Gait & Posture 39 (2014) 1097–1102
Contents lists available at ScienceDirect
Gait & Posture journal homepage: www.elsevier.com/locate/gaitpost
Mediolateral foot placement ability during ambulation in individuals with chronic post-stroke hemiplegia Angelika Zissimopoulos a,b,*, Rebecca Stine c, Stefania Fatone a, Steven Gard a,b,c a
Northwestern University Prosthetics-Orthotics Center, Department of Physical Medicine & Rehabilitation, Feinberg School of Medicine, Chicago, IL, USA Northwestern University, Department of Biomedical Engineering, Evanston, IL, USA c Jesse Brown VA Medical Center, Department of Veterans Affairs, Chicago, IL, USA b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 15 May 2013 Received in revised form 12 December 2013 Accepted 22 January 2014
Mediolateral (ML) foot placement is an effective way to redirect the lateral trajectory of the body center of mass (BCoM) during ambulation, but has only been partly characterized in the chronic post-stroke population despite their increased risk for falling [1]. During able-bodied gait, the locomotor system coordinates lower limb swing phase kinematics such that an appropriate ML foot placement occurs upon foot contact. Muscle weakness and abnormal motor patterns may impair foot placement ability poststroke. The purpose of this study was to characterize ML foot placement ability during post-stroke ambulation by quantifying ML foot placement accuracy and precision, for the both sound and affected feet. Age matched able-bodied individuals were recruited for comparison. All participants were instructed to target step widths ranging from 0 to 45% leg length, as marked on the laboratory floor. Results of this study confirmed that ML foot placement accuracy and precision were significantly lower for the post-stroke group as compared to the control group (p = 0.0). However, ML foot placement accuracy and precision were not significantly different between the affected and sound limbs in the poststroke group. The lowest accuracy for post-stroke subjects was observed at both extreme step width targets (0 and 45%). Future work should explore potential mechanisms underlying these findings such as abnormal motor coordination, lower limb muscle strength, and abnormal swing phase movement patterns. ß 2014 Elsevier B.V. All rights reserved.
Keywords: Mediolateral foot placement Stroke Gait analysis Balance Human locomotion
1. Introduction The ability to modify mediolateral (ML) foot placement is important for safe and efficient ambulation [2,3]. ML foot placement is an effective way to redirect the ML component of the body center of mass (BCoM) trajectory during ambulation [4]. To achieve the desired foot placement at the beginning of stance phase, adjustments to the locomotor system (likely via hip ab/ adduction) occur during the preceding swing phase [2]. Pathological populations (e.g., persons post-stroke) with impairments of neuromuscular functioning that affect swing phase may be unable to effectively adjust foot placement during ambulation. If ML redirection is inadequate, the ML component of BCoM may progress beyond the functional base of support (BoS), and result in instability or a fall. Therefore, the ML location of the foot is
* Corresponding author at: Northwestern University Prosthetics-Orthotics Center, 680 North Lake Shore Drive, Suite 1100, Chicago, IL 60611, USA. Tel.: +1 3125035730. E-mail address: [email protected] (A. Zissimopoulos). 0966-6362/$ – see front matter ß 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.gaitpost.2014.01.015
important because it establishes the general limits within which the BCoM must reside to maintain safe forward progression [1]. Stroke is the leading cause of long-term disability [5]. Hemiparesis and impaired motor coordination resulting from stroke lead to myriad abnormal gait characteristics. Specifically, abnormal ankle, knee, and hip control during swing phase may impair ML foot placement ability. While a direct link has not been established between ML foot placement ability and falls, ML foot placement is important for maintaining balance during ambulation, the task during which most falls occur in the post-stroke population [6,7]. ML foot placement ability, defined here as the accuracy and precision of targeted ML foot placements, remains uncharacterized during post-stroke locomotion. However, ML foot placement accuracy for both the affected and sound limbs of post-stroke individuals attempting mid-swing ML foot placement adjustments during supported and unsupported single-step tasks have been reported [8]. Notably, ML foot placement accuracy was lowest when subjects were unsupported and were aiming for medial targets. Accuracy increased when subjects completed the same task with an external support frame. These findings suggest that subjects may have prioritized balance, using wider foot
1098
A. Zissimopoulos et al. / Gait & Posture 39 (2014) 1097–1102
placements to ensure stability, at the expense of experimental task accuracy for narrow targets [8]. Likewise, ML foot placement during ambulation may be indicative of the overall balance capabilities of the post-stroke locomotor system. The purpose of the present study was to characterize bilateral ML foot placement ability during post-stroke gait. We defined ‘‘ability’’ as both the accuracy and precision (i.e., variability) of ML foot placement in response to step width targets ranging from 0 to 45% of the subject’s leg length (LL). To reduce the confounding influence of acute neuromuscular recovery, we recruited individuals in the chronic phase of post-stroke recovery (defined here as at least one year post-stroke). Additionally, an age-matched convenience sample of able-bodied individuals served as a control group. We hypothesized that post-stroke individuals would have reduced ML foot placement ability with their affected foot compared to both their sound foot and controls. We hypothesized that ML foot placement ability for the affected foot of individuals post-stroke would be greatest at target step widths near their preferred step width. We did not expect ML foot placement ability to be step width dependent for the sound foot or for controls. The results of this study will contribute to our understanding of ML foot placement in persons post-stroke, an important mechanism for maintaining dynamic balance. 2. Methods Individuals with chronic post-stroke hemiplegia were recruited from a nearby rehabilitation hospital, from a research subject database in our center, and from the surrounding community. Subjects were required to be greater than one year post-stroke, 18 years of age or older, able to understand simple instructions, and able to walk with shoes but without a cane or other assistive device for at least 12 m. Subjects were excluded if they had other comorbidities that would affect gait or could not ambulate without assistive devices. We also recruited a convenience sample of ablebodied individuals with no underlying gait-related pathologies. The University’s Institutional Review Board approved this study. Subjects provided written informed consent prior to testing. Subject-specific information recorded for all subjects included gender, age, height, weight, dominant limb, and shod foot length; and, for stroke subjects, affected side and number of years poststroke. The target step width for each condition was based on leg length (LL) measured bilaterally from the anterior superior iliac spine (ASIS) to the ipsilateral medial malleolus with subjects in a supine position, the standard in clinical practice. If LL discrepancies were present between the limbs, an average LL was calculated. Shod foot length was measured using a Ritz stick (Woodrow Engineering Company, WI). Affected limb data for the post-stroke subjects were compared to data from the non-dominant limb of controls. The modified Helen Hayes full-body marker set, a standard marker configuration used in clinical gait analysis [9], defined the placement of passive retro-reflective markers for kinematic data collection. Marker positions were recorded at 120 Hz with an eight-camera digital real-time motion capture system (Motion Analysis Corporation, Santa Rosa, CA, USA). Both experimental groups were tested at the following randomized step widths: 0%, 15%, 30%, and 45% LL. For reference, able-bodied step widths are typically 12% LL [10]. Data from previous experiments in our laboratory indicate that average step width post-stroke is 20% LL (18.8 cm) when subjects walk at their preferred walking speed without assistive devices [11]. The range of step widths chosen for the present study was intended to challenge both subject groups at step widths narrower and wider than their preferred step width. Step widths were normalized by LL to account for the biomechanical effects of stature. Tape placed on
the laboratory floor at each selected step width indicated target ML foot placement. Subjects were instructed to walk at a comfortable walking speed while placing one foot on each line or as close as possible. In addition to their preferred walking speed, control subjects were tested at a walking speed that matched the preferred walking speed of post-stroke subjects to determine if ML foot placement ability displayed speed dependent characteristics. Laboratory personnel demonstrated the experimental task to ensure that study participants understood verbal instructions. Subjects completed six walking trials for each target step width and rested as needed throughout the experiment. The standard marker set does not provide sufficient information to locate the perimeter of the shoe at each ML foot placement during the walking trials. Since this was necessary to categorize each ML foot placement as a success or miss, a digital representation of the outline of the shoe was created using a technique previously developed in our laboratory [12]. Following collection of the static trial, laboratory personnel traced an ink outline of each foot onto paper placed on the floor prior to data collection. Subjects stepped off the paper and laboratory personnel digitized the shoe outline by manually tracing the ink outline with a retro-reflective marker. Coordinate system transformations were used to calculate each shoe outline location during walking trials. For the post-stroke group, six walking trials were used to calculate preferred walking. Subjects walked back and forth across a 10 m walkway at their preferred walking speed, shod but without assistive devices or step width restrictions. Following data collection, mean walking speed was calculated for the post-stroke group. Individuals whose preferred walking speed was more than two standard deviations away from the group mean were excluded from subsequent analyses. Marker position data were smoothed using a 4th order bidirectional Butterworth filter at an effective cut-off frequency of 6 Hz [13], using Cortex software (MAC, Santa Rosa, CA, USA). Gait events were generated using OrthoTrak software (MAC, Santa Rosa, CA, USA). Global foot outlines from the static trial were transformed into local foot coordinates using custom MATLAB1 (The MathWorks Inc., Natick, MA, USA) programs. Local foot outline coordinates were calculated for each frame of the walking trial before being transformed back to global coordinates for analysis. The following criteria helped categorize results: ML foot placement was a success if it was located directly on the target line, a lateral miss if the entire foot outline was lateral to the target line, or a medial miss if the entire foot outline was medial to the target line. Thus, the frequency of successes, lateral misses, and medial misses characterized the directional bias of ML foot placement for each step width condition. The distance between the ankle center and the target line quantified ML foot placement accuracy for each foot strike. The standard deviation of this distance for all foot strikes within a step width condition quantified ML foot placement precision. A mixed two-way repeated measures multivariate analysis of variance (MANOVA) was used to test whether ML foot placement ability (i.e., accuracy and precision) was different between the affected foot for the post-stroke group and the non-dominant foot for the control group (the between-subjects factor = group [stroke/ control] and the within-subject factor = step width [0%, 15%, 30%, and 45% LL]). A two-way repeated measures MANOVA was used to test whether ML foot placement ability was different between the affected and sound limbs (wherein the within subjects factors = step width [0%, 15%, 30%, and 45% LL] and leg [affected/sound]). If residuals were not normally distributed or if variances were not homogeneous between the two groups, a transformation was applied.
A. Zissimopoulos et al. / Gait & Posture 39 (2014) 1097–1102
The relationship between ML foot placement ability and target step width was assessed using polynomial contrasts for step width. 3. Results Fifteen subjects with chronic post-stroke hemiplegia completed the study. Preferred walking speed for the post-stroke group was 0.69 0.26 m/s (mean one standard deviation). Two subjects displayed preferred walking speeds that were greater than two standard deviations about the mean (0.09 m/s and 1.27 m/s), excluding them from subsequent analyses. Preferred walking speed for the remaining 13 subjects was 0.69 0.17 m/s. Six able-bodied subjects, whose ages were within one standard deviation of the mean age of the post-stroke group, participated in this study. Table 1 summarizes these and other subject specific characteristics. Fig. 1 depicts the frequency with which post-stroke subjects successfully placed each foot on the target line. Control subjects were >99% accurate on average for all step width conditions. No multivariate effect of walking speed on ML foot placement accuracy and precision was detected in control subjects (F4,2 = 1.242, p = 0.492). Therefore, accuracy and precision data from control subjects’ preferred walking speed were used for the present analysis. ML foot placement data were transformed using a natural log transformation so that variances between stroke and control data were homogeneous and residuals were normally distributed. The between-subjects multivariate main effect of group (F2,16 = 16.897, p = 0.0, Wilk’s Lambda = 0.321) indicating that 67.9% of the variance in foot placement accuracy and precision was accounted for by group: control or post-stroke (Figs. 2 and 3). Univariate between-group tests were significant (p < 0.025 corrected for the number of dependent variables), indicating that both precision (F1,17 = 34.219, p = 0.000) and accuracy (F1,17 = 15.620, p = 0.001) differed across groups. Within the post-stroke group, the multivariate main effect of leg (F2,11 = 1.807, p = 0.210, Wilk’s Lambda = 0.753) and interaction between step width and leg (F6,7 = 3.042, p = 0.086, Wilk’s
1099
Lambda = 0.277) were not significant. Hence, there was no difference in ML foot placement ability between the sound and affected limbs, and both limbs responded similarly to the experimental task. A significant multivariate main effect for step width (F6,7 = 32.445, p = 0.0, Wilk’s Lambda = 0.035) was detected, indicating that 96.5% of the variance in foot placement accuracy and precision within the post-stroke group was accounted for by target step width. A significant quadratic relationship was detected for both accuracy (F1,12 = 69.408, p = 0.0) and precision (F6,7 = 22.069, p = 0.01) such that post-stroke subjects were most accurate at 15% LL, the target step width closest to published values for preferred step width (20.00 0.04% LL). Fig. 4 illustrates the change in walking speed at each target step width for post-stroke (Fig. 4A) and control (Fig. 4B) subjects. In the 0% LL target step width, the decrease in walking speed for poststroke subjects was 44%, to a median speed of 0.37 m/s (range 0.28–0.52 m/s). While able-bodied subjects also decreased their walking speed, the magnitude differed; median decrease in walking speed was 14% of preferred speed for the 0% LL target step width, to a median speed of 1.04 m/s (range 0.78–1.30 m/s). 4. Discussion The purpose of this study was to evaluate ML foot placement ability of individuals with chronic post-stroke hemiplegia during gait because ML foot placement has been identified as an effective solution to the challenge of ML BCoM trajectory redirection in ablebodied persons [2]. However, this strategy was impaired in our post-stroke subjects. The results from the present study support the hypothesis that ML placement of the affected foot is less accurate and more variable for persons with chronic post-stroke hemiplegia than a control population comparable in age. Contrary to our hypotheses, there were no differences in either accuracy or precision between affected and sound limbs in our post-stroke subjects. However, both sound and affected limb accuracy and precision were significantly different from control
Table 1 Subject characteristics. Persons with post-stroke hemiplegia Subject
Gender
Age (yrs)
Height (m)
Mass (kg)
Leg length (m)
Time post-stroke (yrs)
Affected side
1 2 3 4 5 6 7 8 9 10 11 12 13
M M M F F F M F F M M M M
54.7 51.9 59.1 62.9 58.3 44.9 56.3 63.6 58.4 66.3 50.8 47.1 44.3
1.75 1.95 1.72 1.54 1.66 1.58 1.71 1.60 1.68 1.71 1.79 1.68 1.78
76.0 65.0 66.0 63.0 88.3 56.8 112.5 58.5 67.8 79.5 85.8 93.3 67.4
0.95 1.05 0.92 0.82 0.87 0.82 0.89 0.87 0.92 0.87 0.93 0.83 0.99
1.0 8.4 25.6 19.8 2.4 17.8 18.0 8.4 13.8 16.9 4.9 14.1 16.2
R R R L R R R R L R R L L
Avg SD
67% M
55.3 7.1
1.70 0.11
0.90 0.07
12.9 7.3
69% R
75.4 16.1
Control subjects Subject
Gender
1 2 3 4 5 6
F F F M M M
Avg SD
50% M
Age (yrs) 54.0 58.3 56.3 53.2 58.2 56.0 56 2.1
Avg, average; SD, standard deviation; yrs, years; M, male; F, female; R, right; L, left.
Height (m)
Mass (kg)
Leg length (m)
1.59 1.62 1.53 1.84 1.70 1.75
65.1 65.8 62.3 125.0 79.4 96.8
0.85 0.89 0.82 0.95 0.89 0.93
1.67 0.11
82.4 24.5
0.89 0.05
[(Fig._1)TD$IG]
1100
A. Zissimopoulos et al. / Gait & Posture 39 (2014) 1097–1102
Fig. 1. Frequency of accurate foot placements for post-stroke subjects. Bold horizontal lines represent the median percent of foot strikes for each condition. The 25th and 75th percentiles of the data are graphically represented by the upper and lower edges of the boxes. Whiskers extend to data not considered outliers while crosses represent outliers. (A) Frequency of foot placements lateral to the target for each step width target. (B) Frequency of foot placements on the target line for each step width target. (C) Frequency of foot placements medial to the target for each step width target. Data for the affected limb are shown in the top row and in the bottom row for the sound limb.
subjects. All subjects were instructed to place each foot on the target line or as close as possible, which allowed subjects to prioritize the accuracy of one limb over the other. Results suggest that subjects either did not prioritize one foot over the other, or that the bilateral nature of walking prevented prioritizing accuracy of one foot over the other. These findings corroborate previous
[(Fig._2)TD$IG]
Fig. 2. ML foot placement error measured as distance from target to ankle center and normalized by leg length for all limbs of interest.
reports of bilateral ML foot placement errors during single step tasks [8], extending understanding of post-stroke ML foot placement ability to walking. These findings have implications for post-stroke rehabilitation. Future studies should investigate strategies to improve ML foot placement bilaterally, to evaluate whether training can afford greater adjustability of ML foot placement and, therefore, step width post-stroke. Inspection of individual subject data revealed that four subjects always walked with more accurate affected limb ML foot placement, four always walked with more accurate sound limb ML foot placements and five used a combination of strategies depending upon the target step width. While the location of ML foot placements was constrained in this experimental protocol, subjects were free to achieve the task using any temporospatial, [(Fig._3)TD$IG]kinetic, or kinematic compensations available to them. Future
Fig. 3. Within subject variability in foot placement errors, normalized by leg length.
[(Fig._4)TD$IG]
A. Zissimopoulos et al. / Gait & Posture 39 (2014) 1097–1102
1101
Fig. 4. Change in walking speed between preferred and that during targeted conditions: (A) subjects with post-stroke hemiplegia and (B) control subjects.
work should explore whether subject characteristics such as muscle strength or type of compensatory actions (e.g., hip hiking and circumduction) predict ML foot placement ability. Information regarding predictor variables for ML foot placement ability may help guide rehabilitation strategies and further research. One compensation investigated in the present study was the reduction in walking speed in response to the experimental task (Fig. 4). Of note is the 44% reduction in walking speed, from a median of 0.65 m/s (range 0.62–0.83 m/s) to a median of 0.37 m/s (range 0.28–0.52 m/s) in the 0% LL condition. The dynamic nature of locomotion is questionable at these slow speeds. While this may suggest that dynamic stability is reduced, relationships between walking speed and gait stability remain unclear [14,15]. Medial ML foot placement errors at the widest target step width (Fig. 2, 45% LL condition) may also provide interesting insight regarding stability during ambulation. With each step, the BCoM is transferred laterally toward the support foot, with the BCoM remaining medial or just within the support foot during single support. The gravitational moment (medial to the support foot) is countered by the support hip abductors and a passive rotation torque (Ia) about the hip such that the body remains upright during the single support phase [16]. When transitioning toward the sound limb, the lateral distance the BCoM will travel (and therefore the magnitude of the destabilizing gravitational moment) depends upon lateral momentum supplied by the affected foot. Insufficient lateral momentum toward the support foot could result in a large gravitational moment and premature ‘‘fall’’ back toward the affected foot. Therefore, the widest step width poststroke individuals are able to achieve may be dependent on both affected limb strength and use of compensations to propel the body toward the sound support limb. In summary, the inability to achieve the wide ML foot placements in the 45% LL target condition may reflect an inability to generate sufficient lateral momentum during ambulation, yielding medial errors and narrower step widths. We also cannot rule out the possibility that ML foot placement ability was precluded by abnormal motor coordination. In persons with chronic post-stroke hemiplegia, coupling of hip adduction and knee extension torques has previously been reported during isometric hip torque generating tasks [17], and co-activation of
knee extensors (rectus femoris) and hip adductors (adductor longus) has been reported during isometric stretch reflex experiments [18]. To increase step width, individuals would likely extend their knee and abduct their hip in late swing. It is possible that as subjects attempted to ‘‘reach’’ for the wider target, coupling between knee extension and hip adduction necessitated hip hiking for toe-clearance purposes, reducing the maximum ML foot displacement of the affected foot. However, this manifestation of this abnormal across-joint coupling during gait requires further research. It should be noted that results from this study might not generalize to a more severely affected post-stroke population where foot placement impairments are likely more pronounced. Additionally, the relatively small sample size of the current study may limit generalizability. Finally, results may not apply to subjects using orthoses, canes, or other assistive devices. Despite these limitations, this study provides useful information regarding ML foot placement during post-stroke ambulation. ML motion of the BCoM is related to foot placement and the ability to modify foot placement in response to environmental obstacles and external perturbations is necessary for safe ambulation. While no direct link was established, inaccuracies in ML foot placement during locomotion may be related to the increased incidence of falls during walking reported within the chronic post-stroke population [6,7,19]. Given the particular difficulty subjects had with the narrowest step width condition (0% LL), rehabilitation that targets skills such as tandem walking may be particularly beneficial to this population. Likewise, studies investigating joint torque coupling, sound limb strength, and frontal plane compensations post-stroke may improve understanding of errors in the widest step width condition (45% LL). Further research is needed to investigate the potential mechanisms underlying bilateral ML foot placement impairments post-stroke. Acknowledgements The authors would like to acknowledge use of the Jesse Brown VA Medical Center Motion Analysis Research Laboratory (JBVAMCMARL) for data collection. The authors also acknowledge Richard Harvey, MD, and Elliot Roth, MD, at the Rehabilitation Institute of
1102
A. Zissimopoulos et al. / Gait & Posture 39 (2014) 1097–1102
Chicago (RIC) Center for Stroke Rehabilitation, and Lynda McCracken, CPO, and Christopher Robinson, MBA, CPO, ATC, FAAOP, at the RIC Prosthetics and Orthotics Clinical Center for their assistance with post-stroke subject recruitment and Wendy Murray, PhD, Matthew Tresch, PhD, Yasin Dhaher, PhD, and Andrew Hansen, PhD for their feedback regarding study design. Finally, the authors would like to acknowledge funding support from the Orthotic and Prosthetic Education and Research Foundation (Grant # RFA OPERF-2010-FA-1) (Recipient: Angelika Zissimopoulos), the Dr. John N. Nicholson Fellowship (Recipient: Angelika Zissimopoulos), and the National Institute on Disability and Rehabilitation Research (NIDRR) Department of Education (Grant # H133E080009) (Principal Investigators: Steven Gard and Stefania Fatone). The opinions contained in this publication are those of the grantee and do not necessarily reflect those of the Department of Education. Conflict of interest statement The authors have no conflicts of interest to disclose. References [1] Balasubramanian CK, Neptune RR, Kautz SA. Foot placement in a body reference frame during walking and its relationship to hemiparetic walking performance. Clin Biomech 2010;25:483–90. [2] Kuo AD, Donelan JM. Dynamic principles of gait and their clinical implications. Phys Ther 2010;90:157–74. [3] Redfern MS, Schumann T. A model of foot placement during gait. J Biomech 1994;27:1339–46. [4] Kuo AD. Stabilization of lateral motion in passive dynamic walking. Int J Robot Res 1999;18:917–30.
[5] Roger VL, Go AS, Lloyd-Jones DM, Benjamin EJ, Berry JD, Borden WB, et al. Heart disease and stroke statistics – 2012 update: a report from the American Heart Association. Circulation 2012;125:e2–20. [6] Hyndman D, Ashburn A, Stack E. Fall events among people with stroke living in the community: circumstances of falls and characteristics of fallers. Arch Phys Med Rehab 2002;83:165–70. [7] Jorgensen L, Engstad T, Jacobsen BK. Higher incidence of falls in long-term stroke survivors than in population controls: depressive symptoms predict falls after stroke. Stroke 2002;33:542–7. [8] Nonnekes JH, Talelli P, de Niet M, Reynolds RF, Weerdesteyn V, Day BL. Deficits underlying impaired visually triggered step adjustments in mildly affected stroke patients. Neurorehabil Neural Repair 2010;24:393–400. [9] Kadaba MP, Ramakrishnan HK, Wootten ME. Measurement of lower extremity kinematics during level walking. J Orthop Res 1990;8:383–92. [10] Donelan JM, Kram R, Kuo AD. Mechanical and metabolic determinants of the preferred step width in human walking. Proc Biol Sci 2001;268:1985–92. [11] Fatone S, Hansen AH. Effect of ankle-foot orthosis on roll-over shape in adults with hemiplegia. J Rehabil Res Dev 2007;44:11–20. [12] Saha D, Gard S, Fatone S, Ondra S. The effect of trunk-flexed postures on balance and metabolic energy expenditure during standing. Spine 2007;32:1605–11. [13] Winter DA. Biomechanics and motor control of human movement. 4th ed. New York: Wiley; 2010. [14] Krasovsky T, Lamontagne A, Feldman AG, Levin MF. Reduced gait stability in high-functioning poststroke individuals. J Neurophysiol 2013;109:77–88. [15] Hak L, Houdijk H, Steenbrink F, Mert A, VanderWurff P, Beek PJ, et al. Speeding up or slowing down? Gait adaptations to preserve gait stability in response to balance perturbations. Gait Posture 2012;36:260–4. [16] MacKinnon CD, Winter DA. Control of whole body balance in the frontal plane during human walking. J Biomech 1993;26:633–44. [17] Cruz TH, Dhaher YY. Evidence of abnormal lower-limb torque coupling after stroke: an isometric study. Stroke 2008;39:139–47. [18] Finley JM, Perreault EJ, Dhaher YY. Stretch reflex coupling between the hip and knee: implications for impaired gait following stroke. Exp Brain Res 2008;188:529–40. [19] Weerdesteyn V, de Niet M, van Duijnhoven HJ, Geurts AC. Falls in individuals with stroke. J Rehabil Res Dev 2008;45:1195–213.
Contents lists available at ScienceDirect
Gait & Posture journal homepage: www.elsevier.com/locate/gaitpost
Mediolateral foot placement ability during ambulation in individuals with chronic post-stroke hemiplegia Angelika Zissimopoulos a,b,*, Rebecca Stine c, Stefania Fatone a, Steven Gard a,b,c a
Northwestern University Prosthetics-Orthotics Center, Department of Physical Medicine & Rehabilitation, Feinberg School of Medicine, Chicago, IL, USA Northwestern University, Department of Biomedical Engineering, Evanston, IL, USA c Jesse Brown VA Medical Center, Department of Veterans Affairs, Chicago, IL, USA b
A R T I C L E I N F O
A B S T R A C T
Article history: Received 15 May 2013 Received in revised form 12 December 2013 Accepted 22 January 2014
Mediolateral (ML) foot placement is an effective way to redirect the lateral trajectory of the body center of mass (BCoM) during ambulation, but has only been partly characterized in the chronic post-stroke population despite their increased risk for falling [1]. During able-bodied gait, the locomotor system coordinates lower limb swing phase kinematics such that an appropriate ML foot placement occurs upon foot contact. Muscle weakness and abnormal motor patterns may impair foot placement ability poststroke. The purpose of this study was to characterize ML foot placement ability during post-stroke ambulation by quantifying ML foot placement accuracy and precision, for the both sound and affected feet. Age matched able-bodied individuals were recruited for comparison. All participants were instructed to target step widths ranging from 0 to 45% leg length, as marked on the laboratory floor. Results of this study confirmed that ML foot placement accuracy and precision were significantly lower for the post-stroke group as compared to the control group (p = 0.0). However, ML foot placement accuracy and precision were not significantly different between the affected and sound limbs in the poststroke group. The lowest accuracy for post-stroke subjects was observed at both extreme step width targets (0 and 45%). Future work should explore potential mechanisms underlying these findings such as abnormal motor coordination, lower limb muscle strength, and abnormal swing phase movement patterns. ß 2014 Elsevier B.V. All rights reserved.
Keywords: Mediolateral foot placement Stroke Gait analysis Balance Human locomotion
1. Introduction The ability to modify mediolateral (ML) foot placement is important for safe and efficient ambulation [2,3]. ML foot placement is an effective way to redirect the ML component of the body center of mass (BCoM) trajectory during ambulation [4]. To achieve the desired foot placement at the beginning of stance phase, adjustments to the locomotor system (likely via hip ab/ adduction) occur during the preceding swing phase [2]. Pathological populations (e.g., persons post-stroke) with impairments of neuromuscular functioning that affect swing phase may be unable to effectively adjust foot placement during ambulation. If ML redirection is inadequate, the ML component of BCoM may progress beyond the functional base of support (BoS), and result in instability or a fall. Therefore, the ML location of the foot is
* Corresponding author at: Northwestern University Prosthetics-Orthotics Center, 680 North Lake Shore Drive, Suite 1100, Chicago, IL 60611, USA. Tel.: +1 3125035730. E-mail address: [email protected] (A. Zissimopoulos). 0966-6362/$ – see front matter ß 2014 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.gaitpost.2014.01.015
important because it establishes the general limits within which the BCoM must reside to maintain safe forward progression [1]. Stroke is the leading cause of long-term disability [5]. Hemiparesis and impaired motor coordination resulting from stroke lead to myriad abnormal gait characteristics. Specifically, abnormal ankle, knee, and hip control during swing phase may impair ML foot placement ability. While a direct link has not been established between ML foot placement ability and falls, ML foot placement is important for maintaining balance during ambulation, the task during which most falls occur in the post-stroke population [6,7]. ML foot placement ability, defined here as the accuracy and precision of targeted ML foot placements, remains uncharacterized during post-stroke locomotion. However, ML foot placement accuracy for both the affected and sound limbs of post-stroke individuals attempting mid-swing ML foot placement adjustments during supported and unsupported single-step tasks have been reported [8]. Notably, ML foot placement accuracy was lowest when subjects were unsupported and were aiming for medial targets. Accuracy increased when subjects completed the same task with an external support frame. These findings suggest that subjects may have prioritized balance, using wider foot
1098
A. Zissimopoulos et al. / Gait & Posture 39 (2014) 1097–1102
placements to ensure stability, at the expense of experimental task accuracy for narrow targets [8]. Likewise, ML foot placement during ambulation may be indicative of the overall balance capabilities of the post-stroke locomotor system. The purpose of the present study was to characterize bilateral ML foot placement ability during post-stroke gait. We defined ‘‘ability’’ as both the accuracy and precision (i.e., variability) of ML foot placement in response to step width targets ranging from 0 to 45% of the subject’s leg length (LL). To reduce the confounding influence of acute neuromuscular recovery, we recruited individuals in the chronic phase of post-stroke recovery (defined here as at least one year post-stroke). Additionally, an age-matched convenience sample of able-bodied individuals served as a control group. We hypothesized that post-stroke individuals would have reduced ML foot placement ability with their affected foot compared to both their sound foot and controls. We hypothesized that ML foot placement ability for the affected foot of individuals post-stroke would be greatest at target step widths near their preferred step width. We did not expect ML foot placement ability to be step width dependent for the sound foot or for controls. The results of this study will contribute to our understanding of ML foot placement in persons post-stroke, an important mechanism for maintaining dynamic balance. 2. Methods Individuals with chronic post-stroke hemiplegia were recruited from a nearby rehabilitation hospital, from a research subject database in our center, and from the surrounding community. Subjects were required to be greater than one year post-stroke, 18 years of age or older, able to understand simple instructions, and able to walk with shoes but without a cane or other assistive device for at least 12 m. Subjects were excluded if they had other comorbidities that would affect gait or could not ambulate without assistive devices. We also recruited a convenience sample of ablebodied individuals with no underlying gait-related pathologies. The University’s Institutional Review Board approved this study. Subjects provided written informed consent prior to testing. Subject-specific information recorded for all subjects included gender, age, height, weight, dominant limb, and shod foot length; and, for stroke subjects, affected side and number of years poststroke. The target step width for each condition was based on leg length (LL) measured bilaterally from the anterior superior iliac spine (ASIS) to the ipsilateral medial malleolus with subjects in a supine position, the standard in clinical practice. If LL discrepancies were present between the limbs, an average LL was calculated. Shod foot length was measured using a Ritz stick (Woodrow Engineering Company, WI). Affected limb data for the post-stroke subjects were compared to data from the non-dominant limb of controls. The modified Helen Hayes full-body marker set, a standard marker configuration used in clinical gait analysis [9], defined the placement of passive retro-reflective markers for kinematic data collection. Marker positions were recorded at 120 Hz with an eight-camera digital real-time motion capture system (Motion Analysis Corporation, Santa Rosa, CA, USA). Both experimental groups were tested at the following randomized step widths: 0%, 15%, 30%, and 45% LL. For reference, able-bodied step widths are typically 12% LL [10]. Data from previous experiments in our laboratory indicate that average step width post-stroke is 20% LL (18.8 cm) when subjects walk at their preferred walking speed without assistive devices [11]. The range of step widths chosen for the present study was intended to challenge both subject groups at step widths narrower and wider than their preferred step width. Step widths were normalized by LL to account for the biomechanical effects of stature. Tape placed on
the laboratory floor at each selected step width indicated target ML foot placement. Subjects were instructed to walk at a comfortable walking speed while placing one foot on each line or as close as possible. In addition to their preferred walking speed, control subjects were tested at a walking speed that matched the preferred walking speed of post-stroke subjects to determine if ML foot placement ability displayed speed dependent characteristics. Laboratory personnel demonstrated the experimental task to ensure that study participants understood verbal instructions. Subjects completed six walking trials for each target step width and rested as needed throughout the experiment. The standard marker set does not provide sufficient information to locate the perimeter of the shoe at each ML foot placement during the walking trials. Since this was necessary to categorize each ML foot placement as a success or miss, a digital representation of the outline of the shoe was created using a technique previously developed in our laboratory [12]. Following collection of the static trial, laboratory personnel traced an ink outline of each foot onto paper placed on the floor prior to data collection. Subjects stepped off the paper and laboratory personnel digitized the shoe outline by manually tracing the ink outline with a retro-reflective marker. Coordinate system transformations were used to calculate each shoe outline location during walking trials. For the post-stroke group, six walking trials were used to calculate preferred walking. Subjects walked back and forth across a 10 m walkway at their preferred walking speed, shod but without assistive devices or step width restrictions. Following data collection, mean walking speed was calculated for the post-stroke group. Individuals whose preferred walking speed was more than two standard deviations away from the group mean were excluded from subsequent analyses. Marker position data were smoothed using a 4th order bidirectional Butterworth filter at an effective cut-off frequency of 6 Hz [13], using Cortex software (MAC, Santa Rosa, CA, USA). Gait events were generated using OrthoTrak software (MAC, Santa Rosa, CA, USA). Global foot outlines from the static trial were transformed into local foot coordinates using custom MATLAB1 (The MathWorks Inc., Natick, MA, USA) programs. Local foot outline coordinates were calculated for each frame of the walking trial before being transformed back to global coordinates for analysis. The following criteria helped categorize results: ML foot placement was a success if it was located directly on the target line, a lateral miss if the entire foot outline was lateral to the target line, or a medial miss if the entire foot outline was medial to the target line. Thus, the frequency of successes, lateral misses, and medial misses characterized the directional bias of ML foot placement for each step width condition. The distance between the ankle center and the target line quantified ML foot placement accuracy for each foot strike. The standard deviation of this distance for all foot strikes within a step width condition quantified ML foot placement precision. A mixed two-way repeated measures multivariate analysis of variance (MANOVA) was used to test whether ML foot placement ability (i.e., accuracy and precision) was different between the affected foot for the post-stroke group and the non-dominant foot for the control group (the between-subjects factor = group [stroke/ control] and the within-subject factor = step width [0%, 15%, 30%, and 45% LL]). A two-way repeated measures MANOVA was used to test whether ML foot placement ability was different between the affected and sound limbs (wherein the within subjects factors = step width [0%, 15%, 30%, and 45% LL] and leg [affected/sound]). If residuals were not normally distributed or if variances were not homogeneous between the two groups, a transformation was applied.
A. Zissimopoulos et al. / Gait & Posture 39 (2014) 1097–1102
The relationship between ML foot placement ability and target step width was assessed using polynomial contrasts for step width. 3. Results Fifteen subjects with chronic post-stroke hemiplegia completed the study. Preferred walking speed for the post-stroke group was 0.69 0.26 m/s (mean one standard deviation). Two subjects displayed preferred walking speeds that were greater than two standard deviations about the mean (0.09 m/s and 1.27 m/s), excluding them from subsequent analyses. Preferred walking speed for the remaining 13 subjects was 0.69 0.17 m/s. Six able-bodied subjects, whose ages were within one standard deviation of the mean age of the post-stroke group, participated in this study. Table 1 summarizes these and other subject specific characteristics. Fig. 1 depicts the frequency with which post-stroke subjects successfully placed each foot on the target line. Control subjects were >99% accurate on average for all step width conditions. No multivariate effect of walking speed on ML foot placement accuracy and precision was detected in control subjects (F4,2 = 1.242, p = 0.492). Therefore, accuracy and precision data from control subjects’ preferred walking speed were used for the present analysis. ML foot placement data were transformed using a natural log transformation so that variances between stroke and control data were homogeneous and residuals were normally distributed. The between-subjects multivariate main effect of group (F2,16 = 16.897, p = 0.0, Wilk’s Lambda = 0.321) indicating that 67.9% of the variance in foot placement accuracy and precision was accounted for by group: control or post-stroke (Figs. 2 and 3). Univariate between-group tests were significant (p < 0.025 corrected for the number of dependent variables), indicating that both precision (F1,17 = 34.219, p = 0.000) and accuracy (F1,17 = 15.620, p = 0.001) differed across groups. Within the post-stroke group, the multivariate main effect of leg (F2,11 = 1.807, p = 0.210, Wilk’s Lambda = 0.753) and interaction between step width and leg (F6,7 = 3.042, p = 0.086, Wilk’s
1099
Lambda = 0.277) were not significant. Hence, there was no difference in ML foot placement ability between the sound and affected limbs, and both limbs responded similarly to the experimental task. A significant multivariate main effect for step width (F6,7 = 32.445, p = 0.0, Wilk’s Lambda = 0.035) was detected, indicating that 96.5% of the variance in foot placement accuracy and precision within the post-stroke group was accounted for by target step width. A significant quadratic relationship was detected for both accuracy (F1,12 = 69.408, p = 0.0) and precision (F6,7 = 22.069, p = 0.01) such that post-stroke subjects were most accurate at 15% LL, the target step width closest to published values for preferred step width (20.00 0.04% LL). Fig. 4 illustrates the change in walking speed at each target step width for post-stroke (Fig. 4A) and control (Fig. 4B) subjects. In the 0% LL target step width, the decrease in walking speed for poststroke subjects was 44%, to a median speed of 0.37 m/s (range 0.28–0.52 m/s). While able-bodied subjects also decreased their walking speed, the magnitude differed; median decrease in walking speed was 14% of preferred speed for the 0% LL target step width, to a median speed of 1.04 m/s (range 0.78–1.30 m/s). 4. Discussion The purpose of this study was to evaluate ML foot placement ability of individuals with chronic post-stroke hemiplegia during gait because ML foot placement has been identified as an effective solution to the challenge of ML BCoM trajectory redirection in ablebodied persons [2]. However, this strategy was impaired in our post-stroke subjects. The results from the present study support the hypothesis that ML placement of the affected foot is less accurate and more variable for persons with chronic post-stroke hemiplegia than a control population comparable in age. Contrary to our hypotheses, there were no differences in either accuracy or precision between affected and sound limbs in our post-stroke subjects. However, both sound and affected limb accuracy and precision were significantly different from control
Table 1 Subject characteristics. Persons with post-stroke hemiplegia Subject
Gender
Age (yrs)
Height (m)
Mass (kg)
Leg length (m)
Time post-stroke (yrs)
Affected side
1 2 3 4 5 6 7 8 9 10 11 12 13
M M M F F F M F F M M M M
54.7 51.9 59.1 62.9 58.3 44.9 56.3 63.6 58.4 66.3 50.8 47.1 44.3
1.75 1.95 1.72 1.54 1.66 1.58 1.71 1.60 1.68 1.71 1.79 1.68 1.78
76.0 65.0 66.0 63.0 88.3 56.8 112.5 58.5 67.8 79.5 85.8 93.3 67.4
0.95 1.05 0.92 0.82 0.87 0.82 0.89 0.87 0.92 0.87 0.93 0.83 0.99
1.0 8.4 25.6 19.8 2.4 17.8 18.0 8.4 13.8 16.9 4.9 14.1 16.2
R R R L R R R R L R R L L
Avg SD
67% M
55.3 7.1
1.70 0.11
0.90 0.07
12.9 7.3
69% R
75.4 16.1
Control subjects Subject
Gender
1 2 3 4 5 6
F F F M M M
Avg SD
50% M
Age (yrs) 54.0 58.3 56.3 53.2 58.2 56.0 56 2.1
Avg, average; SD, standard deviation; yrs, years; M, male; F, female; R, right; L, left.
Height (m)
Mass (kg)
Leg length (m)
1.59 1.62 1.53 1.84 1.70 1.75
65.1 65.8 62.3 125.0 79.4 96.8
0.85 0.89 0.82 0.95 0.89 0.93
1.67 0.11
82.4 24.5
0.89 0.05
[(Fig._1)TD$IG]
1100
A. Zissimopoulos et al. / Gait & Posture 39 (2014) 1097–1102
Fig. 1. Frequency of accurate foot placements for post-stroke subjects. Bold horizontal lines represent the median percent of foot strikes for each condition. The 25th and 75th percentiles of the data are graphically represented by the upper and lower edges of the boxes. Whiskers extend to data not considered outliers while crosses represent outliers. (A) Frequency of foot placements lateral to the target for each step width target. (B) Frequency of foot placements on the target line for each step width target. (C) Frequency of foot placements medial to the target for each step width target. Data for the affected limb are shown in the top row and in the bottom row for the sound limb.
subjects. All subjects were instructed to place each foot on the target line or as close as possible, which allowed subjects to prioritize the accuracy of one limb over the other. Results suggest that subjects either did not prioritize one foot over the other, or that the bilateral nature of walking prevented prioritizing accuracy of one foot over the other. These findings corroborate previous
[(Fig._2)TD$IG]
Fig. 2. ML foot placement error measured as distance from target to ankle center and normalized by leg length for all limbs of interest.
reports of bilateral ML foot placement errors during single step tasks [8], extending understanding of post-stroke ML foot placement ability to walking. These findings have implications for post-stroke rehabilitation. Future studies should investigate strategies to improve ML foot placement bilaterally, to evaluate whether training can afford greater adjustability of ML foot placement and, therefore, step width post-stroke. Inspection of individual subject data revealed that four subjects always walked with more accurate affected limb ML foot placement, four always walked with more accurate sound limb ML foot placements and five used a combination of strategies depending upon the target step width. While the location of ML foot placements was constrained in this experimental protocol, subjects were free to achieve the task using any temporospatial, [(Fig._3)TD$IG]kinetic, or kinematic compensations available to them. Future
Fig. 3. Within subject variability in foot placement errors, normalized by leg length.
[(Fig._4)TD$IG]
A. Zissimopoulos et al. / Gait & Posture 39 (2014) 1097–1102
1101
Fig. 4. Change in walking speed between preferred and that during targeted conditions: (A) subjects with post-stroke hemiplegia and (B) control subjects.
work should explore whether subject characteristics such as muscle strength or type of compensatory actions (e.g., hip hiking and circumduction) predict ML foot placement ability. Information regarding predictor variables for ML foot placement ability may help guide rehabilitation strategies and further research. One compensation investigated in the present study was the reduction in walking speed in response to the experimental task (Fig. 4). Of note is the 44% reduction in walking speed, from a median of 0.65 m/s (range 0.62–0.83 m/s) to a median of 0.37 m/s (range 0.28–0.52 m/s) in the 0% LL condition. The dynamic nature of locomotion is questionable at these slow speeds. While this may suggest that dynamic stability is reduced, relationships between walking speed and gait stability remain unclear [14,15]. Medial ML foot placement errors at the widest target step width (Fig. 2, 45% LL condition) may also provide interesting insight regarding stability during ambulation. With each step, the BCoM is transferred laterally toward the support foot, with the BCoM remaining medial or just within the support foot during single support. The gravitational moment (medial to the support foot) is countered by the support hip abductors and a passive rotation torque (Ia) about the hip such that the body remains upright during the single support phase [16]. When transitioning toward the sound limb, the lateral distance the BCoM will travel (and therefore the magnitude of the destabilizing gravitational moment) depends upon lateral momentum supplied by the affected foot. Insufficient lateral momentum toward the support foot could result in a large gravitational moment and premature ‘‘fall’’ back toward the affected foot. Therefore, the widest step width poststroke individuals are able to achieve may be dependent on both affected limb strength and use of compensations to propel the body toward the sound support limb. In summary, the inability to achieve the wide ML foot placements in the 45% LL target condition may reflect an inability to generate sufficient lateral momentum during ambulation, yielding medial errors and narrower step widths. We also cannot rule out the possibility that ML foot placement ability was precluded by abnormal motor coordination. In persons with chronic post-stroke hemiplegia, coupling of hip adduction and knee extension torques has previously been reported during isometric hip torque generating tasks [17], and co-activation of
knee extensors (rectus femoris) and hip adductors (adductor longus) has been reported during isometric stretch reflex experiments [18]. To increase step width, individuals would likely extend their knee and abduct their hip in late swing. It is possible that as subjects attempted to ‘‘reach’’ for the wider target, coupling between knee extension and hip adduction necessitated hip hiking for toe-clearance purposes, reducing the maximum ML foot displacement of the affected foot. However, this manifestation of this abnormal across-joint coupling during gait requires further research. It should be noted that results from this study might not generalize to a more severely affected post-stroke population where foot placement impairments are likely more pronounced. Additionally, the relatively small sample size of the current study may limit generalizability. Finally, results may not apply to subjects using orthoses, canes, or other assistive devices. Despite these limitations, this study provides useful information regarding ML foot placement during post-stroke ambulation. ML motion of the BCoM is related to foot placement and the ability to modify foot placement in response to environmental obstacles and external perturbations is necessary for safe ambulation. While no direct link was established, inaccuracies in ML foot placement during locomotion may be related to the increased incidence of falls during walking reported within the chronic post-stroke population [6,7,19]. Given the particular difficulty subjects had with the narrowest step width condition (0% LL), rehabilitation that targets skills such as tandem walking may be particularly beneficial to this population. Likewise, studies investigating joint torque coupling, sound limb strength, and frontal plane compensations post-stroke may improve understanding of errors in the widest step width condition (45% LL). Further research is needed to investigate the potential mechanisms underlying bilateral ML foot placement impairments post-stroke. Acknowledgements The authors would like to acknowledge use of the Jesse Brown VA Medical Center Motion Analysis Research Laboratory (JBVAMCMARL) for data collection. The authors also acknowledge Richard Harvey, MD, and Elliot Roth, MD, at the Rehabilitation Institute of
1102
A. Zissimopoulos et al. / Gait & Posture 39 (2014) 1097–1102
Chicago (RIC) Center for Stroke Rehabilitation, and Lynda McCracken, CPO, and Christopher Robinson, MBA, CPO, ATC, FAAOP, at the RIC Prosthetics and Orthotics Clinical Center for their assistance with post-stroke subject recruitment and Wendy Murray, PhD, Matthew Tresch, PhD, Yasin Dhaher, PhD, and Andrew Hansen, PhD for their feedback regarding study design. Finally, the authors would like to acknowledge funding support from the Orthotic and Prosthetic Education and Research Foundation (Grant # RFA OPERF-2010-FA-1) (Recipient: Angelika Zissimopoulos), the Dr. John N. Nicholson Fellowship (Recipient: Angelika Zissimopoulos), and the National Institute on Disability and Rehabilitation Research (NIDRR) Department of Education (Grant # H133E080009) (Principal Investigators: Steven Gard and Stefania Fatone). The opinions contained in this publication are those of the grantee and do not necessarily reflect those of the Department of Education. Conflict of interest statement The authors have no conflicts of interest to disclose. References [1] Balasubramanian CK, Neptune RR, Kautz SA. Foot placement in a body reference frame during walking and its relationship to hemiparetic walking performance. Clin Biomech 2010;25:483–90. [2] Kuo AD, Donelan JM. Dynamic principles of gait and their clinical implications. Phys Ther 2010;90:157–74. [3] Redfern MS, Schumann T. A model of foot placement during gait. J Biomech 1994;27:1339–46. [4] Kuo AD. Stabilization of lateral motion in passive dynamic walking. Int J Robot Res 1999;18:917–30.
[5] Roger VL, Go AS, Lloyd-Jones DM, Benjamin EJ, Berry JD, Borden WB, et al. Heart disease and stroke statistics – 2012 update: a report from the American Heart Association. Circulation 2012;125:e2–20. [6] Hyndman D, Ashburn A, Stack E. Fall events among people with stroke living in the community: circumstances of falls and characteristics of fallers. Arch Phys Med Rehab 2002;83:165–70. [7] Jorgensen L, Engstad T, Jacobsen BK. Higher incidence of falls in long-term stroke survivors than in population controls: depressive symptoms predict falls after stroke. Stroke 2002;33:542–7. [8] Nonnekes JH, Talelli P, de Niet M, Reynolds RF, Weerdesteyn V, Day BL. Deficits underlying impaired visually triggered step adjustments in mildly affected stroke patients. Neurorehabil Neural Repair 2010;24:393–400. [9] Kadaba MP, Ramakrishnan HK, Wootten ME. Measurement of lower extremity kinematics during level walking. J Orthop Res 1990;8:383–92. [10] Donelan JM, Kram R, Kuo AD. Mechanical and metabolic determinants of the preferred step width in human walking. Proc Biol Sci 2001;268:1985–92. [11] Fatone S, Hansen AH. Effect of ankle-foot orthosis on roll-over shape in adults with hemiplegia. J Rehabil Res Dev 2007;44:11–20. [12] Saha D, Gard S, Fatone S, Ondra S. The effect of trunk-flexed postures on balance and metabolic energy expenditure during standing. Spine 2007;32:1605–11. [13] Winter DA. Biomechanics and motor control of human movement. 4th ed. New York: Wiley; 2010. [14] Krasovsky T, Lamontagne A, Feldman AG, Levin MF. Reduced gait stability in high-functioning poststroke individuals. J Neurophysiol 2013;109:77–88. [15] Hak L, Houdijk H, Steenbrink F, Mert A, VanderWurff P, Beek PJ, et al. Speeding up or slowing down? Gait adaptations to preserve gait stability in response to balance perturbations. Gait Posture 2012;36:260–4. [16] MacKinnon CD, Winter DA. Control of whole body balance in the frontal plane during human walking. J Biomech 1993;26:633–44. [17] Cruz TH, Dhaher YY. Evidence of abnormal lower-limb torque coupling after stroke: an isometric study. Stroke 2008;39:139–47. [18] Finley JM, Perreault EJ, Dhaher YY. Stretch reflex coupling between the hip and knee: implications for impaired gait following stroke. Exp Brain Res 2008;188:529–40. [19] Weerdesteyn V, de Niet M, van Duijnhoven HJ, Geurts AC. Falls in individuals with stroke. J Rehabil Res Dev 2008;45:1195–213.
