in ISS, we iden- QTR was impossible, we determined the sampling tified the ISS as a strong predictor of rehabilitation frames through 1) the development of a predictive discharge. However, some important differences logistic predictive model of rehabilitation discharge, between our study and Wagner et al.’s19 have to be using the best available data; and 2) the application of noted. Wagner et al.’s19 relative risk of rehabilita- that model to data from two metropolitan level I and tion discharge was identified in data that included eight urban or rural level II trauma centers, to iden- only TBI subjects discharged to either rehabilita- tify all potential survey subjects. tion settings or home. Our study involved all The differences observed between the level I trauma patients with every possible discharge des- and level II datasets used in this study (Table 1) are tination. Wagner et al.19 considered the ISS a con- reflective of the Quebec trauma-care system, which tinuous predictor, whereas we identified its most is designed to admit, either directly or through discriminant form in our data. The ISS categories transfers, the most severely injured patients to that we developed for rehabilitation discharge pre- level I trauma centers. Indeed, the critically injured diction are different from those broadly used to trauma patients first brought to level II centers predict mortality.31 Although much more research who require brain and/or spine surgeries and other is needed to fully understand its role in the matter, ultraspecialized acute care are actually transferred our results indicate that the ISS might have a to level I trauma centers. unique function in predicting rehabilitation dis- The multivariate predictive model included 11 charge in general trauma populations. predictors of rehabilitation discharge that, alto- Age and ALOS also were expected to predict gether, were used to identify 2535 subjects, among rehabilitation discharge. Wrigley et al.,17 in a very whom 808 were living in metropolitan areas, 798 different sample (younger; TBI only), also found in urban areas, and 929 in rural areas. that age and ALOS, both used as continuous vari- Our results generally confirm the a priori hy- ables, were related to referral to inpatient rehabil- pothesis that using the QTR straightforwardly to itation. Our analyses suggest that specific age and determine the sampling frames of trauma survivors ALOS levels can be used to estimate the proportion discharged to rehabilitation from level II trauma of patients discharged to rehabilitation services in a centers would have severely underestimated their general trauma population; this could be useful

570 Sirois et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 and practical in clinical settings. In our data, ALOS among the metropolitan subjects may be explained was, by far, the strongest predictor of rehabilitation by the higher number of falls in metropolitan areas discharge. This could probably be explained by the and the higher mortality related to falls in subjects fact that ALOS in itself encompasses a number of over 65 yrs of age in urban and rural areas.33 The factors potentially related to rehabilitation dis- differences in ALOS between the metropolitan and charge, such as the severity of injuries and medical the urban/rural areas could be largely explained by and/or surgical complications. some specific aspects of the healthcare system in Contrary to what was expected, according to each area. For instance, in some of the urban and our model, trauma survivors with automobile rural areas, the inpatient rehabilitation services are (SAAQ) insurance coverage were less likely to be provided in the rehabilitation unit of the same discharged to rehabilitation. This association, al- hospital. Such patients are, thus, not coded as though statistically significant, is very close to null discharged to rehabilitation in the hospital trauma (odds ration: 0.96; 95% confidence interval: 0.94– registries, because they are still considered acute- 0.98). Given the predictive context of the analyses care stay. Thus, in those hospitals, the ALOS was that did not consider confounding between vari- artificially longer. The higher pressure for beds in ables, the automobile insurance coverage odds ra- metropolitan centers might also contribute to the tio in the model could be a statistical artifact at- shorter ALOS observed in those areas. The differ- tributable to model specification. This will have to ences observed between the three sampling frames be clarified in a further study. might also be attributable, to some degree, to the The impressive magnitude of the lower-limb fact that the predictive model developed with data injuries’ odds ratio (3.0) in the model is probably from level I trauma centers might not be fully linked to the important representation, in our data, adapted for use in level II data. of senior subjects (n ϭ 1734, mean age ϭ 75.9 yrs) One important limitation of this study is that who suffered from hip fractures as a result of falls other outcome destinations in the trauma regis- (92.1% of hip fractures). Most of these subjects are tries, such as home with or without help and long- routinely referred to in- or outpatient rehabilita- term care facility most likely involve some provision tion services. This also explains the sharp increase of rehabilitation services. However, one cannot as- in rehabilitation discharges at the ISS of 9, which sume that all patients coded as sent home or to a corresponds to the ISS for isolated hip fractures. long-term care facility were indeed discharged to The inclusion of hip-fracture patients in the some form of rehabilitation service. It would have QTR results in a more elderly sample and in a been abusive to include those destinations in our larger proportion of falls as a mechanism of injury rehabilitation outcome variable. Because these two compared with some other North American trauma destinations are associated with less intense reha- registries. Case criteria for entry into statewide bilitation services, we can hypothesize that the trauma registries vary considerably.32 Any compar- predictive model developed in this study identified ison of study results (such as ours) that emerge the most severe trauma survivors discharged to from North American trauma registries requires postacute rehabilitation. Thus, the complete and caution regarding these criteria. true population of trauma survivors discharged to a Taken altogether, the 11 predictors consti- broad range of rehabilitation services remains un- tuted a predictive model that generated 15.9% of known. false-negative subjects (wrongly predicted as no This work reflects some of the obstacles asso- rehabilitation discharge) and 40.9% of false-posi- ciated with the use of large administrative data- tives. This proportion of false-positives might seem bases such as trauma registries for research pur- high; however, knowing a priori that we were deal- poses. The use of databases is generally faster and ing with some remaining miscoding of the dis- cheaper when the research focuses on questions charge destination in the dataset used to build the that the database was designed to answer. In North predictive model, we were confident that a fair American trauma-care systems such as Quebec’s, proportion of those subjects would be found to be trauma registries have been used for the evaluation true positives (i.e., discharged to some form of of prehospital and acute-care effectiveness with re- rehabilitation services) once interviewed. This will gard to mortality, acute complications, and re- have to be confirmed in the second phase of the source use, most often linking them to injury se- study. For the same reason, the performance indi- verity. Consequently, much attention has been cators reported in Tables 2 and 3 must be consid- given by registry managers to controlling the qual- ered with caution. ity of “early” trauma variables such as emergency Some differences were observed between the and acute-care medical procedures, the coding of metropolitan, urban, and rural sampling frames of injuries, and mortality/survival outcome. Trauma trauma survivors generated by our method (Table registries have not been traditionally designed to 4). The older age and higher frequency of falls accurately ascertain the various rehabilitation out-

July 2007 Predicting Discharge to Rehabilitation 571 comes or to be used for rehabilitation research. among the most severe trauma survivors who Besides, administrative and research use of the could be discharged to rehabilitation services. It QTR has emerged exclusively from large metropol- also led to the beginning of an understanding of itan level I trauma centers. These centers have the specific role of some important variables in the more incentive to maintain high-quality data, in- prediction of postacute rehabilitation discharge in cluding outcome data. Consequently, we ran into a general population of trauma survivors. These substantial difficulties when we attempted to use results would certainly need to be replicated with the databases to identify trauma survivors dis- data from other trauma registries. charged to various rehabilitation settings outside To our knowledge, this is the first study to the large metropolitan areas. The extent of the examine the implications of determining sampling discrepancies observed in the proportions of reha- frames of trauma survivors discharged to rehabili- bilitation discharges in the QTR between the level tation services through an acute-care trauma reg- I (28–39%) and level II trauma centers (0–9%) was istry. Scrutinized examination of the data has re- not expected, because the latter do admit large vealed important flaws in the quality of outcome proportions of severely injured patients (ISS Ն 15) data that prevent researchers from knowing the who are likely to require postacute rehabilitation. complete size and nature of that population. If We did not expect the outcome data from urban statewide trauma registries have to serve some and rural level II trauma centers to reflect the purpose in trauma/rehabilitation research, the administrative organization of the healthcare set- quality of their outcome data, other than living tings in those areas rather than the “real” postacute status, has to improve considerably to facilitate this events to such an extent. In most urban or rural emerging field in trauma research. areas, the in- or outpatient rehabilitation services were provided in one of the hospital wards or by REFERENCES rehabilitation centers administratively merged with 1. Beatty P, Hagglund K, Neri M, et al: Access to health care the trauma centers. Transfers to such rehabilitation services among people with chronic or disabling conditions: units or centers were interpreted as a continuation of patterns and predictors. Arch Phys Med Rehabil 2003;84: 1417–25 the acute-care phase. 2. Foster M, Tilse C, Fleming J: Referral to rehabilitation Instead of giving up on the use of the QTR to following traumatic brain injury: practitioners and the pro- identify trauma survivors discharged to rehabilita- cess of decision-making. Soc Sci Med 2004;59:1867–78 tion services, we tried to “bypass” the obstacles by 3. Dumont C, Vincent C, Beauregard L, Fougeyrollas P: E´ tude using the analytical procedure presented in this descriptive sur la re´gionalisation des services de reádapta- tion en de´ficience physique dans le centre et l’est du Que´- paper. Even considering the difficulties we encoun- bec. Ruptures 2000;7:23–40 tered, the method used to specify the sampling 4. Vincent C, Dumont C, Be´gin L, et al: La re´gionalisation des frames for this study was faster and cheaper. In- services de reádaptation, ce qu’en pensent les usagers dans deed, the whole process, including authorizations le centre et l’est du Que´bec. Can Fam Physician 2001;47: from each trauma center’s research and ethics re- 1769–75 view board, data preparation, model building, and 5. Canada Health Action: Building on the legacy—Volume II. Striking a Balance Working Group Synthesis Report. Avail- identification of the potential survey subjects, re- ablet at: http://www.hc-sc.gc.ca/english/care/health_forum/ quired the equivalent of one full-time research publications/finvol2/balance. Accessed May 1, 2007. position for 6 mos. Reaching subjects across the 6. AERDPQ: En Toute Urgence. Que´bec, Canada, Association province through different rehabilitation settings des e´tablissements en reádaptation physique du Que´bec, (which are much more numerous than trauma 1999 centers) would have been much more time- and 7. Gervais M: La Mise en Place de Re´seaux de Neurotraumatologie: Leçons Apprises d’Expe´riences Que´- resource consuming. bećoises. Que´bec, Canada, Universite´ Laval, 2000 Obviously, the survey results will have to con- 8. Levy P, Lemeshow S: Sampling of Population: Methods and firm that all the trauma survivors identified by our Applications, ed 3. New York, NY John Wiley & Sons, 1999 method were, in fact, discharged to some rehabil- 9. Jurkovich G, Mock C: Systematic review of trauma system itation services. The differences observed among effectiveness based on registry comparisons. J Trauma the three sampling frames generated by our 1999;47:S46–55 method will have to be accounted for in the anal- 10. Mann N, Mullin R, Mackenzie E: Systematic review of pub- lished evidence regarding trauma system effectiveness. yses and interpretations of the survey results that J Trauma 1999;47:S25–33 will be presented in a future paper. 11. Maier R, Rhodes M: Trauma performance improvement, in Rivara FP, Cummings P, Koepsell TD, Grossman DC, Maier CONCLUSION RV (eds): Injury Control. Cambridge, UK, Cambridge Uni- The challenges encountered in this survey in versity Press, 2001, pp 236–49 trying to specify reliable sampling frames were not 12. Owen J, Bolenbaucher R, Moore M: Trauma registry databases: a comparison of data abstraction, interpretation trivial. It was achieved with the help of statistical and entry at two level I trauma centers. J Trauma 1999;46: techniques, which, most likely, identified those 1100–4

572 Sirois et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 13. Rodenberg H: The Florida trauma system: assessment of a 22. Bernard PM: Re´gression Logistique: Notes de Cours, vol II. trauma statewide data base. Injury 1996;27:205–8 Available at: http://w3.res.ulaval.ca/cours-epm-64312 14. Joseph L, Be´lisle P, Tamin H, Sampalis J: Selection bias 23. Hosmer D, Lemeshow S: Applied Logistic Regression, ed 2. found in interpreting analyses with missing data for the New York, NY, John Wiley & Sons, 2000 prehospital index for trauma. J Clin Epidemiol 2004;57: 24. Champion H, Sacco W, Copes W, et al: A revision of trauma 147–53 scores. J Trauma 1989:623–9 15. Haas J: Admission to rehabilitation centers: selection of 25. O’Keefe G, Jurkovich G: Measurement of injury severity and patients. Arch Phys Med Rehabil 1988;69:329–32 co-morbidity, in Rivara FP, Cummings P, Koepsell TD, 16. Osberg J, Unsworth C: Trauma-rehabilitation connections: Grossman DC, Maier RV (eds): Injury Control. Cambridge, discharge and admission decisions for children. Pediatr UK, Cambridge University Press, 2001, pp 32–46 Rehabil 1997;1:131–46 26. Moore L, Lavoie A, Lesage N, et al: Multiple imputations of 17. Wrigley J, Yoels W, Webb C, Fine P: Social and physical the Glasgow coma scale. J Trauma 2005;59:698–704 factors in the referral of people with traumatic brain inju- 27. Association for the Advancement of Automotive Medicine: ries to rehabilitation. Arch Phys Med Rehabil 1994;75: The Abbreviated Injury Scale. Des Plaines, Ill, AAAM, 1990 149–55 28. Baker S, O’Neill B, Haddon W, et al: The Injury Severity 18. Emhoff T, McCarthy M, Cushman M, et al: Functional Score: a method for describing patients with multiple inju- scoring of multi-trauma patients: who ends up where? ries and evaluating emergency care. J Trauma 1974:187–96 J Trauma 1991;31:1227–32 29. Altman D, Royston P: What do we mean by validating a 19. Wagner A, Hammon F, Grisby J, Norton H: The value of prognostic model? Stat Med 2000;19:453–473 trauma scores: predicting discharge after traumatic brain 30. Moore C: SAS Macro CVLR (Cross-Validation for Logistic injury. Am J Phys Med Rehabil 2000;79:235–42 Regression). Athens, Ga, University of Georgia, 2000 20. Chan L, Doctor J, Temkin N, et al: Discharge disposition 31. Copes W, Champion H, Sacco W, et al: The Injury Severity from acute care after traumatic brain injury: the effect of Score revisited. J Trauma 1988:69–77 insurance type. Arch Phys Med Rehabil 2001;82:1151–4 32. Mock C: Case series and trauma registries, in Rivara FP, 21. Retchin S, Brown R, Yeh S et al: Outcome of stroke patients Cummings P, Koepsell TD, Grossman DC, Maier RV (eds): in Medicare fee for services and managed care. JAMA 1997: Injury Control. Cambridge, UK, Cambridge University 119–24 Press, 2001, pp 168–82

July 2007 Predicting Discharge to Rehabilitation 573 Authors: Inaćio Teixeira da Cunha-Filho, PT, PhD Danielle Aparecida Gomes Pereira, PT Walking Andre´ Maurıćio Borges de Carvalho, MD Leilane Campedeli, PT Michelle Soares, PT Joyce de Sousa Freitas, PT RESEARCH ARTICLE Affiliations: From the Centro Universita´rio de Belo Horizonte, UNI-BH Brazil, Belo Horizonte, Brazil. The Reliability of Walking Tests in Correspondence: People with Claudication All correspondence and requests for reprints should be addressed to Inaćio Teixeira da Cunha-Filho, PT, PhD, Rua Ernani Agrıćola 50/302, ABSTRACT 30455-760 Belo Horizonte, MG Brazil. Teixeira da Cunha-Filho I, Gomes Pereira DA, Borges de Carvalho AM, Campedeli L, Soares M, de Sousa Freitas J: The reliability of walking tests in Disclosures: people with claudication. Am J Phys Med Rehabil 2007;86:574–582. This study was partially funded by Objective: To contrast the reliability of the 6-min walking test (6MW) Centro Universita´rio de Belo Horizonte. with that of the shuttle walking test (SWT) in patients with intermittent claudication attributable to peripheral obstructive arterial disease (POAD), 0894-9115/07/8607-0574/0 and to examine the relationships of the walking tests, medical outcomes, American Journal of Physical Medicine & Rehabilitation and hemodynamic variables. Copyright © 2007 by Lippincott Design: Twenty-three patients were randomly assigned to perform Williams & Wilkins both walking tests on two different occasions. Total distance walked (DW), DOI: 10.1097/PHM.0b013e31806de721 time of pain onset (PO), and time of limiting claudicating symptom (TLS) were analyzed. Results: The reliability coefficients (ICC) generated with the 6MW for DW, PO, and TLS were 0.84, 0.81, and 0.63, respectively, and the coefficients of variation (CV) across trials were 18, 31.8, and 20.3%, respectively. With the SWT, the ICCs were 0.95, 0.72, and 0.90 for each variable, respectively, whereas the CV was 10.9% for DW, 26.8% for PO, and 9.1% for TLS. There was a significant correlation between DW and double product with the SWT (r ϭ 0.51, P ϭ 0.01), but there was no correlation with the 6MW. Conclusion: Both the 6MW and the SWT are reliable walking tests for claudicant patients. However, the reliability coefficients obtained with the SWT are higher, the measurements are less variable, and there were better associations between this walking test and hemodynamic variables.

Key Words: Reliability, Walking Test, Claudication

574 Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 and pediatric dysfunctions, and its correlation to Peripheral arterial occlusive disease (PAOD) af- aerobic capacity has been considered satisfactory.9 fects 12% of the older population in the United This test, therefore, can also be used to compare States.1 In Brazil, about 11 million individuals over functional capacity before and after intervention, the age of 45 have a high probability of developing although one disadvantage is that its velocity is the disease.2 Its incidence is higher among men, determined by the patients themselves, despite the and its prevalence is two times higher among Af- supervision and encouragement from the evalua- rican descendents and eight times more common tor, who, according to the previous standardiza- among people with low levels of education. One tion, is supposed to prompt the patient during the typical symptom of PAOD is intermittent claudica- performance.8 tion (IC), which is defined as pain, pressure, or More recently, another overground walking cramping associated with walking in one or both test has been proposed to evaluate functional ca- lower limbs, affecting either distal or proximal pacity among patients with different pathologies.10 muscle groups. IC is caused by decreases in blood This is the incremental shuttle walking test (SWT), flow, leading to tissue hypoxia and, consequently, which, besides imposing a similar challenge closer reduction of the functional activities of the patient. to the patient’s usual way of walking than that The claudicating symptoms usually seem to be a observed with the 6MW, has the advantage of hav- function of the distance and/or speed of walking ing the patient’s speed controlled externally, allow- and can be relieved by rest.3 The distance walked by ing for a progressive increase of the walking effort. patients with PAOD can be decreased as much as This is a low-cost, easily administered test, and 50% compared with healthy individuals of the because of its external control of speed, it has same age, because of the claudicating symptom.4,5 better potential for revealing the patient’s func- The patient’s functional compromise depends tional capacity. However, the SWT is not overly both on the level of arterial obstruction and on used with POAD patients, and its psychometric aerobic capacity, which may be further aggravated characteristics still need to be evaluated with this by physical inactivity secondary to the claudicating population and contrasted with different protocols symptom. Therefore, the more inactive the patient concerning functional assessment.11 The investiga- becomes, the more aerobic capacity deteriorates, tion of simpler protocols that could deliver reliable leading to earlier muscle fatigue and a reduced results is necessary, especially in areas where eco- willingness to walk.5 Hence, the walking limitation nomic restrictions are present. is one of the most important hallmarks related to Therefore, the objective of this study was to functional impairment among patients with PAOD. compare the reliability of the 6MW and the SWT One of the ways to assess the severity of PAOD and to contrast their association with measures of is based on the magnitude of the patient’s func- self-perceived levels of health that are based on the tional compromise while walking. The literature Medical Outcomes Survey (MOS-SF36) and hemo- describes different protocols to assess function ac- dynamic variables. cording to the time of pain-free walking or maximal walking distance, as limited by the claudicat- METHODS ing symptoms.1,4,6,7 The use of treadmill protocols Subjects has the advantage of controlling the speed and/or Forty-two patients with PAOD were contacted inclination during the ambulation, allowing for at the Center for Treatment and Study of Periph- comparisons before and after the intervention. eral Vascular Disease from the Educational Foun- However, the reproducibility of the time or dis- dation of Belo Horizonte (UNI-BH) and the Clinical tance of pain-free walking or maximal limiting Hospital of the Federal University of Minas Gerais, claudication symptoms with these protocols are Brazil. Patients had known diagnoses of PAOD con- not yet completely satisfactory. Moreover, walking firmed either by a Doppler scan study and/or by the on a treadmill is not universally tolerated by pa- clinical presentation and classified as Fontaine tients, because it does not represent typical ambu- stage 2 of the disease. Patient selection was inde- lation. In addition, the equipment is expensive and pendent of gender, ethnicity, or age. Patients with is not available in all treatment centers, particu- an asymptomatic PAOD, nonvascular claudicating larly lower-income communities.6,7 pain, unstable angina, uncontrolled diabetes, rest- Alternative ways to evaluate functional limita- ing high blood pressure, EKG suggestive of acute tions using walking protocols are based on endur- ischemia or uncontrolled arrhythmias, a recent ance testing performed overground.8 The 6-min episode of pulmonary embolism, or any other neu- walking test (6MW) is a submaximal endurance test romuscular deficit precluding walking were ex- that calculates the distance after a subject has cluded from the study. The project was approved by walked for 6 mins. It has been used for patients the local ethics committee at the UNI-BH, and all with rheumatic, neurologic, cardiac, pulmonary, participants gave their written informed consent.

July 2007 Reliability of Walking Tests 575 Procedures To evaluate subjective perceptions of health- related quality of life, the Medical Outcomes Study The 6MW was conducted in a 60-m quadran- Short Form (SF-36) was used. It consists of 36 gular course, with ground marks every 2 m. Each items, arranged in eight domains, evaluating phys- patient was oriented to walk as quickly as possible, ical functioning, role limitations attributable to without running or jogging, for 6 min. At each physical problems, general health, bodily pain, so- minute, a verbal command was given to the patient cial functioning, role limitations attributable to by the physical therapist to encourage the perfor- emotional problems, mental health, and vitality. mance. Before beginning the testing, the patient This study grouped the first four items as the was fitted with a Polar heart rate (HR) monitor so physical health domain, and the last four items that the HR could be registered each minute within were designated the mental health domain. The the test. Before and immediately after the test, statistical analysis was conducted on the basis of blood pressure was measured. The last recorded these two domains. The patient responded to the HR and blood pressure were used for statistical SF-36 during an interview conducted by the phys- analysis. The rate of perceived exertion (RPE) was ical therapist during the first visit to the clinic.13,14 assessed by the Borg scale at the end of the test.12 If the patient needed to stop the testing because of Statistical Analysis tiredness or claudication, he or she was allowed to rest and eventually resume the test so that the total Descriptive statistics were presented as means Ϯ of 6 mins was completed. standard deviations (SD). Data were checked for nor- Two cones were set apart on the ground at a mality (Kolmogorov–Smirnov), and the paired t test distance of 10 m to establish the course where the was used to analyze differences between the test– SWT was conducted.10 The patient started the test retest procedures. When sample sizes were unequal, by walking back and forth around the cones, the independent t test was applied. The intraclass guided by a speed that had been determined previ- correlation coefficient (ICC2,1) was used to assess the 15 ously by an audio signal played on a compact disc reproducibility of the measures. The reliability of player. The initial speed was 0.50 m/sec and was the measures also was checked by the coefficient of increased by 0.17 m/sec each minute up to the end variation (CV) obtained between the two evaluations. of the test, which could last up to 12 mins. The Pearson coefficients (r) were used to verify associa- speed during the last stage was 2.37 m/sec. After tions among the variables of interest. When data the initial audio signal, the patient was instructed departed from normality, nonparametric procedures to walk to the other cone by the next sound. If the were used. An alpha value of 5% was set for statistical patient reached that cone before the sound, he or significance. she was supposed to continue to walk in place until the audio signal was again heard. After completing RESULTS one stage (1 min of walking), a different set of Of the 42 patients contacted, 12 did not join sounds was heard so that the patient would know the project because they lacked transportation, one that the speed was going to increase. At each new was excluded from the analysis for not showing up stage, the time between the audio signals would for the retest, one for uncontrolled arrhythmia, decrease and the patient had to increase his or her two because of leg pain at rest, and three because speed accordingly. The test was terminated when they were not claudicants. Of the 23 patients who the patient could not reach the next cone, for two actually participated in the study, 9 were women consecutive turns, by the time the audio signal was and 14 were men. Table 1 provides patient demo- heard. The same procedures for recording HR and graphics, medications, and associated pathologies. blood pressure with the 6MW were repeated with All patients tolerated the procedures, and there the SWT. were no medical complications or emergencies The time of pain onset and the time of limiting during the study. pain (i.e., time when the patient could no longer There was no difference between the mean walk because of the high intensity of the claudicat- distances covered by the patients between the two ing symptom) during both tests were recorded. The evaluations with the 6MW or with the SWT (Table

total distance walked also was registered for both 2). The ICC2,1 values indicate that both walking tests. To evaluate the reproducibility of the two tests are reliable, although the SWT generated a tests, they were repeated a second time the next higher coefﬁcient. The CV values for this variable week, and the order of administration was ran- were 18% with the 6MW and 10.9% with the SWT. domly assigned. There was an interval of 20 mins The mean walking distance obtained with the 6MW between the two tests, and before the actual data was statistically higher compared with the distance collection, the patients were familiarized with the obtained with the SWT, in evaluation 1 (P ϭ 0.017) procedures. and in evaluation 2 (P ϭ 0.007).

576 Cunha-Filho et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 TABLE 1 Patient characteristics (n ϭ 23) Age, yrs 64.00 ؎ 11.01 Height, cm 162.09 ؎ 7.65 Body weight, kg 68.10 ؎ 15.59

Associate Conditions n Medications n

Hypertension 21 Beta blocker 6 Diabetes 7 Diuretic 13 Emphysema 1 Digital 1 Stable angina 2 ACE inhibitor 9 Cardiac arrhythmia 2 Calcium-channel blocker 8 Congestive heart failure 1 AT-1 receptor blocker 3 Thromboangiitis obliterans 1 Central alpha-adrenergic agonist 1 Dyslipidemia 8 Vasodilator 17 Smoking 7 Antiagregating agents 17 Ex-smoking 12 Anticoagulant agents 17 Hypolipemiant 4 Oral hypoglycemiant 7 Antidepressant 3 Antiinﬂamatory 16 Nitrate 3

Only eight patients during the first evaluation evaluation. The reproducibility of this measure was ϭ and nine patients during the second evaluation high (ICC2,1 0.81) and significant, and the variabil- reached limiting claudicating pain with the 6MW, ity was 31.8%. With the SWT, three patients did not and no statistical difference was observed between report pain on evaluation 1, one patient did not report the two evaluations (independent t test). The ICC2,1 pain in either evaluation, one reported pain during of 0.63 that was evaluated with eight patients, the first evaluation, and one reported pain during the although modest, was significant (Table 3). As for second evaluation. There was no statistical difference the SWT, all patients reached limiting claudicating between the mean time of pain onset between the two pain, and there was no statistical difference be- evaluations (independent t test), and the reproduc- ϭ tween the two evaluations. The reproducibility of ibility assessed with 20 patients was high (ICC2,1 ϭ this measurement was high (ICC2,1 0.90) and 0.72) and significant. The CV for this variable with the significant (Table 3). When the time to reach lim- SWM was 26.8%. iting claudicating pain was compared between the The correlations among distance walked, time two tests (Table 4), the patients who reached this of pain onset, and time of limiting pain were also symptom during the 6MW had an earlier onset analyzed for consistency of results within each test. compared with the SWT, both in evaluation 1 (P ϭ For the 6MW, the Pearson correlation between 0.019) and evaluation 2 (P ϭ 0.016). The variability distance walked and time of pain onset was 0.41 of the measures with the 6MW was higher (CV ϭ (P ϭ 0.057), whereas the Spearman correlation 20.3%) compared with the SWT (CV ϭ 9.1%). between distance walked with time for limiting As for the time to onset of pain, only one patient symptom was 0.21 (P ϭ 0.60). There was a strong, during the second evaluation did not report pain with significant Spearman correlation between time for the 6MW. The mean time did not differ statistically pain onset and time of limiting symptom (␳ ϭ 0.92, from evaluation 1 to 2, but there was a tendency to P ϭ 0.01). With the SWT, Pearson correlations reduce the time to onset of pain during the second were 0.75 (P ϭ 0.0001) for distance walked and

TABLE 2 Distance walked, m

a b n Evaluation 1 Evaluation 2 P ICC2,1 P 6MW 23 388.38 Ϯ 126.09 386.20 Ϯ 122.54 0.88 0.84 0.00005 SWT 23 297.83 Ϯ 132.35 292.61 Ϯ 124.29 0.55 0.95 0.00005

a 6MW, 6-min walking test; SWT, shuttle walking test; ICC2,1, intraclass correlation coefficient model 2,1. Significance level between evaluation 1 and evaluation 2; b significance level of the ICC. n, number of participants.

July 2007 Reliability of Walking Tests 577 TABLE 3 Time of limiting claudicating pain, secs

a b n Evaluation 1 Evaluation 2 P ICC2,1 P 6MW 8/9* 246.56 Ϯ 59.72 236.00 Ϯ 67.29 0.74 0.63 0.03 SWT 23† 327.65 Ϯ 90.60 332.00 Ϯ 90.30 0.63 0.90 Ͻ0.0001

† * Independent t test; paired t test; 6MW, 6-min walking test; SWT, shuttle walking test; ICC2,1, intraclass correlation coefficient model 2,1. a Significance level between evaluation 1 and evaluation 2; b significance level of the ICC. n, number of participants.

time of pain onset, 0.77 (P ϭ 0.0001) for time of health components with the performance on the pain onset and time of limiting symptom, and 0.93 two walking tests. The 6MW had high, signiﬁcant (P ϭ 0.0001) for distance walked and time for correlations only when the time of pain onset and limiting symptom. the physical and mental components of the SF-36 There was no difference between the mean were analyzed. Correlations among the two com- values of HR, systolic blood pressure (SBP) or di- ponents of the SF-36 and the performance during astolic blood pressure (DBP), or RPE between eval- the SWT were present (ranging from 0.33–0.48) uations 1 and 2 in either the 6MW or the SWT, with distance walked, time of pain onset, and time although there was a tendency for these values to for limiting claudicating pain. These correlations be higher with the SWT. The SWT generated were statistically signiﬁcant only with the mental higher mean values of HR and DBP than the 6MW health component. The correlation between time

during the second evaluation (Table 5). The ICC2,1 of pain onset and the physical health component for HR was 0.61 (P Ͻ 0.0001) for the 6MW and was during the SWT was low and not significant. 0.81 (P Ͻ 0.0001) for the SWT. As for the SBP, this coefficient was 0.75 (P Ͻ 0.0001) and 0.39 (P ϭ DISCUSSION 0.03) for the 6MW and SWT, respectively. The objective of this study was to contrast the There was no correlation between the rate– reliability of the 6MW and SWT in registering the pressure product (HR ϫ SBP) with distance walked time of pain onset and time of limiting claudicating and time of limiting claudicating pain during the symptom while walking, as well as the total dis- 6MW, whereas with the SWT, these correlations tance covered by the patients with POAD. Both were r ϭ 0.51 (P ϭ 0.01) and r ϭ 0.44 (P ϭ 0.03), tests generated high reliability coefficients for dis- respectively. The rate of perceived exertion did not tance walked, although the SWT demonstrated a correlate with any of the hemodynamic variables or higher coefficient and less variability than did the with the performance during the walking tests. 6MW. The findings observed with the 6MW are Twenty-two patients completed the SF-36. One corroborated by the literature, not only in relation patient’s data were incomplete and were excluded to the distance walked but also in relation to the from analysis. The mean score for physical func- other variables analyzed. In other studies, for ex- tioning was 44.31 Ϯ 17.55; for physical role, ample, the distance walked varied from 382 Ϯ 12 to 44.32 Ϯ 38.52; for bodily pain, 55.23 Ϯ 20.99; for 433 Ϯ 11 m with patients with PAOD,14,16–19 and general health, 58.95 Ϯ 16.31; for vitality, 55.23 Ϯ the test–retest reliability was high with small vari- 17.96; for social functioning, 72.3 Ϯ 25.80; for ability (r ϭ 0.94 and CV ϭ 10.4%).8 This tendency emotional role, 45.42 Ϯ 46.63; and for mental can be observed when the 6MW is used with pa- health, 59.23 Ϯ 15.72. The items related to physi- tients with congestive heart failure (ICC Ͼ0.95, cal functioning and for physical and emotional CV Ͼ1.9%),9,20 although the distances walked are limitations had the lowest scores. Table 6 relates more variable.9 When the variability of distance the correlations between the physical and mental walked during both tests of this study was con-

TABLE 4 Time of pain onset, secs

a b n Evaluation 1 Evaluation 2 P ICC2,1 P 6MW 23/22* 154.26 Ϯ 80.29 134.45 Ϯ 68.10 0.38 0.81 Ͻ0.001 SWT 21* 235.47 Ϯ 79.09 240.00 Ϯ 93.94 0.85 0.72 Ͻ0.001

* Independent t test; 6MW, 6-min walking test; SWT, shuttle walking test; ICC2,1, intraclass correlation coefficient model 2,1. a Significance level between evaluation 1 and evaluation 2; b significance level of the ICC based on 22 paired observations for the 6MW and on 20 paired observations for the SWT. n, number of participants.

578 Cunha-Filho et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 TABLE 5 Hemodynamic variables and rates of perceived exertion (RPE) 6MW SWT

Evaluation 1 Evaluation 2 Evaluation 1 Evaluation 2 P* P**

HR, bpm 108.83 Ϯ 13.74 109.13 Ϯ 15.53 112.48 Ϯ 18.84 114.7 Ϯ 16.47 0.22 0.039 SBP, mm Hg 157.17 Ϯ 21.68 156.52 Ϯ 21.45 160.87 Ϯ 23.72 161.09 Ϯ 25.94 0.40a 0.44a DBP, mm Hg 83.48 Ϯ 11.52 79.13 Ϯ 14.11 85.22 Ϯ 10.82 83.04 Ϯ 12.95 0.52 0.05 RPE 12.83 Ϯ 2.76 11.87 Ϯ 2.94 12.57 Ϯ 2.76 11.91 Ϯ 2.97 0.60a 0.88a 6MW; 6-min walking test; SWT, shuttle walking test; SBP, systolic blood pressure; DBP, dystolic blood pressure. * Signif- icance level between 6MW and SWT at evaluation 1; ** signiﬁcance level between 6MW and SWT at evaluation 2. a Wilcoxon.

trasted, a CV of 18%, as observed with the 6MW, walked between two tests was approximately would mean that if a patient in one evaluation 515 m, the test–retest reproducibility coefficient reached a distance of 300 m, his or her distance was 0.98, and the CV was 6.9%.24 Distances covered walked in a subsequent test could vary by 54 m. with patients with limiting airflow disease varied With the SWT, a CV of 10.9% would mean that the from approximately 272 m25 to 391 m,26 and the variability in another test would be 32.7 m. reliability of the SWT in this population is also The comparison between the data obtained high.9 As for patients with claudication secondary with SWT and the literature was less fruitful, be- to spinal stenosis, the mean walking distance with cause this test has not been used universally with the SWT was 150 m, and test–retest reliability was PAOD patient populations. Manfredini et al.21 have high (ICC ϭ 0.90).27 Thus, the SWT is reliable in proposed a diverse protocol to evaluate claudicat- assessing the distance walked, not only with clau- ing patients’ performance with an incremental dicating patients secondary to PAOD but with dif- shuttle walking test to verify the reliability of speed ferent pathologies as well. of pain onset, rather than time, but they do not Fewer than half of the patients analyzed in this indicate the distance walked. Zwierska et al.22 have study (39%) had to interrupt the walk because of reported a lower mean distance walked (217 Ϯ the limiting claudicating symptom during the 17 m) after three assessments with SWT as com- 6MW. Despite this test being conducted with the pared with the present study, with high reliability therapist encouraging the patient during the per- indexes (ICC ϭ 0.88; CV close to 16%). These formance to walk as quickly as possible, few pa- reliability indexes were smaller than the ones tients reported the limiting symptom; with them, found in our study. Walker at al.11 report a median the correlation between the distance walked and maximal walking distance of 170 m (ranging from the time of the limiting claudicating symptom was 120 to 301 m) and report a CV of less than 10% weak and nonsignificant (␳ ϭ 0.21, P ϭ 0.60). This according to previous research using the SWT. suggests either that the patients had more control There are data on the SWT that are based on during the performance of the test and did not walk different groups of patients. In a study conducted quickly enough, or that the time of the testing (6 with heart-transplant patients, the mean distance mins) was not sufficiently long to allow for the walked with the SWT was close to 400 m, and the limiting symptom to manifest. The data also sug- reliability test–retest of this variable was 0.90.23 In gest that the patients might have walked at lower patients with heart failure, the average distance speeds during the 6MW, because the distances

TABLE 6 Correlation between Medical Outcomes Study Short Form (SF-36) and walking tests Time of Pain Onset, Time of Limiting Distance, m secs Symptom, secs

Component 6MW SWT 6MW SWT 6MW SWT

Physical health 0.15 0.41 0.55* 0.17 Ϫ0.31 0.33 Mental health 0.27 0.48* 0.45* 0.48* Ϫ0.25 0.46* n 22 22 21 20 8 22 6MW, 6-min walking test; SWT, shuttle walking test; n, number of participants. * P Ͻ 0.05.

July 2007 Reliability of Walking Tests 579 walked up to the limiting symptom were statisti- specific intensities for this activity. With the 6MW, cally higher than those with the SWT in both it is possible to get an average time of pain onset, evaluations. The fact that the Spearman correla- but not exactly a precise speed (i.e., walking intention was high and significant between the time of sity), because the patient has more control of his or pain onset and the time of limiting symptom dur- her gait velocity during the performance. In con- ing the 6MW indicates that during this protocol, trast, with the SWT, it is possible to determine the the times to reporting the two symptoms are intensity of walking, because the onset of pain, strongly related. As for the SWT, there was a higher which occurs in a given stage of the test, corre- consistency among those variables because all sponds to a specific walking speed. Therefore, in- Pearson correlations were strong and statistically dividualized walking speeds below, at, or higher significant. than the onset of pain, or at a percentage of the With the SWT, 100% of the patients had to limiting claudicating speed, can be prescribed dur- interrupt the test because of the limiting claudi- ing rehabilitation. cating symptom. Moreover, there was an excellent It has been suggested that the SWT generates ϭ Ͻ association (r 0.93%, P 0.0001) between the a higher cardiovascular load because it is progres- time of limiting claudicating symptom and the sive and symptom limiting compared with other distance walked during the SWT. This higher cor- walking tests with submaximal features, such as relation observed with the SWT indicates that this the 6MW.23,24 Yet, the data gathered in the present test may be more consistent when evaluating the study do not confirm this assumption. The mean claudicating patients. These data, when analyzed in HR and DBP were higher with the SWT only during conjunction with the hemodynamic variables, sug- the second evaluation, but there was no difference gest that the SWT imposed a more uniform load on with the SBP and RPE between the two trials. the patient than did the 6MW. It means that the Nonetheless, there was a tendency for HR, SBP, patients had better external control of their perfor- and DBP to be higher with the SWT in both eval- mance while performing with the SWT protocol. uations when compared with the 6MW. The mod- This study also confirms that the time to pain erate but significant correlation between the dou- onset is more prone to variability than is the total ble product with distance walked and with time of distance walked, as observed in different protocols. limiting claudicating symptom during the SWT In a study with claudicating patients during the suggest that performance with this test is more 6MW, about 69% of 64 patients reported pain onset consistent—that is, the physical performance has in a mean time of 135 Ϯ 59 secs. The test–retest stronger correlations with the cardiovascular pa- correlation of this variable was an ICC of 0.75 and rameters—but this does not imply that it is more a CV of 47%.8 Other studies on reliability generated strenuous for the patients when compared with the a high coefficient (0.95) for time of pain onset 4 6MW. during the 6MW ; when the test was conducted 25 with the constant treadmill protocol, the coeffi- Vagaggini et al., however, compared the cient was 0.55, and when the protocol was progres- 6MW with the SWT in patients with chronic ob- sive, the coefficient was 0.89.28 In one study con- structive pulmonary disease and did not report that ducted with 10 claudicating patients secondary to the latter imposed a higher cardiorespiratory over- PAOD using a diverse incremental shuttle-walking load for these patients. There was no substantial test protocol, the test–retest reproducibility of pain difference between the distances reached, or be- onset reported on speed was 0.98.20 Walker et al.11 tween the cardiorespiratory variables analyzed, in 25 also report that the reliability of time and distance either test. There was a strong, significant corre- of pain onset were high with the SWT, with a CV lation between the distances walked in both tests less than 10%. In the present study, the reliability (Spearman coefficient ϭ 0.85, P Ͻ 0.0005)—a cor- coefficient for time of pain onset during the 6MW relation superior to the one obtained in the present ϭ study (Pearson correlation ϭ 0.54, P ϭ 0.007). In (ICC2,1 0.81) was slightly higher than the coef- ϭ another study on chronic obstructive pulmonary ficient observed during the SWT (ICC2,1 0.72); however, the CV obtained with the SWT had 5% disease patients, even though the distances ob- less variability (26.8% vs. 31.8%) in this variable tained with the 6MW and the SWT were similar, compared with the 6MW. only the latter generated strong, significant corre- These walking tests are used more often to lations with cardiorespiratory parameters.26 In that evaluate walking performance before and after in- study, the authors conclude that the SWT is more tervention. However, because the main rehabilita- accurate than the 6MW for evaluating the perfor- tive procedure suggested for patients with PAOD is mance of patients with chronic obstructive pulmo- based on walking, the association between symp- nary disease.26 With heart failure patients, the peak toms and time, distance, or speed obtained from HR and RPE were higher with the 6MW than with these tests can be potentially useful in prescribing the SWT, although there was a tendency for a

580 Cunha-Filho et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 higher oxygen consumption with SWT compared (DAOP). Available at: http://www.transdoreso.org/estudo_ with the 6MW.24 cardiologia.shtml. Accessed May 6, 2006 The literature suggests that the SF-36, partic- 3. Stewart KJ, Hiatt WR, Regensteiner JG, Hirsch AT: Exercise training for claudication. N Engl J Med 2002;347:1941–51 ularly in the physical function domain, is capable 4. Gardner AW, Skinner JS, Vaughan NR, Bryant CX, Smith of detecting changes after an intervention program LK: Comparison of three progressive exercise protocols in with POAD patients.11,29 There are other studies in peripheral vascular occlusive disease. Angiology 1992;43: which the SF-36 scores did not change,15 or in 661–71 which the improvement varied from 38 to 67% 5. Green S: Haemodynamic limitations and exercise perfor- 14 mance in peripheral arterial disease. Clin Physiol Funct with the domains related to physical health. With Imaging 2002;22:81–91 patients who had sequelae from poliomyelitis, the 6. Labs KH, Nehler MR, Roessner M, Jaeger KA, Hiatt WR: distance walked during the 6MW correlated mod- Reliability of treadmill testing in peripheral arterial disease: erately (r ϭ 0.67) with the physical domain of the a comparison of a constant load with a graded load tread- SF-3630; for heart failure patients, this correlation mill protocol. Vasc Med 1999;4:239–46 ϭ 31 7. Cachovan M, Rogatti W, Woltering F, et al : Randomized with physical function was r 0.62. Bauman and reliability study evaluating constant-load and graded-exer- 32 Arthur report that the correlation between phys- cise treadmill test for intermittent claudication. Angiology ical health and the distance walked during the 1999;50:193–200 6MW was 0.57, and the SF-36 scores were lower on 8. Montgomery PS, Gardner AW: The clinical utility of a six- minute walk test in peripheral arterial occlusive disease the physical function domain. Therefore, the SF- patients. J Am Geriatr Soc 1998;46:706–11 36, even though not specific for patients with 9. Solway S, Brooks D, Lacasse Y, Thomas S: A qualitative PAOD, adds information on the general health sta- systematic overview of the measurement properties of func- tus of the claudicating patient, particularly in the tional walk tests used in the cardiorespiratory domain. physical health domain. In the present study, the Chest 2001;119:256–70 subscales based on physical functioning (44.31 Ϯ 10. Singh SJ, Morgan MDL, Scott S, Walters D, Hardman AE: Development of a shuttle walking test of disability in pa- 17.55) and role physical (44.32 Ϯ 38.56) generated tients with chronic airways obstruction. Thorax 1992;47: lower scores, except for the subscale role emo- 1019–24 tional. With the 6MW, the correlations between 11. Walker RD, Nawaz S, Wilkinson CH, Saxton JM, Pockley AG, physical and mental health were either weak and/or Wood RF: Influence of upper- and lower-limb exercise training on cardiovascular function and walking distances in nonsignificant or negative, except for the correla- patients with intermittent claudication. J Vasc Surg 2000; tion with time of pain onset. With the SWT, the 31:662–9 magnitude of the correlations of the mental and 12. Dawes HN, Barker KL, Cockburn J, Scott NRO, Wade D: physical health components were all positive and Borg’s rating of perceived exertion scales: do the verbal anchors mean the same for different clinical groups? Arch generated scores between 0.33 and 0.48, except for Phys Med Rehabil 2005;86:912–6 time of pain onset, which was 0.17. Therefore, 13. Ciconelli RM: Traduçaõ para o portugueˆs e validaçaõ do when the patterns of the correlation between the questiona´rio gene´rico de avaliaçaõ de qualidade de vida different domains of the SF-36 and the SWT are ЉMedical outcomes study 36-item short-form health survey (SF-36) [thesis]. Saõ Paulo, Brazil, Universidade Federal de analyzed, they are more favorable to the validity of Saõ Paulo, Escola Paulista de Medicina, 1997 this walking test in evaluating a patient’s general 14. Regensteiner JG, Steiner JF, Hiatt WR: Exercise training perception of health-related aspects than is the improves functional status in patients with peripheral ar- 6MW for people with POAD. terial disease. J Vasc Surg 1996;23:104–15 15. Portney GL, Watkins MP: Foundations of Clinical Re- CONCLUSION search—Applications to Practice, ed 1. Norwalk, Conn, Appleton & Lange, 1993 This study indicates that both walking tests are 16. Gardner AW, Katzel LI, Sorkin JD, et al: Exercise rehabili- reliable and useful for clinical purposes with pa- tation improves functional outcomes and peripheral circu- tients who claudicate because of PAOD. The SWD, lation in patients with intermittent claudication: a random- however, may be preferable with this population ized controlled trial. J Am Geriatr Soc 2001;49:755–62 17. Gardner AW, Montgomery PS, Flinn WR, Katzel LI: The because it has generated higher indexes of test– effect of exercise intensity on the response to exercise retest reliability and lower variability. It also has rehabilitation in patients with intermittent claudication. demonstrated more consistency among the test J Vasc Surg 2005;42:702–9 outcomes (distance and symptoms), stronger cor- 18. Atkins LM, Gardner AW: The relationship between lower relations with hemodynamic variables, and more extremity functional strength and severity of peripheral arterial disease. Angiology 2004;55:347–55 uniform association with measures of subjective 19. McDermott MM, Liu K, Greenland P, et al: Funcitonal perception of health-related quality of life (SF-36). decline in peripheral arterial disease. Associations with the ankle brachial index and leg symptons. JAMA 2004;292: REFERENCES 453–61 1. Dormandy JA, Rutherford RB: Management of peripheral 20. Kervio G, Ville NS, Leclercq C, Daubert JC, Carre´ F: Inten- arterial disease (PAD). TransAtlantic inter-society consen- sity and daily reliability of the six-minute walk test in sus (TASC) working group. J Vasc Surg 2000;31(suppl): moderate chronic heart failure patients. Arch Phys Med S5–13 Rehabil 2004;85:1513–8 2. Nascimento MR: Doença arterial obstrutiva perife´rica 21. Manfredini F, Mangolini C, Mascoli F, et al: An incremental

July 2007 Reliability of Walking Tests 581 test to identify the pain threshold speed in patients with 27. Pratt RK, Fairbank JC, Virr A: The reliability of the Shuttle intermittent claudication. Circ J 2002;66:1124–7 Walking Test, the Swiss Spinal Stenosis Questionnaire, the 22. Zwierska I, Nawaz S, Walker RD, Wood RFM, Pockley AG, Oxford Spinal Stenosis Score, and the Oswestry Disability Saxton JM: Treadmill versus shuttle walk tests of walking Index in the assessment of patients with lumbar spinal ability in intermittent claudication. Med Sci Sports Exerc stenosis. Spine 2002;27:84–91 2004;36:1835–40 28. Gardner AW, Skinner JS, Cantwell BW, Smith LK: Progres- 23. Lewis ME, Newall C, Townend JN, Hill SL, Bonser RS: sive vs single-stage treadmill tests for evaluation of claudi- Incremental shuttle walk test in the assessment of patients cation. Med Sci Sports Exerc 1991;23:402–8 for heart transplantation. Heart 2001;86:183–7 29. Coyne SK, Margolis KM, Gilchirist AK, et al: Evaluating 24. Green DJ, Watts K, Rankin S, Wong P, O’Driscoll JG: A effects of method of administration on Walking Impairment comparison of the shuttle and 6 minute walking tests with Questionnaire. J Vasc Surg 2003;38:296–304 measured peak oxygen consumption in patients with heart 30. Noonan VK, Dean E, Dallimore M: The relationship between failure. J Sci Med Sport 2001;4:292–300 self-reports and objective measures of disability in patients 25. Vagaggini B, Taccola M, Severino S, et al: Shuttle walking with late sequelae of poliomyelitis: a validation study. Arch test and 6-minute walking test induce a similar cardiore- Phys Med Rehabil 2000;81:1422–7 spiratory performance in patients recovering from an acute 31. Hamilton DM, Haennel RG: Validity and reliability of the exacerbation of chronic obstructive pulmonary disease. 6-minute walk test in a cardiac rehabilitation population. Respiration 2003;70:579–84 J Cardiopulm Rehabil 2000;20:156–64 26. Onorati P, Antonucci R, Valli G, et al: Non-invasive evalu- 32. Bauman HC, Arthur HM: Relationship between functional ation of gas exchange during a shuttle walking test vs. a exercise capacity and general quality of life in nonsurgical 6-min walking test to assess exercise tolerance in COPD patients with lower-extremity peripheral arterial disease. patients. Eur J Appl Physiol 2003;89:331–6 J Vasc Nurs 1997;15:21–8

582 Cunha-Filho et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 Authors: Michael W.C. Payne, MD, MSc David R. Williams, MD, MSc Physiology Guy Trudel, MD, MSc

Afﬁliations: From the Division of Physical Medicine & Rehabilitation (M.W.C.P., LITERATURE REVIEW G.T.), University of Ottawa, Ottawa, Canada, The Rehabilitation Centre, Ottawa, Canada (M.W.C.P.), Canadian Space Agency, Longueuil, Canada (D.R.W.), and NASA Johnson Space Space Flight Rehabilitation Center, Houston, Texas (D.R.W.).

Correspondence: All correspondence and requests for ABSTRACT reprints should be addressed to Dr. Payne MWC, Williams DR, Trudel G: Review: space-flight rehabilitation. Am J Michael Payne, The Rehabilitation Centre, University of Ottawa, Division Phys Med Rehabil 2007;86:583–591. of PM&R, 505 Smyth Road, Room The weightless environment of space imposes specific physiologic adaptations on 1200, Ottawa, K1H 8M2 Ontario, Canada. healthy astronauts. On return to Earth, these adaptations manifest as physical impairments that necessitate a period of rehabilitation. Physiologic changes result 0894-9115/07/8607-0583/0 from unloading in microgravity and highly correlate with those seen in relatively American Journal of Physical immobile terrestrial patient populations such as spinal cord, geriatric, or decon- Medicine & Rehabilitation ditioned bed-rest patients. Major postflight impairments requiring rehabilitation in- Copyright © 2007 by Lippincott tervention include orthostatic intolerance, bone demineralization, muscular atrophy, Williams & Wilkins and neurovestibular symptoms. Space agencies are preparing for extended- DOI: 10.1097/PHM.0b013e31802b8d09 duration missions, including colonization of the moon and interplanetary exploration of Mars. These longer-duration flights will result in more severe and more prolonged disability, potentially beyond the point of safe return to Earth. This paper will review and discuss existing space rehabilitation plans for major postflight impairments. Evidence-based rehabilitation interventions are imperative not only to facilitate return to Earth but also to extend the safe duration of exposure to a physiologically hostile microgravity environment. Key Words: Space Flight, Rehabilitation, Weightlessness, Disability

The weightless environment of space forces physiologic adaptation on all systems of the human body. A period of disability follows space flight and may last from days to weeks as the body readapts to full gravity. Physiologic adaptation to microgravity is dependent on exposure, with greater levels of disability anticipated on returning home from longer missions. Astronauts’ physiologic adaptations to the extreme environment of space cause impairments and disabilities that may not be fully reversed on return to Earth. To date, space medicine has focused primarily on preflight preparation and potential acute problems of space flight; however, knowledge and understanding of the consequences of weightlessness, combined with the expertise of medical rehabilitation specialists, can converge to extend the hard ceiling of space-mission duration. Disability after a space flight is dependent on three major factors: amount of time spent in space, individual adaptation to weightlessness, and efficacy of countermeasures.1 Because space agencies are currently preparing for longer- duration missions than ever before, greater levels of postflight disability are inevitable. We are currently unable to predict individual adaptation to weightlessness, and therefore we cannot preselect astronauts who adapt with relative

July 2007 Space-Flight Rehabilitation 583 ease to differing gravitational environments. Inte- ization, anemia, immunosuppression, endocrine grated countermeasures have been of great benefit alterations, and neurosensory vestibular dysfunc- in long-duration space flight but are dependent on tion. As such, physiatrists are ideally suited to compliance and not completely efficacious. Thus, a develop optimal rehabilitation strategies for re- well-designed, evidence-based, postflight rehabili- turning astronauts. tation program is needed for the readaptation of Furthermore, rehabilitation patients will profit astronauts to Earth after long-duration flights. from space research. The early emphasis of space Short-duration flights are defined as less than development was with regards to physics, engi- 30 days, encompassing typical shuttle missions of neering, and hardware, not biological sciences. 10–17 days. Most data on space physiology have Thus far, spinoffs directly benefiting rehabilitation been derived from these short-term stays in space. patients have mainly been technology transfers.6–8 Long-duration missions, including a 6-mo-average Recently, NASA identified the human as an essen- stay on the International Space Station, are dis- tial element within the systems of space flight, tinct from short-duration flights in regards to their calling for increased attention to the biology and increased emphasis on countermeasures. Eventu- physiology of human space flight. ally, there may need to be a category for ultra-long- This paper will review space-related physio- duration space travel corresponding to the time logic adaptations in three sections: (1) cardiovas- required for extended lunar colonization or inter- cular, (2) musculoskeletal, subdivided into muscle planetary travel to Mars (approximately 3 yrs) or and bone, and (3) neurovestibular. Other affected beyond. To date, the longest uninterrupted expo- systems (i.e., gastrointestinal, hematologic, renal, sure to space has been 437 days (Cosmonaut Valery etc.) are less amenable to rehabilitation and will Polyakov). The longest cumulative time in space was not be reviewed here. 747 days throughout a total of three flights (Cosmo- naut Sergei Avdeyev). Rehabilitating astronauts to CARDIOVASCULAR return to 1 g (Earth gravity) after years of space travel Cardiovascular deconditioning seen with space will prove to be the most challenging. flight includes decreases in circulating blood vol- Data on space adaptation and rehabilitation are ume, diastolic blood pressure, left ventricular sparse and have been derived mainly from limited mass, and ventricular stroke volume.9–11 These studies on astronauts and from Earth-based models alterations are dose dependent on exposure to mi- of weightlessness. The ideal study is frequently crogravity and may lead to dizziness, tachycardia, compromised by payload restrictions and limited palpitations, or reduced exercise capacity. Arrhyth- time available for the crew members who need to mias do not routinely occur during or after flight.12 act as in-flight scientists on top of other high- The most immediate and troublesome manifesta- priority tasks. Additionally, a potentially large tion for the astronaut on return to Earth is post– amount of rehabilitation data may not have been space-flight orthostatic intolerance (PSOI).13 systematically collected for publication. As a result, PSOI is caused by insufficient perfusion of the existing space-flight studies lack the statistical brain and presents as syncope or presyncope (light- power (low subject numbers), control groups, and headedness, dizziness, graying out, or systolic blinded interventions that provide the essential blood pressure below 70 mm Hg).14 First fully tenants of evidence-based practice. Nevertheless, documented after the 34-hr Mercury flight in 1963, existing studies provide the foundation for future PSOI occurs in 20–64% of returning astronauts.15 research. Some astronauts are predisposed to this postflight The bed-rest model, used to mimic human complication and do not develop resistance to PSOI weightlessness, constitutes a direct link with reha- with subsequent flights.16 Women are particularly bilitation medicine. In bed-rest studies, subjects susceptible to PSOI.17 Long-duration space flight follow a strict protocol of lying down in bed at a greatly increases the incidence of PSOI compared 6-degree head-down tilt for days to months. with that of short missions.14,16 Physiatrists provide care to people whose condi- The mechanism of PSOI is likely multifacto- tions limit their mobility or confine them to bed. rial, although it is not yet completely understood. Some of these populations, such as spinal cord Extensive reviews of this topic have been pub- injury patients,2 severely deconditioned or nutri- lished.18 Early hypotheses for contributing factors tionally deficient patients,3 and the elderly,4,5 un- were mechanical in nature and included (a) fluid dergo physiologic changes that highly correlate loss and blood-volume contraction,19 (b) muscular with those seen in astronauts. Physiatrists rou- wasting, (c) impaired cardiorespiratory perfor- tinely deal with problems similar to both space and mance,9,20 and (d) increased venous compliance.21 bed-rest models, including cardiovascular decondi- More recent hypotheses include (e) vestibular oto- tioning, fluid shifts, muscular atrophy and decon- lith organ morphological and neurological plastic- ditioning, nutritional problems, bone demineral- ity with altered cardiovascular control,22 (f) post-

584 Payne et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 flight hypoadrenergic responses or impairment of per month.35 At this rate, BMD could theoretically the adrenergic loop,17 and (g) impaired barorecep- decrease by half during a mission to Mars and may tor reflex arcs.23 Furthermore, space-flight anemia be incompatible with safe return to 1 g. The de- with short-duration24 and long-duration25 flights crease in BMD is thought to result from decreased may contribute to insufficient brain oxygenation. mechanical loading in microgravity and is often Countermeasures with oral saline fluid loading likened to an accelerated age-related osteoporosis immediately before landing initiates the readapta- seen on Earth. tion procedures.26 Lower-body negative-pressure Decreases in BMD vary between individuals devices create a vacuum around the lower body and and are site dependent.36 The greatest amount of temporarily simulate dependent blood distribution bone loss is in larger, locomotion and Earthbound seen in normal gravity. Despite encouraging early weight-bearing bones of the lower extremity, in- experimental data, lower-body negative-pressure cluding the distal tibia, femoral neck, pelvis, and devices have not yet demonstrated fully protective lumbar vertebrae. BMD remains essentially un- effects against PSOI to justify their routine use in changed in non–weight-bearing bones of the upper space. During reentry, astronauts wear advanced limbs37 and actually increases in the skull.25 Tra- crew escape suits that contain antigravity suits and becular bone density decreases earlier and by a liquid cooling garments. The antigravity suit con- greater percentage than cortical bone,37 although sists of inflatable bladders on the legs and abdo- more absolute mineral mass is lost from cortical men, and the liquid cooling garment circulates bone.38 At the cellular level, decreased mechanical thermoelectrically cooled water through a network forces on bone may directly or indirectly act to stop of tubes covering the body. These suits offer tran- the differentiation of mesenchymal stem cells into sient cardiovascular protection by supporting the osteoblasts.39–41 blood pressure and limiting increases in heart Bone mineral depletion seems only partially re- rate.27 Within minutes of return to 1 g, medical versible on return to Earth because BMD failed to rehabilitation management may include further recover to preflight values after 6 mos37 and in lim- fluid resuscitation and, rarely, the use of inotropic ited 5-yr follow-up data.42 Progressive degradation of agents. Fludrocortisone, a mineralocorticoid,28 has the skeletal system, unless arrested by countermea- been shown to improve plasma volume, but it does sures, may be a factor limiting space travel.43 It is not reduce the symptoms of PSOI, suggesting that unknown whether bone loss is slowed or whether it reduced fluid volume alone is not responsible for eventually plateaus with long-duration flights. Com- PSOI.29 Midodrine, an alpha-1 agonist, may limit plicated osteoporosis, including fractures of the hip, PSOI.30 radius, or spine, have not been documented in space- Readaptation to increased gravity is the most flight literature, but these conditions will need mon- physiologically demanding phase of space flight, in itoring with longer flights. BMD provides only a sur- part because of the immediate consequences of rogate measure for true fracture risk. PSOI. Although syncope is accommodated on re- Bone rehabilitation aims at preventing bone turn to Earth by a medical ground crew, astronauts loss and promoting osteogenesis by mechanical will not be afforded such ground support in either loading while avoiding fractures in osteopenic a planned or an emergency egress on the moon bone. The greatest osteogenic effects come from (0.16 g) or Mars (0.36 g). Hindered human perfor- repeated short-duration high-strain activities di- mance on landing can put lives and the mission at rected to a specific bone site.44 In-flight countermea- risk, because life-support systems must be config- sures for bone loss have typically included exercise on ured immediately during this critical period. Re- treadmills, cycle ergometers, and resistance exercise de- covery from PSOI, with the disappearance of hypo- vices. Russian cosmonauts also use Penguin loading tension and presyncope, is usually complete by 3 suits to produce 70% of axial static and dynamic body days17 to 2 wks31,32 after landing. Recovery of heart weight during treadmill training.26 Unfortunately, rate, blood pressure, stroke volume, left ventricular the ideal countermeasures are as yet unknown; in- mass, and cardiorespiratory responses is expected vestigation is underway to augment existing counter- to occur during rehabilitation.10,33 measures with short-term, high-impact mechanical stimuli,45 with extremely low-magnitude, high-frequency mechanical stimuli,34,46 or with regularly MUSCULOSKELETAL: BONE scheduled centrifugation.47 Perhaps the best known effect of space flight is The role of bisphosphonate medications in the loss of bone mass, which is accelerated by a space-flight osteoporosis is as yet unclear. These factor of 10 compared with Earth.34 Bone mineral medications inhibit osteoclastic resorption of bone, density (BMD) in weight-bearing bones typically but the relative contribution of increased bone decreases by approximately 1.0% per month, al- resorption vs. decreased bone formation in space though it may decrease in some sites by up to 2.0% has not yet been delineated.36 The use of bone-

July 2007 Space-Flight Rehabilitation 585 forming pharmacological agents such as parathy- Despite the use of countermeasures, maximal roid hormone, fluoride, or insulin-like growth fac- strength was reduced to 45% of preflight values tor has not been established in this population, after 180 days on Mir.62 Additionally, postflight although this is a logical research option to inves- muscles have lost explosive power, greatly exceed- tigate. Calcium and vitamin D are essential for bone ing the moderate loss of muscle bulk. Functional integrity and must be represented in the diet, both muscle losses may be attributable to changes in during flight and rehabilitation phases. Recent evi- neural activation.59 Conduction velocity in axon dence supports the role of vitamin K in inhibiting terminals is decreased with space flight,63 and neu- osteoclastic bone resorption in space flight, and vita- romuscular junctions undergo significant struc- min K should be incorporated in the diet.48 tural remodeling in postural muscles exposed to Postfracture rehabilitation of osteoporotic pa- microgravity.61 tients is aimed at mobilization and pain control by The discord between mass, force, and power using a supervised, progressive, resistive exercise may also be attributable to altered biomechanical program. Vibration platforms seem promising for properties of muscles in microgravity. Higher64 terrestrial and space-flight osteoporosis but are and lower65,66 musculotendinous stiffness has been still under early investigation.46,49 reported after space flight, depending on the measurement techniques employed. Flywheel exercise MUSCULOSKELETAL: MUSCLE resistance blunts the decrease but does not prevent Performance of activities of daily living in alterations in tendon stiffness during bed rest.66 space, as on Earth, depends on sufficient functional Impaired control of length–tension relationships muscle mass. However, the force needed to per- during contraction may contribute to functional form these activities is reduced in space. Progres- weakness and predispose to injury. sive muscular atrophy in weightlessness has been The efficacy of countermeasures for muscular consistently documented.20,50–55 In the absence of atrophy (aerobic exercises on cycle ergometers and countermeasures, muscle mass plateaus at about treadmills, resistance exercises, and leucin-rich two thirds of initial mass after approximately 270 diet) is not known because protocols have not been days.20 Muscular integrity is needed for emergency standardized and studied. Extrapolating from bed- maneuvers, high-performance duties such as ex- rest data, the ideal muscular countermeasures travehicular activities (e.g., space walks), using would include isometric and isotonic high-inten- tools, mobility, and, possibly, to limit the degree of sity impact exercises.50 Eccentric contraction PSOI upon return to Earth. The loss of muscle training with the flywheel resistance device has pre- mass, force, and power, abnormal reflex patterns, served knee-extensor mass, blunted planterflexion and increased fatigability from space flight corre- muscle loss, hypertrophied chronically unloaded sponds in many aspects to the deconditioning suf- muscle, and attenuated bone loss in bed-rest stud- fered by bed-rest patients and elderly patients on ies.67–69 Regular, rapid muscle contractions during Earth.4,56 flight may prevent explosive force deficits seen on Actual and simulated weightlessness have return to Earth. demonstrated that atrophy is greatest in postural Because countermeasures cannot yet preserve antigravity muscles.20,55,57 Loss in muscular volume muscles, astronauts require neuromuscular read- is greatest in ankle plantarflexors, followed by dorsi- aptation to gravity and currently receive physical flexors, knee extensors, knee flexors, and lower- therapy similar to a regimen for generalized decon- back muscles.58 Short-duration space flight results ditioning. Restoration of muscle mass and strength in a decrease in fiber cross-sectional area but no occurs at a rate similar to, or possibly shorter than, change in fiber number. Unlike animal models, the rate of initial atrophy.57,58 We could not find where type I fibers preferentially atrophy, humans any studies comparing the efficacy of various reha- in microgravity display atrophy of both type I and bilitation protocols applied to astronauts after ex- type II fibers.59,60 Atrophy is determined more by tended missions. muscle function than fiber composition, because One last peculiar concern is the persistent atrophy of both types is largest in predominantly postflight muscle soreness reported by astronauts slow-twitch, postural muscles and milder in pre- up to months following return to Earth.60 The exact dominantly fast-twitch, nonpostural muscles.61 cause of this soreness is unknown, but it could be This atrophy has been attributed to an imbalance related to damage in fibers preferentially atrophied between increased myofilament protein degrada- during weightlessness and, subsequently, challenged tion and decreased synthesis when exposed to mi- with eccentric loading.50,53 Rat muscle biopsies that crogravity.54 were normal in flight demonstrated extensive sarco- However, more clinically important than mere disruption and edema within hours of returning structural changes in muscle mass, decreases in to1g.70 The closest clinical corollary to this con- force and power have been measured in space.60 dition may be delayed-onset muscle soreness after

586 Payne et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 excessive athletic activities, or that reported by rehabilitation.78 Preﬂight training and onboard rehabilitation patients after a period of non–weight countermeasures of dual adapted states are being bearing.71 Nonsteroidal antiinﬂammatory medica- tested.78,79 No preventative medications are known. tions and massage may provide symptomatic re- Similar to PSOI, potentially high-risk situations lief.72 Muscle soreness may limit both the intensity can be averted when returning to Earth, but pre- of rehabilitation exercises and the use of aggressive vention might not be possible on lunar or Martian eccentric activities. soils in the absence of ground rescue personnel.

NEUROVESTIBULAR RADIATION RISKS AND Astronauts experience a number of vestibular REHABILITATION and sensory symptoms during and after flight, in- Space-flight radiation poses severe and cur- cluding postural ataxia, peripheral proprioceptive rently irreversible threats to astronauts and may deficits, gaze instability, oscillopsia, impaired spa- result in sizable risks in interplanetary missions. tial orientation, visual vection illusions, and func- Long-duration exposure may contribute to cata- tional motor planning problems.73 Comprehensive racts, immune system dysfunction, sterility, dam- testing can detect complex vestibular symptoms in age to the central nervous system, genetic mutation, all astronauts within a few hours of landing. Space and cancers.80–85 Lingering effects of radiation on an motion sickness, which has been likened to but is astronaut during the rehabilitation phase is unknown different from terrestrial motion sickness, is to date. Strategies for risk mitigation revolve around caused by exposure to provocative real or apparent shielding, because no rehabilitation interventions can motion and is estimated to affect 80–90% of all reverse radiation exposure.86 crew members when changing gravitational fields.73 Neurovestibular disturbances have been CURRENT POSTFLIGHT attributed to impaired integration of central and REHABILITATION STRATEGIES peripheral signals from various sensory systems. In 1997, NASA developed a postflight rehabil- Changing visual reference frames, otolith organ itation plan for crew members returning from unloading, and peripheral proprioceptive deficits in long-duration missions.87 This document was microgravity are believed to play key roles.74,75 based on short-duration flights and before habita- Neurovestibular disturbances are common in tion of the International Space Station. The plan astronauts after space flight as sensory systems outlines personnel, equipment, facilities, and readapt to gravity. The most common symptoms schedules for rehabilitation. Medical support is are clumsiness (reported by 69% of astronauts), overseen by the crew surgeon and may involve difficulty walking in a straight line (66%), persist- consultation with physical medicine and rehabili- ing sensation aftereffects (60%), vertigo while tation specialists as needed. Currently, interna- walking (32%), vertigo while standing (29%), nau- tional crews returning from the International sea (15%), difficulty concentrating (10%), and Space Station initiate rehabilitation efforts at the vomiting (8%).76 landing site regardless of location. North American Although symptoms are generally described as astronauts continue their rehabilitation as soon as mild or moderate in intensity,76 they may be suf- possible in Houston, Texas, and Russian cosmo- ficient to affect astronaut well-being or mission nauts return to Star City, Russia; we will review the safety. In space, impaired neuromotor performance former in detail. Astronauts have 2 hrs/day of pro- would endanger the crew during emergency egress tected rehabilitation time for the first 45 days, from an orbiter or could delay planned construc- during which time rehabilitation interruptions (for tion goals with planetary missions. These symp- appearances, interviews, etc.) must be approved by toms have not been lessened by any current coun- medical operations. The rehabilitation plan is di- termeasures.76 vided into four phases. Symptoms are typically self-limited to a few Phase 0, the landing/egress phase, incorpo- weeks in duration but can recur intermittently and rates activities from shuttle-hatch opening until unpredictably during this time. These paroxysmal arrival in Houston. This usually occurs entirely on disturbances increase the risk for falls during re- landing day (return ϩ 0 days) but may include the habilitation or in routine tasks such as driving, next day (return ϩ 1 day), depending on the land- running, or maneuvering stairs. Motor actions that ing site. Immediate measures are aimed at symp- are fully automated before flight may require tomatic treatment of orthostatic intolerance and greater cognitive awareness and visual inputs to vestibular dysfunction. achieve a similar motor outcome during readapta- Phase 1 follows until return ϩ 3 days and tion.77 Rehabilitation exercise prescription needs addresses the acute effects of readaptation. Priori- to accommodate these unpredictable limitations. ties of this phase include rest, family time, psycho- Research is underway to guide neurovestibular logical support, physical assessment, circadian

July 2007 Space-Flight Rehabilitation 587 rhythm adjustment, and injury prevention. Physi- but also for establishing colonies on Mars, where cal activities begin with stretching exercises, mas- partial gravity exists after weightless space travel. sage, and assisted walking as tolerated. Research in rehabilitation after space flight should Phase 2 lasts from return ϩ 4 days until re- seek to delineate the apparent asymmetric plastic- turn ϩ 14 days and focuses on incremental phys- ity of the human nervous system and homeostatic ical conditioning. Stretching and massage con- adaptability to maximize functional performance of tinue, and progressive strengthening is introduced astronauts during transitions between such gravi- using resistance devices. Astronauts receive cardio- tational forces. vascular fitness training with the use of standard Although the investment into countermea- treadmills and cycle ergometers. Hydrotherapy is sures both in time and equipment may not be advocated early and, based on the water depth, rewarding for short-duration flights, countermea- provides graded levels of lower-extremity loading. sures seem indispensable in long-duration flights. Fatigue is a common problem during this phase Individual exercise-countermeasure compliance and may require frequent rest periods. Astronauts may correlate with the variability seen in postflight returning from short-duration flights often feel disability, but this needs to be verified in future that they have returned to baseline on completion missions. of this phase. Phase 3 completes the rehabilitation activities Inevitably, unexpected medical events such as by day of return ϩ 45 days. The goal of this final injury, fracture, or severe medical illness (i.e., re- phase is to obtain preflight levels of fitness in quiring operative intervention) will occur in 88 preparation for return-to-flight status. Medical su- space. In low Earth orbit, astronauts may return pervision continues but is less intense. Aerobic and to Earth for medical management and rehabilita- anaerobic activities continue, with increased focus tion. This rehabilitation must proficiently address on proprioceptive and agility training. Use of a both the acute problem as well as the complica- vibration platform has recently been initiated to tions of postflight readaptation. stimulate bone formation, although its efficacy in If returning to Earth is timely or possible, as the astronaut population needs to be established.46 during interplanetary exploration, in-flight reha- The crew surgeon may extend the rehabilitation bilitation after acute injury may be needed before program as indicated. return to active flight duty. Arguably, this in-flight Each phase of this rehabilitation plan, at each rehabilitation will be even more pressing than its phase, is based on short-term mission experience. terrestrial counterpart because of the reliance of In practical application, it is highly customized to the mission on each crew member. Although re- the crew’s medical status and training preferences. dundancy among crew members is built into all Anecdotally, astronauts and therapists believe essential operations, every astronaut contributes these exercises and duration of rehabilitation are special skills to the mission; disability of a single sufficient for full return to preflight physical sta- person among a small crew could have serious tus. Only recently have standardized outcome mea- consequences on scientific experimentation, pilot- sures been collected on strength, agility, and over- ing, or, potentially, mission safety itself. Aborting all fitness before and after long-duration flights. or rescuing a mission would result in extreme Thus far, the number of astronauts returning from delays and cost. For these reasons, the rehabilita- Mir or International Space Station astronauts has tion strategy needs to be tailored based on the been too small to draw solid conclusions. These physical impairment and healing potential in mi- raw data will eventually be used to build an evi- crogravity. dence-based postflight rehabilitation program. Rehabilitation protocols must also strive to be ADAPTING TO CHANGING evidence based. Current recommendations have GRAVITATIONAL ENVIRONMENTS evolved from extrapolations of terrestrial clinical Physiologic space-flight data suggest that it is experience with patients and bed-rest subjects. Re- more difficult to return to gravity than to adapt to search of long-duration space-flight rehabilitation microgravity. This is remarkable because it means amplifies the difficulties of rehabilitation research that after millennia on Earth, human beings can seen on Earth: minimal sample sizes (only a few adapt quickly to an exceptionally different environ- people per year), variable medical and therapeutic ment. Yet, after just weeks away, humans require interventions, nonstandardized outcome measure- prolonged and extensive rehabilitation in readapt- ments, and limited availability of outcomes because ing back home. The most demanding physical re- of issues of confidentiality. Simulation studies using quirements of space travel occur when transition- Earth-based models to mimic weightlessness (bed ing between gravitational fields. This has obvious rest or water immersion) only capture some of the implications, not only when returning to Earth, physiologic adaptations of space travel.

588 Payne et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 SUMMARY 8. DiSante E: Technology transfer: From space exploration to occupational therapy. Am J Occup Ther 1978;32:171–4 Astronauts are a unique population of physi- 9. Antonutto G, di Prampero PE: Cardiovascular decondition- cally and mentally fit individuals who undergo ing in microgravity: Some possible countermeasures. Eur physiologic adaptation during weightlessness, only J Appl Physiol 2003;90:283–91 to find themselves disabled when returning to our 10. Perhonen MA, Franco F, Lane LD, et al: Cardiac atrophy normal environment. Countermeasures have re- after bed rest and spaceflight. J Appl Physiol 2001;91: 645–53 duced but not eliminated impairments to astro- 11. Martin DS, South DA, Wood ML, Bungo MW, Meck JV: nauts on landing. As the duration of space flight Comparison of echocardiographic changes after short- and continues to lengthen, rehabilitation becomes in- long-duration spaceflight. Aviat Space Environ Med 2002; creasingly important. The commonalities between 73:532–6 people with physical disabilities treated by physia- 12. Charles JB, Yelle J, Sawin CF, Charles JB, Yelle J, Sawin CF: Lessons from operational cardiovascular studies in space. trists and astronauts on reentry to gravity support Med Sci Sports Exerc 1996 28(10 suppl):S18–22 the need for a greater contribution of physiatric 13. Yates BJ, Holmes MJ, Jian BJ: Plastic changes in processing expertise to the rehabilitation of astronauts. of graviceptive signals during spaceflight potentially con- Research on the effects of weightlessness and tribute to postflight orthostatic intolerance. J Vestib Res 2003;13:395–404 the development of evidence-based rehabilitation 14. Meck JV, Reyes CJ, Perez SA, Goldberger AL, Ziegler MG: protocols are needed, especially for cases of landing Marked exacerbation of orthostatic intolerance after long- on a virgin surface with no ground crew. Efficient vs. short-duration spaceflight in veteran astronauts. Psy- space rehabilitation strategies can, in turn, benefit chosom Med 2001;63:865–73 the rehabilitation care of disabling conditions on 15. Buckey JC Jr, Lane LD, Levine BD, et al: Orthostatic intolerance after spaceflight. J Appl Physiol 1996;81:7–18 Earth. 16. Martin DS, Meck JV: Presyncopal/non-presyncopal out- Thus far, humankind has not reached a thresh- comes of post spaceflight stand tests are consistent from old limit of exposure to weightlessness beyond flight to flight. Aviat Space Environ Med 2004;75:65–7 which the body cannot readapt to Earth. Specific 17. Waters WW, Ziegler MG, Meck JV: Postspaceflight ortho- space-flight rehabilitation strategies will help fur- static hypotension occurs mostly in women and is predicted by low vascular resistance. J Appl Physiol 2002; ther extend this limit. 92:586–94 18. Convertino VA: Mechanisms of microgravity induced ortho- ACKNOWLEDGMENTS static intolerance: Implications for effective countermeasures. J Gravit Physiol 2002;9:1–13 The authors wish to thank Natalie Hirsh and 19. Leach CS, Alfrey CP, Suki WN, et al: Regulation of body Patrick Sullivan at the Canadian Space Agency’s fluid compartments during short-term spaceflight. J Appl Astronaut Office, Longueuil, Quebec, and Dr. Lou Physiol 1996;81:105–16 Moreno, MD, at the Kennedy Space Center, Cape 20. diPrampero PE, Narici MV: Muscles in microgravity: From Canaveral, Florida, for assistance with reference fibres to human motion. J Biomech 2003;36:403–12 21. Tyberg JV, Hamilton DR: Orthostatic hypotension and the materials. Dr. Payne received funding from the role of changes in venous capacitance. Med Sci Sports Exerc Canadian Space Agency’s Operational Space Medi- 1996;28(10 suppl):S29–31 cine Group to participate in an Aerospace Medicine 22. Yates BJ, Kerman IA: Post-spaceflight orthostatic intolerance: Elective at the Kennedy Space Center provided by Possible relationship to microgravity-induced plasticity in the the NASA-KSC Biomedical Office. vestibular system. Brain Res Brain Res Rev 1998;28:73–82 23. Meck JV, Waters WW, Ziegler MG, et al: Mechanisms of postspaceflight orthostatic hypotension: low alpha1-adren- REFERENCES ergic receptor responses before flight and central auto- 1. Kozlovskaya IB: Countermeasures for long-term space nomic dysregulation postflight. Am J Physiol Heart Circ flights. Lessons learned from the Russian space program. Physiol 2004;286:H1486–95 J Gravit Physiol 2002;9:313–7 24. Alfrey CP, Udden MM, Leach-Huntoon C, Driscoll T, Pickett MH: Control of red blood cell mass in spaceflight. J Appl 2. Edgerton VR, Roy RR, Hodgson JA, et al: How the science Physiol 1996;81:98–104 and engineering of spaceflight contribute to understanding the plasticity of spinal cord injury. Acta Astronaut 2000;47: 25. Grigoriev AI, Bugrov SA, Bogomolov VV, et al: Main medical 51–62 results of extended flights on space station Mir in 1986– 1990. Acta Astronaut 1993;29:581–5 3. Maillet A, Beaufrere B, Di Nardo P, Elia M, Pichard C: Weightlessness as an accelerated model of nutritional dis- 26. Kozlovskaya IB, Grigoriev AI: Russian system of counter- turbances. Curr Opin Clin Nutr Metab Care 2001;4:301–6 measures on board of the International Space Station (ISS): The first results. Acta Astronaut 2004;55:233–7 4. Biolo G, Heer M, Narici M, Strollo F: Microgravity as a model of ageing. Curr Opin Clin Nutr Metab Care 2003;6: 27. Perez SA, Charles JB, Fortner GW, Hurst V, Meck JV: 31–40 Cardiovascular effects of anti-G suit and cooling garment during space shuttle re-entry and landing. Aviat Space 5. Wang E: Age-dependent atrophy and microgravity travel: Environ Med 2003;74:753–7 What do they have in common?. FASEB J 1999;13(suppl): 28. Shi SJ, South DA, Meck JV: Fludrocortisone does not pre- S167–74 vent orthostatic hypotension in astronauts after spaceflight. 6. Allen JR: Technology advances the frontiers of rehabilita- Aviat Space Environ Med 2004;75:235–9 tion. J Rehabil 1974;40:14–21 29. Waters WW, Platts SH, Mitchell BM, Whitson PA, Meck JV: 7. Helberg L, Chiaruttini H, Hartwig Q: Space research and Plasma volume restoration with salt tablets and water after the disabled. Rehabil Rec 1967;8:1–3 bed rest prevents orthostatic hypotension and changes in

July 2007 Space-Flight Rehabilitation 589 supine hemodynamic and endocrine variables. Am J Physiol of a low level mechanical stress device in a clinical osteo- Heart Circ Physiol 2005;288:H839–47 porosis study. Osteoporos Int 2004;15:918–26 30. Platts SH, Ziegler MG, Waters WW, Mitchell BM, Meck JV: 50. Fitts RH, Riley DR, Widrick JJ: Physiology of a microgravity Midodrine prescribed to improve recurrent post-spaceflight environment invited review: Microgravity and skeletal mus- orthostatic hypotension. Aviat Space Environ Med 2004;75: cle. J Appl Physiol 2000;89:823–39 554–6 51. Akima H, Kawakami Y, Kubo K, et al: Effect of short- 31. Vasilyeva TD, Bogomolov VV: Medical rehabilitation follow- duration spaceflight on thigh and leg muscle volume. Med ing long-term space missions. Acta Astronaut 1991;23: Sci Sports Exerc 2000;32:1743–7 153–6 52. Riley DA: Is skeletal muscle ready for long-term spaceflight 32. Cooke WH, Ames JEIV, Crossman AA, et al: Nine months in and return to gravity?. Adv Space Biol Med 1999;7:31–48 space: Effects on human autonomic cardiovascular regula- 53. Riley DA, Bain JL, Thompson JL, et al: Decreased thin tion. J Appl Physiol 2000;89:1039–45 filament density and length in human atrophic soleus mus- 33. Trappe T, Trappe S, Lee G, Widrick J, Fitts R, Costill D: cle fibers after spaceflight. J Appl Physiol 2000;88:567–72 Cardiorespiratory responses to physical work during and 54. Vandenburgh H, Chromiak J, Shansky J, Del Tatto M, Le- following 17 days of bed rest and spaceflight. J Appl Physiol maire J: Space travel directly induces skeletal muscle atro- 2006;100:951–7 phy. FASEB J 1999;13:1031–8 34. Rubin C, Xu G, Judex S: The anabolic activity of bone tissue, 55. LeBlanc A, Rowe R, Schneider V, Evans H, Hedrick T: suppressed by disuse, is normalized by brief exposure to Regional muscle loss after short duration spaceflight. Aviat extremely low-magnitude mechanical stimuli. FASEB J Space Environ Med 1995;66:1151–4 2001;15:2225–9 56. Adams GR, Caiozzo VJ, Baldwin KM: Skeletal muscle 35. LeBlanc A, Schneider V, Shackelford L, et al: Bone mineral unweighting: spaceflight and ground-based models. J Appl and lean tissue loss after long duration space flight. J Mus- Physiol 2003;95:2185–201 culoskelet Neuronal Interact 2000;1:157–60 57. Tesch PA, Berg HE, Bring D, Evans HJ, LeBlanc AD: Effects 36. Turner RT: Invited review: What do we know about the of 17-day spaceflight on knee extensor muscle function and effects of spaceflight on bone? J Appl Physiol 2000;89:840–7 size. Eur J Appl Physiol 2005;93:463–8 37. Vico L, Collet P, Guignandon A, et al: Effects of long-term 58. LeBlanc AD, Schneider VS, Evans HJ, Pientok C, Rowe R, microgravity exposure on cancellous and cortical weight- Spector E: Regional changes in muscle mass following 17 bearing bones of cosmonauts. Lancet 2000;355:1607–11 weeks of bed rest. J Appl Physiol 1992;73:2172–8 38. Lang T, LeBlanc A, Evans H, Lu Y, Genant H, Yu A: Cortical 59. Berg HE, Larsson L, Tesch PA: Lower limb skeletal muscle and trabecular bone mineral loss from the spine and hip in function after 6 wk of bed rest. J Appl Physiol 1997;82: long-duration spaceflight. J Bone Miner Res 2004;19: 182–8 1006–12 60. Fitts RH, Riley DR, Widrick JJ: Functional and structural 39. McBeath R, Pirone DM, Nelson CM, Bhadriraju K, Chen CS: adaptations of skeletal muscle to microgravity. J Exp Biol Cell shape, cytoskeletal tension, and RhoA regulate stem 2001;204:3201–8 cell lineage commitment. Dev Cell 2004;6:483–95 61. Deschenes MR, Wilson MH, Kraemer WJ: Neuromuscular 40. Colleran PN, Wilkerson MK, Bloomfield SA, Suva LJ, adaptations to spaceflight are specific to postural muscles. Turner RT, Delp MD: Alterations in skeletal perfusion with Muscle Nerve 2005;31:468–74 simulated microgravity: a possible mechanism for bone remodeling. J Appl Physiol 2000;89:1046–54 62. Antonutto G, Capelli C, Girardis M, Zamparo P, di Prampero PE: Effects of microgravity on maximal power of lower 41. Ontiveros C, Irwin R, Wiseman RW, McCabe LR: Hypoxia limbs during very short efforts in humans. J Appl Physiol suppresses runx2 independent of modeled microgravity. 1999;86:85–92 J Cell Physiol 2004;200:169–76 63. Ruegg DG, Kakebeeke TH, Gabriel JP, Bennefeld M: Con- 42. Tilton FE, Degioanni JJ, Schneider VS: Long-term fol- duction velocity of nerve and muscle fiber action potentials low-up of Skylab bone demineralization. Aviat Space Envi- after a space mission or a bed rest. Clin Neurophysiol ron Med 1980;51:1209–13 2003;114:86–93 43. Holick MF: Microgravity-induced bone loss–will it limit 64. Lambertz D, Perot C, Kaspranski R, Goubel F: Effects of human space exploration? Lancet 2000;355:1569–70 long-term spaceflight on mechanical properties of muscles 44. Lanyon LE: Using functional loading to influence bone in humans. J Appl Physiol 2001;90:179–88 mass and architecture: Objectives, mechanisms, and rela- 65. Goubel F: Changes in mechanical properties of human tionship with estrogen of the mechanically adaptive process muscle as a result of spaceflight. Int J Sports Med 1997; in bone. Bone 1996;18(1 suppl):37S–43S 18(suppl 4):S285–7 45. Goodship AE, Cunningham JL, Oganov V, Darling J, Miles 66. Reeves ND, Maganaris CN, Ferretti G, Narici MV: Influence AW, Owen GW: Bone loss during long term space flight is of 90-day simulated microgravity on human tendon me- prevented by the application of a short term impulsive chanical properties and the effect of resistive countermea- mechanical stimulus. Acta Astronaut 1998;43:65–75 sures. J Appl Physiol 2005;98:2278–86 46. Rubin C, Judex S, Muir J, Lee B, Adler B, Qin Y-X: Retention 67. Alkner BA, Tesch PA: Knee extensor and plantar flexor of skeletal musculature and postural status with a non- muscle size and function following 90 days of bed rest with invasive, extremely low level mechanical signal: a ground- or without resistance exercise. Eur J Appl Physiol 2004;93: based evaluation of efficacy. Abstract presented at Universi- 294–305 ties’ Space Research Association Bioastronautics Investigators’ 68. Alkner BA, Tesch PA: Efficacy of a gravity-independent Meeting, Galveston, TX, 2005 resistance exercise device as a countermeasure to muscle 47. Clement G, Pavy-Le Traon A: Centrifugation as a counter- atrophy during 29-day bed rest. Acta Physiol Scand 2004; measure during actual and simulated microgravity: A re- 181:345–57 view. Eur J Appl Physiol 2004;92:235–48 69. Tesch PA, Trieschmann JT, Ekberg A: Hypertrophy of 48. Heer M: Nutritional interventions related to bone turnover chronically unloaded muscle subjected to resistance exer- in European space missions and simulation models. Nutri- cise. J Appl Physiol 2004;96:1451–8 tion 2002;18:853–6 70. Riley DA, Ellis S, Slocum GR, et al: In-flight and postflight 49. Hannan MT, Cheng DM, Green E, Swift C, Rubin CT, Kiel changes in skeletal muscles of SLS-1 and SLS-2 spaceflown DP: Establishing the compliance in elderly women for use rats. J Appl Physiol 1996;81:133–44

590 Payne et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 71. MacIntyre DL, Eng JJ, Allen TJ: Recovery of lower limb seur-Clausen P, Pavy-LeTraon A: Effects of cosmonaut ves- function following 6 weeks of non-weight bearing. Acta tibular training on vestibular function prior to spaceflight. Astronaut 2005;56:792–800 Eur J Appl Physiol 2001;85:539–45 72. Cheung K, Hume P, Maxwell L: Delayed onset muscle 80. NASA, Charles JB. NASA Critical Path Roadmap. National soreness: Treatment strategies and performance factors. Aeronautics and Space Administration. Available at: http:// Sports Med 2003;33:145–64 bioastroroadmap.nasa.gov/index.jsp 73. Reschke MF, Bloomberg JJ, Harm DL, Paloski WH, Layne C, 81. Cucinotta FA, Schimmerling W, Wilson JW, et al: Space McDonald V: Posture, locomotion, spatial orientation, and radiation cancer risks and uncertainties for Mars missions. motion sickness as a function of space flight. Brain Res Radiat Res 2001;156:682–8 Brain Res Rev 1998;28:102–17 82. Cucinotta FA, Manuel FK, Jones J, et al: Space radiation and 74. Lackner JR, DiZio P: Motor function in microgravity: Move- cataracts in astronauts. Radiat Res 2001;156:460–6 ment in weightlessness. Curr Opin Neurobiol 1996;6: 83. Rastegar N, Eckart P, Mertz M: Radiation-induced cataract 744–50 in astronauts and cosmonauts. Graefes Arch Clin Exp Oph- 75. McCall GE, Goulet C, Boorman GI, Roy RR, Edgerton VR: thalmol 2002;240:543–7 Flexor bias of joint position in humans during spaceflight. 84. Vazquez ME: Neurobiological problems in long-term deep Exp Brain Res 2003;152:87–94 space flights. Adv Space Res 1998;22:171–83 76. Bacal K, Billica R, Bishop S: Neurovestibular symptoms 85. Dicello JF: The impact of the new biology on radiation risks following space flight. J Vestib Res 2003;3:93–102 in space. Health Phys 2003;85:94–102 77. Bloomberg JJ, Mulavara AP: Changes in walking strategies 86. Wilson JW, Shinn JL, Tripathi RK, et al: Issues in deep space after spaceflight. IEEE Eng Med Biol Mag 2003;22:58–62 radiation protection. Acta Astronaut 2001;49:289–312 78. Cohen HS: Update on the status of rehabilitative counter- 87. NASA. Postflight Rehabilitation Plan. Document JSC measures to ameliorate the effects of long-duration expo- 27050. Johnson Space Center, NASA, 1997 sure to microgravity on vestibular and sensorimotor func- 88. Billica RD, Simmons SC, Mathes KL, et al: Perception of the tion. J Vestib Res 2003;13:405–9 medical risk of spaceflight. Aviat Space Environ Med 1996; 79. Clement G, Deguine O, Parant M, Costes-Salon MC, Vas- 67:467–73

July 2007 Space-Flight Rehabilitation 591 Authors: Lois D. Hedman, PT, MS Jane E. Sullivan, PT, DHS Stroke Marjorie J. Hilliard, PT, MS Deborah M. Brown, DPT

Afﬁliations: From the Department of Physical CASE REPORT Therapy and Human Movement Sciences, Northwestern University Feinberg School of Medicine, Chicago, Illinois. Neuromuscular Electrical Correspondence: Stimulation During Task-Oriented All correspondence and requests for reprints should be addressed to Lois Exercise Improves Arm Function for D. Hedman, PT, MS, Department of Physical Therapy and Human an Individual with Proximal Arm Movement Sciences, Northwestern University Feinberg School of Medicine, 645 N. Michigan Avenue Dysfunction After Stroke Suite 1100, Chicago, IL 60611.

Disclosures: ABSTRACT This case report was presented as a poster at the Combined Sections Hedman LD, Sullivan JE, Hilliard MJ, Brown DM: Neuromuscular electrical Meeting of the American Physical stimulation during task-oriented exercise improves proximal arm dysfunction after Therapy Association in San Antonio, stroke. Am J Phys Med Rehabil 2007;86:592–596. TX, February, 2001 This case report examined the effectiveness of a home program using neuro- Rehabilicare, Inc. (New Brighton, muscular electrical stimulation (NMES) during voluntary task-oriented exercise to MN) supplied the muscle stimulator used in this study. Empi, Inc., (St. achieve functional and impairment improvements for an individual with primarily Paul, MN) supplied the electrodes proximal arm paresis after a stroke. The subject initially achieved a Fugl–Meyer used in this study. Assessment (FMA) score of 58/66, but she reported minimal functional use of her involved, dominant arm. The 6-wk intervention consisted of NMES-assisted 0894-9115/07/8607-0592/0 task practice involving repetitive reaching for and manipulation of small objects American Journal of Physical for three daily 15-min sessions. The subject applied NMES to the deltoid and Medicine & Rehabilitation Copyright © 2007 by Lippincott triceps brachii muscles to augment shoulder flexion and abduction and elbow Williams & Wilkins extension during task practice. Outcome measures included the FMA, the Action Research Arm Test (ARAT), and the Motor Activity Log Quality of Movement DOI: 10.1097/PHM.0b013e31806dc0ce subscale (MAL-QOM). The FMA remained unchanged, but the ARAT and MAL-QOM showed improvements, from the beginning to the conclusion of the intervention, that were maintained at 6-wk follow-up. Key Words: Stroke, Exercise, Physical Therapy, Electrical Stimulation

Half of the people with poststroke hemiparesis have nonfunctional upper extremities (UE) at 4 yrs after onset.1 These ﬁndings are consistent with the expected pattern of greater distal than proximal weakness in paretic UEs2 that results in inability to use the hand in functional tasks. However, deviations from this pattern have been demonstrated, highlighting the presence of signiﬁcant proximal weakness in the UE after stroke3–5 and showing that shoulder muscle strength was as strongly associated with UE function as was grip strength.3 The intensity or amount of exercise therapy time seems to be a critical factor in interventions shown to improve UE function after stroke.6 Neuromus- cular electrical stimulation (NMES) can be used to produce or augment contractions in paretic muscle. This intervention can be used to assist individuals

592 Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 to exercise successfully when nonstimulated exer- participation in the study but was blinded to the cise is not feasible. To date, positive outcomes specific details of the intervention. using electrical stimulation have been reported pri- An NMES-assisted practice task was designed marily in individuals with distal arm dysfunction to incorporate elbow extension with shoulder flex- after stroke.7–13 The purpose of this case report is ion and abduction movements, and to make use of to describe the functional- and impairment-level the subject’s voluntary hand function. The task outcomes of a home program of surface NMES- required her to lift a variety of objects that neces- assisted, task-specific exercise for an individual sitated different handgrips (paper plate, a pen, and with proximal UE dysfunction after stroke. an empty 12-ounce aluminum can) from a table top to a shelf 7 inches above. The shelf was placed in front and toward her involved side, to require CASE REPORT her to abduct and flex her shoulder as she extended The subject was a 43-yr-old, right-handed her elbow. She was instructed to lift the objects in woman who had experienced a stroke 5½ yrs before random order, as quickly as possible, to and from the current intervention. Magnetic resonance im- the shelf. A Rehabilicare NMIII stimulator (Reha- aging performed after the stroke showed a left- bilicare, Inc., New Brighton, MN) and reusable, sided intracerebral hematoma in the occipitopari- self-adhering, 2 ϫ 4-inch pregelled electrodes etal region with midline shift. The hematoma was (Empi Inc., St. Paul, MN) provided stimulation. evacuated, and an arteriovenous malformation of Two stimulation channels were used, one on the the left middle cerebral and left anterior commu- deltoid muscle and one on the triceps brachii mus- nicating arteries was surgically corrected. The sub- cles. A symmetrical biphasic waveform with a stim- ject participated in inpatient and outpatient phys- ulus phase duration of 300 ␮s and a frequency of 35 ical therapy for 10 mos after her stroke. None of Hz was used. Ramp and fall times were each 3 secs. her therapy included electrical stimulation. She Three aspects of the application of intervention was not participating in any formal therapy at the were tailored to enhance the subject’s voluntary time of the present intervention. Written informed effort. First, the subject was instructed to adjust consent was obtained from the subject before par- the stimulus amplitude for each session only as ticipation, in accordance with the university’s of- high as was necessary to allow her to successfully fice for protection of human subjects. complete the task. Second, the subject used a hand The subject’s paretic (right) UE movements switch in her nonparetic hand to activate stimula- were slow. She demonstrated isolated wrist and tion during the portions of the task for which she hand movements but had difficulty combining required assistance. Third, she was instructed to shoulder flexion or abduction with elbow exten- move as quickly as possible as the amplitude sion. She reported that she used her nonparetic UE ramped up for 3 secs. This ramp time provided her preferentially to meet her responsibilities at home with gradually increasing assistance as the move- because doing so required less effort and she could ment became more difficult because of muscle move more quickly. The subject’s goals were to use length and the effects of gravity. The ramping am- her paretic UE more effectively during household plitude was provided to increase the ease and speed and child-care tasks, to return to work as an office of movement beyond what was possible without manager, and to resume driving. stimulation. The subject was instructed to practice Three outcome measures were administered the task repetitively during three 15-min sessions before intervention, midintervention, after inter- per day, 7 days/wk. vention, and at 6-wk follow-up. The Action Re- As the subject’s performance improved, the search Arm Test (ARAT)14 was used to measure UE practice task was modified to make it more chal- function. This test assesses both UEs during per- lenging. During the second half of the 6-wk inter- formance of 19 tasks involving grasp, grip, pinch, vention phase, the practice task involved moving and gross movements. The UE subscale of the objects between the table top and two shelves 12 Fugl–Meyer Assessment (FMA)15 was used to mea- inches above the table, one toward each side; this sure control of movement. The FMA consists of 33 increased the excursion of shoulder movement re- items examining both UE movements, reflex activ- quired. New objects were also added, including a ity, coordination and speed. The Motor Activity Log heavier plastic plate, a variety of coins, and a glass Quality of Movement subscale (MAL-QOM)16 was filled with water. The inclusion of these objects was used to measure the subject’s perception of the intended to vary the type of grip and arm position quality of movement of the paretic arm during the and to increase the amount of force required. Lift- performance of 30 activities of daily living (ADL). ing the filled water glass encouraged accuracy and During a structured interview, the subject rates precision to avoid spillage. perception of the paretic arm from no use to nor- The subject completed the 6-wk home-based mal use. The tester was aware of the subject’s intervention. She was scheduled for a follow-up

July 2007 Neuromuscular Electrical Stimulation 593 TABLE 1 Results of outcome measures testing Midintervention Outcome Measure Preintervention (3 wks) Postintervention 6-wk Follow-up

FMA score (maximum score 58/66 55/66 58/66 59/66 of 66), involved/uninvolved UE ARAT score (maximum score 38/57 40/57 45/57 45/57 of 57), involved/uninvolved UE MAL-QOM mean (SD) score 0.98 (1.42) 2.09 (1.34) 2.78 (0.96) 2.97 (1.1) (maximum score of 5) FMA, Fugl–Meyer Assessment; UE, upper extremity; ARAT, Action Research Arm Test; MAL-QOM, Motor Activity Log Quality of Movement subscale.

reexamination in 6 wks. She was encouraged to use up. No overall changes were seen in her movement her paretic UE as much as possible, but no other control as measured by the FMA. instructions or exercises were given. Several aspects of the intervention may have She used a logbook to record that she prac- contributed to the gains. The subject was able to ticed 92 of the 126 possible sessions, averaging 2.19 completely perform the training tasks only when of the recommended three daily sessions. The stimulation supplemented her voluntary effort. mean session length was 16.75 mins. She did not The success associated with NMES-assisted prac- report any difficulty with electrode positioning or tice may have contributed to her high degree of discomfort during stimulation. The subject’s FMA compliance with the training program and enabled score did not show any change from the beginning sufficient practice for functional improvement. The to the end of the intervention or at 6-wk follow-up. subject’s voluntary participation was also thought The total score on the ARAT increased at midinter- to be a key feature of the training. A trend toward vention and again at postintervention, with the functional improvement has been found in studies increases maintained at follow-up. Results on the in which participants were instructed to attempt MAL-QOM scale indicated that the subject per- voluntary contractions with stimulation,17 as well ceived an improved ability to use her paretic arm at as in studies using electromyography-triggered midintervention and after the intervention, with NMES,18 which requires voluntary contractions. In additional improvement noted at the 6-wk fol- addition, the training in this case report was struc- low-up (Table 1). tured to provide practice that was both task-spe- The subject was contacted 1 yr after the ter- cific and random. Task-specific practice has been mination of the intervention. At that time, she was shown to contribute to enhanced functional car- working in an office 3 days/wk and volunteering at ryover,19 especially when manipulation of objects is her children’s school 2 days/wk. She had obtained included.20,21 Random practice, achieved in this her driver permit and had begun driver training. study through different objects and multiple tra- She felt she had maintained the gains made during jectories, has been reported to be significantly bet- the intervention and that these improvements had ter than massed practice for the retention of arm helped her make changes in her life. function after stroke.22 Two aspects of the intervention were targeted DISCUSSION to address the subject’s specific needs. She had At 4 yrs after stroke onset, 67% of persons with identified the slow speed of her paretic UE as a UE hemiparesis consider nonuse or disuse of the barrier to using this extremity functionally. It was paretic upper extremity to be a major problem. believed that training at higher movement speeds This is true even for some individuals with good might maximize carryover to actual use of her UE motor function.1 The subject in this case report in daily life. She was, therefore, instructed to move noted that she was not able to move her paretic, as quickly as possible. In addition, it was antici- dominant (before stroke) UE fast enough to meet pated that NMES would recruit type II motor her functional needs, even though she had good units23 that the subject was not able to recruit hand and wrist function. After a 6-wk, home-based, without stimulation because she had been moving NMES-assisted training program, the subject dem- at a slow self-selected speed for several years. Move- onstrated improvements in a functional perfor- ment speed was one of the items on the FMA that mance measure (ARAT) and a self-report of her did show improvement from the beginning to the perception of the quality of movement of her pa- end of the intervention, although the improvement retic UE (MAL-QOM) that were retained at follow- was not retained at follow-up. Perhaps the improve-

594 Hedman et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 ment was not robust enough to be carried over into duction, also improved from postintervention to her daily life once the intervention stopped. follow-up. No change was seen in the items shoul- The subject also demonstrated problems with der flexion from 90 to 180 degrees and from hand proximal control at the shoulder, especially when to sacrum. The FMA score gains were offset by a trying to simultaneously extend the elbow and flex score decrease that reflected hyperactivity in UE or abduct the shoulder. Abnormal coupling of el- deep-tendon reflexes at midintervention that per- bow flexion and shoulder abduction torques has sisted through follow-up. Recent reports in the been shown to limit arm excursion in persons with literature have shown variable effects of electrical poststroke hemiparesis.24 Keller et al.25 have dem- stimulation on muscle tone after stroke.26,27 It is onstrated that electrically induced contraction of not clear why the reflex activity increased in asso- triceps brachii has been used successfully to help ciation with the electrically stimulated training in individuals overcome these abnormal torque pat- this case, or what the relationship of this increase terns in the UE after stroke. This subject’s ten- is to the subject’s ability to move her paretic UE. dency to flex her elbow as she elevated her shoulder Perhaps the lack of improvement seen in the FMA was addressed by applying NMES to the shoulder scores may also reflect the relatively low sensitivity flexors and abductors as well as elbow extensors of the FMA to detect clinically meaningful change simultaneously within a task that required these in chronic hemiparesis.28 motions. Repetitive practice of NMES-assisted an- Selecting appropriate outcome measures to tigravity shoulder elevation was intended to in- detect meaningful clinical change is challenging. crease the proximal torque-producing capabilities Concerns about the validity of the Motor Activity required to move the entire limb and to stabilize Log have been raised in the literature.29 However, the distal extremity for function.3 the validity of the MAL-QOM has been supported by Interestingly, there was a trend that the out- significant correlations between pretreatment to come measure items that required significant prox- posttreatment change scores on the MAL-QOM and imal strength did not improve, whereas items that accelerometer readings. The internal consistency required more modest proximal UE strength or and responsiveness of the MAL-QOM were also distal control did improve. On the MAL-QOM, the supported.30 These findings support the use of the subject reported that items such as opening the MAL-QOM as a measure of real-world arm use in refrigerator, turning a door knob, and using a key conjunction with measures of functional perfor- remained most difficult, and she reported the most mance and motor control. dramatic improvements in combing her hair, car- rying objects from place to place, and donning and CONCLUSION doffing shoes and socks. Similarly, items on the This case report demonstrates the potential for ARAT, such as grasping and lifting blocks and balls NMES to enable task-specific practice to improve to shoulder level, did not improve, but manipulat- upper-extremity function in individuals with pri- ing a marble and washer/bolt did improve. Inter- marily proximal motor impairments after stroke. estingly, FMA items involving shoulder flexion and Many aspects of the training may have contributed shoulder abduction to 90 degrees did improve from to the subject’s gains, but the intensity of the postintervention to follow-up. It may be that the training may not have been sufficient to completely intensity of the training was sufficient to improve address the subject’s proximal UE paresis. The FMA her less impaired distal function but insufficient to may not be responsive enough to detect clinical strengthen her more paretic proximal musculature change in chronic hemiparesis. This case provides to participate effectively in function. These findings the framework for designing a clinical trial to com- are consistent with previous findings of greater pare NMES plus task-specific training against task- gains associated with higher pretreatment motor specific training without stimulation. capacity.10,26 The improvements in function and in the sub- REFERENCES ject’s perception of her paretic UE participation in 1. Broeks JG, Lankhorst GJ, Rumping K, Prevo AJ: The long- ADLs notwithstanding, there was no change in her term outcome of arm function after stroke: results of a follow-up study. Disability Rehabil 1999;21:357–64 overall FMA score. The test items whose scores did 2. Colebatch JG, Gandevia SC: The distribution of muscular improve seemed to be more related to the task weakness in upper motor neuron lesions affecting the arm. practice than were those items whose scores did Brain 1989;112:749–63 not change. For example, items pronation/supina- 3. Mercier C, Bourbonnais D: Relative shoulder flexor and tion with elbow extension and movement speed handgrip strength is related to upper limb function after stroke. Clin Rehabil 2004;18:215–21 improved from the beginning to the end of the 4. Bohannon RW, Smith MB: Assessment of strength deficits intervention, although the improvements were not in eight paretic upper extremity muscle groups of stroke retained at follow-up. The items shoulder abduc- patients with hemiplegia. Phys Ther 1987;67:522–5 tion and flexion to 90 degrees, and wrist circum- 5. Andrews AW, Bohannon RW: Distribution of muscle

July 2007 Neuromuscular Electrical Stimulation 595 strength impairments following stroke. Clin Rehabil 2000; outcome in studies using electrical stimulation to improve 14:79–87 motor control of the upper extremity in stroke. J Rehabil 6. Kwakkel G, van Peppen R, Wagenaar RC, et al: Effects of Med 2005;37:65–74 augmented exercise therapy time after stroke: a meta-anal- 19. Winstein CJ, Rose DK, Tan SM, Lewthwaite R, Chui HC, ysis. Stroke 2004;35:2529–39 Azen SP: A randomized controlled comparison of upper- 7. Gritsenko V, Prochazka A: A functional electric stimulation- extremity rehabilitation strategies in acute stroke: a pilot assisted exercise therapy system for hemiplegic hand func- study of immediate and long-term outcomes. Arch Phys tion. Arch Phys Med Rehabil 2004;85:881–5 Med Rehabil 2004;85:620–8 8. Kimberley TJ, Lewis SM, Auerbach EJ, Dorsey LL, Lojovich 20. Trombly CA, Wu CY: Effect of rehabilitation tasks on orga- JM, Carey JR: Electrical stimulation driving functional im- nization of movement after stroke. Am J Occup Ther 1999; provements and cortical changes in subjects with stroke. 53:333–44 Exp Brain Res 2004;154:450–60 21. Wu C, Trombly CA, Lin K, Tickle-Degnen L: A kinematic 9. Sullivan JE, Hedman LD: A home program of sensory and study of contextual effects on reaching performance in neuromuscular electrical stimulation integrated with up- persons with and without stroke: inﬂuences of object avail- per-limb task practice in a patient who is stable after a ability. Arch Phys Med Rehabil 2000;81:95–101 stroke. Phys Ther 2004;84:1045–54 22. Hanlon RE: Motor learning following unilateral stroke. 10. Powell J, Pandyan AD, Granat M, Cameron M, Stott DJ: Arch Phys Med Rehabil 1996;77:811–5 Electrical stimulation of wrist extensors in poststroke 23. Trimble MH, Enoka RM: Mechanisms underlying the train- hemiplegia. Stroke 1999;30:1384–9 ing effects associated with neuromuscular electrical stimu- 11. Alon G, Dar A, Katz-Behiri D, Weingarden H, Nathan R: lation. Phys Ther 1991;71:273–80 Efﬁcacy of a hybrid upper limb neuromuscular electrical 24. Beer RF, Dewald JPA, Dawson ML, Rymer WZ: Target- stimulation system in lessening selected impairments and dependent differences between free and constrained arm dysfunctions consequent to cerebral damage. J Neurol Re- movements in chronic hemiparesis. Exp Brain Res 2004; habil 1998;12:73–9 156:458–70 12. Hummelsheim H, Maier-Loth ML, Eickhof C: The func- 25. Keller T, Ellis MD, Dewald JP: Overcoming abnormal joint tional value of electrical muscle stimulation for the reha- torque patterns in paretic upper extremities using triceps bilitation of the hand in stroke patients. Scand J Rehabil stimulation. Artif Organs 2005;29:229–32 Med 1997;29:3–10 26. Sullivan JE, Hedman LD: Effects of home-based sensory and 13. Pandyan AD, Powell J, Futter C, Granat MH, Stott DJ: motor amplitude electrical stimulation on arm dysfunction Effects of electrical stimulation on the wrist of hemiplegic in chronic stroke. Clin Rehabil 2007;21:142–50 subjects. Physiotherapy 1996;82:184–8 27. Weingarden HP, Zeilig G, Heruti R, et al: Hybrid functional 14. Hsieh CL, Hsueh IP, Chiang FM, Lin PH: Inter-rater reli- electrical stimulation orthosis system for the upper limb: ability and validity of the Action Research Arm Test in effects on spasticity in chronic stable hemiplegia. Am J Phys stroke patients. Age Ageing 1998;27:107–14 Med Rehabil 1998;77:276–81 15. Wood-Dauphinee SL, Williams JI, Shaprio SH: Examining 28. van der Lee JH, Beckerman H, Lankhorst GJ, Bouter LM: outcome measures in a clinical study of stroke. Stroke The responsiveness of the Action Research Arm test and the 1990;21:731–9 Fugl-Meyer Assessment scale in chronic stroke patients. 16. van der Lee JH, Beckerman H, Knol DL, de Vet HC, Bouter J Rehabil Med 2001;33:110–3 LM: Clinimetric properties of the motor activity log for the 29. van der Lee JH, Wagenaar RC, Lankhorst GJ, Vogelaar TW, assessment of arm use in hemiparetic patients. Stroke 2004; Deville WL, Bouter LM: Forced use of the upper extremity 35:1410–4 in chronic stroke patients: results from a single-blind ran- 17. Ring H, Rosenthal N: Controlled study of neuroprosthetic domized clinical trial. Stroke 1999;30:2369–75 functional electrical stimualtion in sub-acute post-stroke 30. Uswatte G, Taub E, Morris D, Vignolo M, McCulloch K: rehabilitation. J Rehabil Med 2005;37:32–6 Reliability and validity of the upper extremity motor activity 18. de Kroon JR, Ijzerman MJ, Chae J, Lankhorst GJ, Zilvold G: log-14 for measuring real-world use. Stroke 2005;36: Relation between stimulation characteristics and clinical 2493–6

596 Hedman et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 Authors: John M. Saroyan, MD, FAAP Christopher J. Winfree, MD Neuropathy William S. Schechter, MD, MS, FAAP David Roye, MD Arnold P. Gold, MD

Afﬁliations: CASE REPORT From the Departments of Anesthesiology and Pediatrics (JMS, WSW), Department of Neurological Surgery (CJW), Department of Orthopaedic Surgery (DR), and Sciatic Neuropathy after Departments of Neurology and Pediatrics (APG), Columbia University Lower-Extremity Trauma College of Physicians and Surgeons, New York, New York. Successful Treatment of an Uncommon Pain and Disability Syndrome in an Adolescent Correspondence: All correspondence and requests for reprints should be addressed to Dr. John Saroyan, Division of Pain ABSTRACT Management, PH5-500, Columbia Saroyan JM, Winfree CJ, Schechter WS, Roye D, Gold AP: Sciatic neuropathy University, Department of Anesthesiology, 622 W. 168th Street, after lower-extremity trauma: successful treatment of an uncommon pain and disabil- New York, NY 10032. ity syndrome in an adolescent. Am J Phys Med Rehabil 2007;86:597–600.

0894-9115/07/8607-0597/0 Lower-extremity trauma is an uncommon but reported cause of sciatic nerve American Journal of Physical injury in children and adolescents. Failure to identify sciatic neuropathy after Medicine & Rehabilitation traumatic injury to the lower extremity may lead to the delayed institution of Copyright © 2007 by Lippincott neuropathic pharmacotherapy, electrodiagnostic testing, physical therapy, and Williams & Wilkins increased risk for the development of complex regional pain syndrome. This article presents a case of an adolescent male with neuropathic pain and weakness DOI: 10.1097/PHM.0b013e31806dbdd2 in the right lower extremity after traumatic injury. Spontaneous recovery of the injured nerve occurred with early institution of pharmacologic and physical therapies. Operative exploration and neurolysis were considered but were not ultimately necessary. Key Words: Neuropathy, Trauma, Sciatic Nerve, Pain Management

Sciatic nerve trauma at the level of the buttock and gluteal fold can occur as a result of a variety of mechanisms. The most common mechanism of injury in children has been reported to be as a complication of intramuscular injection.1 Children as young as 7 yrs old have been reported to sustain thigh-level sciatic, tibial, and peroneal nerve lesions during trauma from additional causes including gunshot wounds, laceration, and stretch with or without fracture.2–4 This article presents an adolescent male with neuropathic pain and weakness in the right lower extremity after simultaneous sacral fracture and stretch injury to the sciatic nerve, sustained during motor vehicle injury. Diagnosis and successful treatment by a multidisciplinary team of pediatric subspecialists is described.

CASE REPORT A 14-yr-old, previously healthy male was struck by an automobile while riding his bicycle; he was dragged underneath the car for 150 feet. He sustained multiple abrasions and lacerations, including a very deep soft-tissue injury to the right buttock, involving the muscle and, possibly, the sciatic nerve, and he required emergency debridement. There was no associated head injury or loss of

July 2007 Sciatic Neuropathy 597 consciousness. Computed tomography scan of the pelvis documented what proved to be a stable ver- tebral fracture in the left hemisacrum. Two days later, the patient reported the new onset of neuropathic pain involving the second and third toes on the right side. The pain intensity was 2 out of 10 on the numeric pain rating scale (NRS). It was constant and described as “pins and needles” sensations and a feeling of “being crushed.” He also had “shooting” (NRS ϭ 9 out of 10) pain that occurred three to four times per day and that lasted from a few minutes to 30 mins. Initially, pain management included high-dose opioids and nonsteroidal antiinflammatory drugs, which provided no relief. FIGURE 2 Right-buttock lesion. The pediatric orthopedic surgeon referred him to the pediatric pain-management specialist 1 mo after his injury. Complaints of constant and spon- ors of the right foot were present. There was weak- taneous intermittent pain persisted. The patient ness on thigh flexion, leg extension, foot dorsiflex- required the use of a wheelchair because bearing ion, and toe extension. The patient was unable to weight on the right foot was extremely painful. On invert or evert the right foot. This was associated examination, there were multiple significant abra- with calf atrophy (Fig. 3). There was hypoesthesia sions of the right lower extremity that involved the involving the right lateral lower leg, the foot, the right tibial crest extending from below the patella fifth toe, and the space between the first and second to the ankle (Fig. 1). There was a right-buttock toes. Allodynia was present in the second and third lesion measuring 11 ϫ 8 cm (Fig. 2). Erythema and toes. Ankle jerk reflex was absent on the right side. edema of the right lower leg and occasional trem- Concurrent psychological evaluation did not reveal clinically significant levels of anxiety or depression. The diagnosis of femoral and sciatic neuropathy was tentatively made, with strong consider- ation given to the development of complex regional

FIGURE 1 Multiple abrasions of the right lower extremity. FIGURE 3 Right-sided calf atrophy.

598 Saroyan et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 pain syndrome type 2. The nerve injury was con- biceps indicate a sciatic nerve injury proximal to sidered to be either at the level of the sacrum, the distal thigh, because this muscle is innervated buttock, and/or right knee. Treatment of pain with by the sciatic nerve in the distal thigh. This finding gabapentin was initiated, and the dose was titrated excludes a peroneal injury solely at the fibular upward. The patient began a vigorous physical head. therapy schedule. A needle electromyogram was Another diagnostic challenge was distinguish- performed 5 mos after the injury. Abnormal spon- ing complex regional pain syndrome type 2 from a taneous activity, including positive sharp waves painful sciatic neuropathy. Severe pain, allodynia, and fibrillations, large-amplitude and long-dura- edema, and erythema were all present in this pa- tion motor units, and reduced recruitment pattern tient after his injury, but these findings did not on maximal effort, was recorded in the right medial spread beyond the confines of the sciatic nerve gastrocnemius, peroneus longus, and biceps short territory. If they had extended into the distribution head. Other muscles in the right leg, including the of another nerve or even the entire lower extrem- anterior tibialis, vastus lateralis, and gluteus me- ity, the diagnosis of complex regional pain syn- dius, showed no abnormal spontaneous activity, drome type 2 would have been made.6 normal motor-unit configuration, and a recruit- Psychological consequences of the child after ment pattern consistent with effort. These results traumatic injury are frequently overlooked. Twen- were consistent with sciatic nerve injury proximal ty-five percent of traffic-injured children (n ϭ 102) to the midthigh, and they correlated with the lo- between 3 and 18 yrs of age are reported to have cation of the buttock wound. Daily outpatient experienced diagnostic posttraumatic stress disor- physical therapy included passive and active range der. Only 46% sought help of any kind.7 Despite of motion in both lower extremities, desensitiza- the absence of significant symptoms, the patient tion of the right lower extremity, and gait training. underwent psychological evaluation at 1 and 6 mos The patient achieved an improvement in pain con- after injury. Psychological counseling, such as in- trol and strengthening of proximal and distal mus- dividual or family therapy, is useful for elucidating culature within 1 mo of the initiation of physical important precipitating or perpetuating factors and and neuropathic pain therapy. Six months after to enhance coping and pain-management strate- injury, the dosage of gabapentin was tapered with- gies. In addition to psychotherapy, biofeedback and out recrudescence of the patient’s pain. Repeat relaxation training are considered to be an integral psychological evaluation showed normal reintegra- part of treatment for complex regional pain syn- tion into family and school activities, with no evi- drome type 1.8 dence of posttraumatic stress disorder. Physiotherapy of the affected extremity represents the cornerstone of treatment after pediatric DISCUSSION traumatic nerve injuries.9,10 Aggressive physical The buttock wound, the presence of hypoes- therapy strengthens uninvolved compensatory thesia in a tibial distribution, allodynia in a super- muscles; it also strengthens involved, partially de- ficial peroneal distribution, and weakness in the nervated muscles and maintains the flexibility of anterior, lateral, and posterior compartments of totally denervated muscles and joints in prepara- the leg, were considered in identifying the location tion for their eventual reinnervation and recovery. and mechanism of nerve injury. It was concluded Without active use of the affected extremity, dele- that the cause was likely to be a stretch or lacera- terious changes such as disuse atrophy, stiffness, tion injury to the sciatic nerve. Though the patient and complex regional pain syndrome can occur. had a sacral fracture and soft-tissue injury at the Analgesia is a crucial adjunct to physiotherapy. level of the right fibular head, his neurological Standard analgesics such as nonsteroidal antiin- examination and electrodiagnostic studies did not flammatory medications and opioids are effective support nerve injuries solely at these locations. for treating pain from bone and soft-tissue injuries, Electrodiagnostic studies performed 3–4 wks but they are not first-line agents for neuropathic after injury are extremely useful for delineating the pain. Frequently described as dysesthetic burning, pattern of traumatic nerve injuries. Delayed elect- lancinating, or “pins and needles” sensations, neu- rodiagnostic examination is necessary to permit ropathic pain is related to aberrant somatosensory denervation changes to occur in the affected mus- processes occurring in the peripheral nervous sys- cles.5 Additionally, these studies may help to dis- tem, central nervous system, or both.11 An exten- tinguish between decreased motor function attrib- sive adult literature of randomized, placebo-con- utable to impaired nerve conduction vs. decreased trolled trials supports the use of gabapentin in a motor activity attributable to pain. Perhaps most variety of neuropathic pain syndromes.12 Pediatric important is the sampling of individual muscles to and adolescent case reports and case series have localize the level of nerve injury. In this case, suggested the off-label use of gabapentin when denervational changes within the short head of the neuropathic pain is present in settings including

July 2007 Sciatic Neuropathy 599 compressive malignancy, phantom limb, posttho- REFERENCES racotomy and complex regional pain syn- 1. Beregeson PS, Singer SA, Kaplan AM: Intramuscular injec- drome.13–16 The starting dose is usually 5 mg/kg tions in children. Pediatrics 1982;70:944–8 per dose at bedtime, with increases every 3 days to 2. Kim DH, Murovic JA, Tiel RL, Kline DG: Management and outcomes in 353 surgically treated sciatic nerve lesions. a maximum of 40 mg/kg per day, divided into J Neurosurg 2004;101:8–17 intervals of every 8 hrs. Regional anesthetic blocks 3. Kim DH, Cho Y-J, Ryu S, Tiel RL, Kline DG: Surgical may also play a role in the management of painful management and results of 135 tibial nerve lesions at the nerve injuries in children.17 Louisiana State University Health Sciences Center. Neuro- Many nerve injuries spontaneously recover surgery 2003;53:1114–25 without the need for surgical nerve repair; how- 4. Kim DH, Kline DG: Management and results of peroneal ever, surgery is indicated in some circumstances. nerve lesions. Neurosurgery 1996;39:312–20 Sharp, lacerating injuries require immediate sur- 5. Stewart JD: Focal Peripheral Neuropathies, ed 3. Philadel- phia, Pa, Lippincott Williams & Wilkins, 2000, pp 11–35 gical repair. Blunt, lacerating injuries—for exam- 6. Thimineur MA, Saberski L: Complex regional pain syn- ple, by jagged metal in motor vehicle injury or a drome type 1 (RSD) or peripheral mononeuropathy? A lawnmower blade—require surgical repair 2–4 wks discussion of three cases. Clin J Pain 1996;12:145–50 after injury to permit margination of devitalized 7. De Vries AP, Kassam-Adams N, Cnaan A, Sherman-Slate E, tissue at the nerve stumps. Nerve injuries from Gahhagher PR: Looking beyond the physical injury: post- compression, gunshot, or traction may require re- traumatic stress disorder in children and parents after pediatric traffic injury. Pediatrics 1991;104:1293–9 pair after a 3- to 4-month delay.18 These lesions typically do not transect the nerve, resulting in a 8. Wilder RT, Berde CB, Wolohan M, et al: Reflex sympathetic dystrophy in children. Clinical characteristics and follow-up neuroma-in-continuity, and they often spontane- of seventy patients. J Bone Joint Surg Am 1992;74:910–9 ously recover without the need for surgery. When 9. Sherry DD, Wallace CA, Kelly C, Kidder M, Sapp L: Short- recovery does not occur after 3–4 mos, open ex- and long-term outcomes of children with complex regional ploration is required to permit nerve-action poten- pain syndrome type 1 treated with exercise therapy. Clin J Pain 1999;15:218–23 tial recordings to be performed across the affected nerve segment. Recovery requires the regrowth of 10. Lee BH, Scharff L, Sethna NF, et al: Physical therapy and cognitive-behavioral treatment for complex regional pain axons down the degenerated segment. The amount syndrome. J Pediatr 2002;141:135–40 of time required is variable, but it is roughly equal 11. Woolf CJ, Mannion RJ: Neuropathic pain: aetiology, symp- to 1 mm/day of damaged nerve segment length, toms, mechanisms and management. Lancet 1999;353: increasing distally along the extremity.5 Surgical 1959–64 repair was ultimately deemed unnecessary because 12. Backonja M, Glanzman RL: Gabapentin dosing for neuro- the patient demonstrated progressive recovery of pathic pain: evidence from randomized, placebo-controlled clinical trials. Clin Ther 2003;25:81–104 his femoral, sciatic, common peroneal, deep pero- 13. Keskinbora K, Pekel AF, Aydinli I: The use of gabapentin in neal, and tibial nerves. a 12 year old boy with cancer pain. Acta Anaesthesiol Scand 2004;48:663–4 CONCLUSION 14. Rusy LM, Troshynki TJ, Weisman SJ: Gabapentin in phan- Rehabilitation of the adolescent or pediatric tom limb pain management in children and young adults: patient presenting with neuropathic pain and mo- report of seven cases. J Pain Symptom Manage 2001;21: tor deficits after traumatic injury include early 78–82 institution of neuropathic pharmacotherapy, elec- 15. McGraw T, Stacey BR: Gabapentin for treatment of neuro- trodiagnostic testing at 3 wks after the injury, pathic pain in a 12-year-old girl. Clin J Pain 1998;14:354–6 rigorous physical therapy for the prevention of 16. Wheeler DS, Vaux KK, Tam DA: Use of gabapentin in the treatment of childhood reflex sympathetic dystrophy. Pedi- disuse, psychological assessment, and consultation atr Neurol 2000;22:220–1 with a peripheral nerve surgeon. 17. Berde CB, Lebel A: Complex regional pain syndromes in For patients who are severely affected by pain children and adolescents. Anesthesiology 2005;102:252–5 and disability, inpatient rehabilitation and regional 18. Winfree CJ: Peripheral nerve injury evaluation and man- anesthetic techniques should be considered. agement. Curr Surg 2005;62:469–76

600 Saroyan et al. Am. J. Phys. Med. Rehabil. ● Vol. 86, No. 7 VISUAL VIGNETTE

A Thoracic Disc Herniation Presenting as an Abdominal Hernia Myron M. LaBan, MD, MMSc, and Gabriel Gorin, MD From the Department of Physical Medicine & Rehabilitation, William Beaumont Hospital, Royal Oak, Michigan. 0894-9115/07/8607-0601/0 American Journal of Physical Medicine & Rehabilitation Copyright © 2007 by Lippincott Williams & Wilkins DOI: 10.1097/PHM.0b013e31806dd497

Thoracic disc herniations (TDHs) occur infrequently, accounting for fewer than 1% of all spinal disc prolapses.1 Although most TDHs present initially with radicular pain, FIGURE 1 Large abdominal-wall hernia (arrow) with tho- others may demonstrate only myelopathic signs and/or, racic disc herniation. even less often, ipsilateral abdominal muscle weakness.2,3 A 75-yr-old male who had been successfully treated with intermittent split-table pelvic traction for a T12–L1 disc herniation returned a year later with new, abrupt- onset, severe left-flank and abdominal pain. The physical examination revealed a large abdominal- wall hernia (Fig. 1) involving both the left upper- and lower-abdominal quadrants. The straight-leg raising test was unrestricted. Sensory and motor testing and the deep tendon reflexeswere all normal. However, there was a notable reduction in the motor units of the muscular abdominal wall. No long tract signs were observed, and the bowels and bladder were controlled. Electroneuromyography demonstrated 4ϩ fibrillatio- nand positive waves in the left paraspinal muscle at the T10–T12 levels, without similar findings in the adjacent FIGURE 2 Thoracic disc herniation (arrow) at T12–L1. abdominal muscle. However, there was a notable reduction in motor units in the abdominal musculature. stabilizing influence of the rib cage. When thoracic radic- The peroneal motor nerve conductions were normal. ular pain is accompanied by myelopathic signs (i.e., hyper- Magnetic resonance imaging revealed a large, left reflexia and Babinski reflexes), they mimic the presence of T12–L1 disc herniation (Fig. 2). With persistent pain and a spinal cord tumor. Although suggestive, an abdominal associated difficulty in walking, the patient was referred to hernia in the presence of radicular thoracic pain is not a spinal surgeon and was restarted on split-table intermit- pathognomonic evidence of a TDH. Other sources of tho- tent pelvic traction after thermal therapy. racic root compromise include, among others, diabetes With continuing conservative treatment, the patient’s mellitus and/or viral exanthems (i.e., poliomyelitis), as well pain slowly remitted. as herpes zoster. TDHs account for 0.24–0.75% of all disc herniations, 1 with an overall incidence of 1/1,000,000 patients per year. REFERENCES This relative paucity of TDHs has been attributed to the 1. Arce CA. Herniated thoracic discs. Neurol Clin 1985;3:383–92 2. Brown CW. The natural history of thoracic discs. Spine 1992;17: All correspondence and requests for reprints should be addressed to Myron M. 97–102 LaBan, MD, MMSc, Department of Physical Medicine & Rehabilitation, William 3. Ozturk C. Far lateral thoracic disc herniation presenting as flank pain. Beaumont Hospital, Royal Oak, MI 48073. Spine 2006;6:201–3

July 2007 Visual Vignette 601