Gait & Posture 35 (2012) 322–327
Contents lists available at SciVerse ScienceDirect
Gait & Posture journal homepage: www.elsevier.com/locate/gaitpost
Gait asymmetries in children with cerebral palsy: Do they deteriorate with running? Harald Bo¨hm *, Leonhard Do¨derlein Orthopaedic Hospital for Children, Behandlungszentrum Aschau GmbH, Bernauerstr. 18, 83229 Aschau i. Chiemgau, Germany
A R T I C L E I N F O
A B S T R A C T
Article history: Received 27 January 2011 Received in revised form 30 September 2011 Accepted 9 October 2011
In children with cerebral palsy (CP) analysis of gait asymmetry can provide insight into the control of walking and may help in guiding the clinician’s treatment decisions. Running is more difficult that walking for the musculoskeletal system, however, in the literature it has been shown that gait deviations associated with CP maybe better tolerated during running. This leads us to the hypothesis that running might increase gait symmetry in patients with CP. Therefore the purpose of this study was to investigate the effect of running on asymmetries in spatio-temporal, kinematic and kinetic gait parameters for children with CP. Twenty-four children with diplegia and 25 with hemiplegia were examined using 3D gait analysis during running and walking. MANOVA on two factors: diagnosis (hemiplegic, diplegic) and movement (walking, running) was conducted on a total of 22 gait parameters. The MANOVA revealed a significant difference in symmetry between walking and running (p < 0.001) and between patients groups (p = 0.004). The detailed analysis of gait parameters demonstrated a significant decrease of symmetry in 13 of the 22 gait parameters investigated, only symmetry of step time was significantly increased. Therefore the hypothesis that gait symmetry improved with running in children with CP can be rejected. Based on the results of this study, asymmetries masked during walking might appear during running. Therefore, analysis of asymmetry of walking and running gives a more comprehensive assessment of the gait pathology for clinical decision making. ß 2011 Elsevier B.V. All rights reserved.
Keywords: Running Asymmetry Gait analysis Cerebral palsy
1. Introduction Symmetry is an important gait feature that is increasingly measured and reported, particularly in hemiplegic patients [1,2] and single leg amputees [3] where one limb is predominantly affected. Moreover asymmetrical behavior of the lower limbs is not limited to impaired population. Asymmetry exists as well in ablebodied population and was found to reflect natural functional differences between the lower extremities [4]. These functional differences were explained by the fact that one leg is mainly responsible for balance and support and the other leg contributes more to propulsion during walking. Clinically, gait asymmetry is important since it may be associated with a number of negative consequences. These are challenges to control balance [1], risk of overuse of the healthy limb, loss of bone mass density and shorter limb length in the involved lower extremity [5,6]. In observational clinical gait
* Corresponding author at: Gait Laboratory, Orthopaedic Hospital for Children, Behandlungszentrum Aschau GmbH, Bernauerstr. 18, 83229 Aschau i. Chiemgau, Germany. Tel.: +49 8052 171 2016. E-mail address:
[email protected] (H. Bo¨hm). 0966-6362/$ – see front matter ß 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.gaitpost.2011.10.003
analysis, symmetry can provide a unique insight into the control of walking, which may differ from more conventional measures such as velocity. This may help in guiding the clinician’s treatment decisions [1,7]. Patients might compensate for their asymmetric pathologies. Therefore instrumented gait analysis shows an abnormal but symmetric gait pattern during walking. Running creates higher loads on the musculo-skeletal-system than walking and therefore requiring a more complex interaction of the neuromuscular system [8]. As a consequence, asymmetries masked during walking might appear in running. Increase of inter-limb asymmetry in running compared to walking has been observed in amputees with prostheses [3,9,10] supporting the assumption that running increases bilateral gait asymmetries. Running is an important functional activity for children with cerebral palsy (CP) to keep up with their peers and therefore it requires particular attention. In one previous study it has been shown that kinematic and kinetic profiles at the ankle joint in children with CP were more similar to normal profiles in running than in walking [11]. This observation suggests that the gait deviations associated with CP maybe better tolerated during running, which leads us to the hypothesis that running might increase gait symmetry in patients with CP. Therefore, the purpose
H. Bo¨hm, L. Do¨derlein / Gait & Posture 35 (2012) 322–327
323
2.3. Data processing and statistical analysis
of this study was to investigate asymmetries in spatio-temporal, kinematic and kinetic gait parameters during walking and running in children with CP.
In the literature, running technique is often defined by the way the foot collides with the ground [12]. In this study collision was defined as forefoot or rear-foot strike. This was determined by the orientation of the heel–toe vector with respect to the walking direction, which was clockwise for forefoot strike and counterclockwise or equal for rear-foot strike. Spatio-temporal gait parameters investigated were step length, step time, and duration of stance, normalized to non-dimensional units [13]. Peak segment angles of the pelvis as well as peak joint angles of hip, knee, and ankle in the sagittal plane were determined for each trial. Further, joint energies of hip, knee and ankle joint during the stance phase were calculated separately for generation and absorption energies by the time integral of the positive and negative joint power curves, respectively. Asymmetry for all gait parameters was calculated for each of the three trials, and was then averaged over three trials. To account for differences in angles curves shapes, root mean square differences (RMS) between sides were calculated for pelvic tilt, hip, knee and ankle flexion. Multivariate analysis of variance (MANOVA) on two factors: diagnosis (hemiplegic, diplegic) and movement (walking, running) was conducted on a total of 22 gait parameters followed by separate ANOVAs on each of the gait parameters. Statistical significance was set at to a = 0.05. Bonferroni correction of a-level was performed to control for multiple tests.
2. Methodology 2.1. Subjects and data collection Forty-nine children and adolescents with CP volunteered to participate in this study. The children had been referred to the Gait Analysis Laboratory at the Orthopedic Hospital in Aschau for evaluation of their gait and assessment for useful orthotic or orthopedic interventions. The children provided written consent, as approved by the local ethics committee. To be included, the children had to be community-level ambulators, using no assistive devices (GMFCS I & II). Further, they should show a phase of double float where neither foot touches the ground while running. All gait and running studies were performed with the children ambulating barefoot. Exclusion criteria were previous Botulinum-toxin injections or previous casting of the lower limbs within 6 months as well as any previous surgical procedures. Twenty-four children had spastic diplegia, age 12.5 years (SD = 3.7), body height 148 cm (SD = 16), body mass 38 kg (SD = 11) and 17/24 were males. In 10/24 diplegic children one limb was diagnosed to be more involved than the other. Twenty-five children had hemiplegia, 12 years (SD = 4), 150 cm (SD = 18), 45 kg (SD = 22) and 18/25 were males. Gait analysis was performed using an eight-camera Vicon MX system with two force plates. The Vicon Plug-in-Gait marker set was used to generate kinematic and kinetic data. The subjects were asked to walk and run at comfortable speed down the 15 m walkway. Five walking and running trials with valid kinetics were collected. The first three consistent trials, determined by visual observation of joint kinematic and kinetics, were used for further data processing.
3. Results Gait velocity was 1.1 m/s (SD = 0.2) for walking and 2.3 m/s (SD = 0.4) for running in both hemi and diplegic patients. During running bilateral forefoot initial contact was observed in 79% and 36% of diplegic and hemiplegic patients, respectively. Unilateral forefoot contact was observed in 21% of diplegic and 64% of hemiplegic patients. In all hemiplegic patients the sole angles at initial contact were lower at the involved side. Asymmetries of gait parameters are shown in Table 1. The MANOVA revealed a significant multivariate main effect for diagnosis (hemiplegic, diplegic), Wilks’ l = 0.589, F(22,71) = 2.31, p = 0.004, and for movement (walk, run) l = 0.344, F(22,71) = 6.05, p < 0.001. The interaction between factors was not significant l = 0.726, F(22,71) = 1.22, p = 0.261. The result of the separate univariate ANOVAs on each of the gait parameters is shown in Table 1. The gait parameters did not show
2.2. Quantification of asymmetry It has been shown that common indices to quantify gait symmetry did not provide any unique differences or contributions to distinguish patients [1]. Further usage of a symmetry index may have major limitations, because differences are reported against their average value. For example, if a large asymmetry is present, the average value does not correctly reflect the performance of either limb. Also, parameters that have large values but relatively small inter-limb differences will tend to lower the index and reflect symmetry [4]. Because of these limitations we used absolute difference in gait parameters between the left and right side without normalization to average values.
Table 1 The asymmetry of 22 gait parameters for hemiplegic and diplegic patients during walking and running is shown. The asymmetry was defined as the absolute difference between the right and left side. Parameters showing an increase in symmetry with running are shown in bold numbers; statistical significance is indicated with an asterisk. ANOVA’s p-values
Asymmetry Hemiplegic Walk Spatio-temporal parameters Steplength 8.7 (5.6) Steptime 0.3 (0.2) Stance [% gaitcycle] 5.8 (3.2) Peak joint angles [8] Pelvis forward tilt stance 1.3 (0.9) Pelvis backward tilt stance 1.2 (1.1) Hip flexion swing 6.4 (5.5) Hip extension 5.2 (4.5) Knee flexion loading response 6.9 (4.5) Knee extension 8.0 (4.7) Knee flexion swing 11.4 (9.8) Ankle dorsiflexion 6.9 (6.4) ankle plantarflexion 7.7 (9.3) Root mean square differences between sides [8] Pelvic tilt 4.4 (2.4) Hip flexion 8.9 (3.0) Knee flexion 9.6 (2.4) Ankle flexion 8.2 (4.8) Joint energies [mJ/kg] Hip generation 111 (81) Hip absorption 51 (46) Knee generation 50 (53) Knee absorption 114 (72) Ankle generation 116 (93) Ankle absorption 58 (48)
Diplegic Run
Walk
7.7 (6.2) 0.1 (0.1) 4.3 (3.5)
Run
Factor 1
Factor 2
Diagnose
Movement
5.5 (4.9) 0.2 (0.2) 3.0 (3.3)
9.6 (7.3) 0.1 (0.1) 3.3 (2.6)
0.653 0.053 0.016
0.139 <0.001* 0.244
Interaction
0.347
2.7 2.5 6.3 7.8 12.0 5.5 11.4 11.3 8.6
(2.3) (1.9) (3.7) (4.4) (8.2) (4.3) (6.6) (8.0) (6.5)
1.3 0.9 4.9 4.9 5.5 6.5 8.8 4.5 6.7
(0.8) (0.7) (5.0) (5.2) (3.6) (5.2) (9.1) (5.1) (6.3)
3.4 2.2 5.1 7.4 9.0 8.6 6.8 6.3 7.5
(2.2) (1.5) (3.5) (7.3) (6.4) (6.5) (5.7) (7.2) (7.2)
0.367 0.218 0.164 0.796 0.096 0.532 0.063 0.048 0.576
<0.001* <0.001* 0.946 <0.001* <0.001* 0.832 0.466 <0.001* 0.443
0.343 0.959
5.8 11.9 12.9 11.5
(3.1) (5.2) (4.8) (6.2)
3.5 6.0 7.6 6.0
(1.8) (3.1) (4.0) (3.4)
4.3 9.1 10.5 8.4
(1.7) (4.7) (4.7) (5.1)
0.052 0.012 0.036 0.056
0.001* <0.001* <0.001* <0.001*
0.271 0.963 0.732 0.349
0.084 0.468 0.588 0.008 0.003 0.575
0.337 <0.001* <0.001* 0.040 0.001* <0.001*
125 92 152 160 178 111
(107) (100) (139) (160) (121) (84)
75 25 48 59 59 62
(83) (19) (47) (52) (52) (54)
93 148 130 84 100 89
(76) (138) (131) (88) (88) (75)
0.935 0.492
0.086
0.010 0.615 0.469 0.257
324
H. Bo¨hm, L. Do¨derlein / Gait & Posture 35 (2012) 322–327
Fig. 1. Joint angles for hemi- and diplegic patients during walking and running. Shown are average values of the involved and healthy or less involved side. In case of symmetric diplegic patients the right side was arbitrarily defined as involved. The vertical lines indicate the duration of stance phase.
significant differences in the asymmetry between hemi and diplegic patients. Significant differences in asymmetry between walking and running were observed for step time, pelvis peak forward and backward tilt, hip peak extension, knee peak flexion in loading response and ankle peak dorsiflexion. Further all RMS differences between sides were significantly increased, as well as the knee and ankle generation energies and the hip and ankle absorption energies. The interaction effect (movement diagnosis) was not significant for any of the significant parameters. All significant gait asymmetries were greater for running compared to walking for both, hemiplegic and diplegic patients, except for the step time, which asymmetry decreased during running. Figs. 1 and 2 as well as Table 2 contain further information about the sign of the gait asymmetries with respect to the involved or more involved leg. For example 88%, 83% and 71% of the
hemiplegic patients had lower ankle, knee and hip generation energy at the involved side, respectively. 4. Discussion The multivariate statistical analysis on 22 gait parameters revealed a significant difference in symmetry between walking and running. The detailed analysis of gait parameters demonstrated a significant decrease of symmetry in 13 of the 22 investigated gait parameters. Only symmetry of step time was significantly increased. Therefore the hypothesis that gait symmetry improved with running in children with CP can be rejected. 4.1. Running performance in children with CP In this study all diplegic patients as well as most of the hemiplegic patients performed running by making initial contact
H. Bo¨hm, L. Do¨derlein / Gait & Posture 35 (2012) 322–327
325
Fig. 2. Joint power curves for hemi and diplegic patients during walking and running. Shown are average values of the involved and healthy or less involved side. In case of symmetric diplegic patients the right side was arbitrarily defined as involved. The vertical lines indicate the duration of stance phase. Joint energies in generation and absorption were obtained by the time integral over the positive and negative power curves.
with the forefoot. In contrast most of the shod runners today make initial contact with the ground heel first. However, experienced barefoot runners avoid heel-first contact, because it hurts owing to repetitive, high-impact forces [12]. Since the patients in this study ran barefoot they comply with normal population under barefoot condition. This running technique with initial forefoot strike complies with the ankle equinus pathology in these patients group reported by Davids et al. [11]. Further, in agreement with the study of Davids et al., peak ankle plantarflexion, knee and hip flexion in swing increased during the progression from walking to running. At the same time peak extension at the knee and hip joints were reduced towards an increased crouch position, whereas peak ankle dorsiflexion increased with running. However the increase in dorsiflexion was only 18 in the study of Davids et al. and about 68 in this study. The reason for the increased dorsiflexion angle reported in this study might be due to the older age (4 years) and heavier body-mass (12 kg) of the patients in this study. The increased body
weight might generate a higher ankle dorsiflexion during running. In addition, the Plug-in-Gait model can overestimate peak ankle dorsiflexion when there is an increased dorsiflexion of the midfoot [14]. This might be the case due to the development of midfoot break in the older children. 4.2. Asymmetry of spatio-temporal parameters In the study of Davids et al. [11], the change in cadence from walking to running was significantly greater in diplegic than in healthy children. Therefore the authors suggest that cadence rather than step length is the preferred option for children with CP to change from walking to running. Better running performance at higher cadence was also shown in this study, because the symmetry of step time was significantly increased from walking to running. One explanation for this behavior might be the margin of stability [15] which increased with smaller step time due to
H. Bo¨hm, L. Do¨derlein / Gait & Posture 35 (2012) 322–327
326
Table 2 The number of patients with positive asymmetry values is shown. The sign of the asymmetry was determined by the difference between the healthy and the involved side. In consequence a positive sign of a gait asymmetry value was obtained when the gait parameter of the involved or more involved leg was lower than the healthy or less involved leg. Hemiplegic (n = 25)
Asymmetric diplegic (n = 10)
Positive sign walk [%]
Positive sign run [%]
Positive sign walk [%]
Positive sign run [%]
46 13 88
50 30 80
10 30 70
33 63 42 13 79
80 30 60 30 50
30 40 30 0 50
58 33 88 58
50 70 60 50
50 70 90 30
63 67 83 83 79 50
70 60 30 80 80 60
70 80 70 40 90 60
Spatio-temporal parameters Steplength 67 Steptime 4 Stance 100 Peak joint angles Pelvis forward tilt stance 50 Pelvis backward tilt stance 29 Hip flexion swing 67 Hip extension 29 Knee flexion loading 58 response 46 Knee extension Knee flexion swing 71 Ankle dorsiflexion 79 Ankle plantarflexion 46 Joint energies Hip generation 71 Hip absorption 83 Knee generation 83 Knee absorption 83 Ankle generation 88 Ankle absorption 46
smaller step length. However step length asymmetry was not significantly changed is this study, so that this observation requires further investigation. In 96% and 87% of the hemiplegic patients a shorter step time was observed at the affected side in walking and running, respectively. A reason for this behavior might be the prolonged duration of the non-paretic limb stance phase shown in the vertical lines in Figs. 1 and 2, in order to load the affected leg for as short time as they can. 4.3. Asymmetry of joint energies Running required significantly more concentric energy in the ankle and knee joints during the stance phase (Fig. 2). With the increased joint energy required, the asymmetry at the ankle and knee joints was also significantly increased. Here the majority of patients showed a lower concentric joint energy at the more involved side. This suggests that the muscles of the more involved side cannot produce more joint energy and therefore an increase in asymmetry was observed. Similar observations were found for the eccentric energy at the hip joint. Whereas for the eccentric energy at the ankle joint only 50% of the hemiplegic patients, absorbed less energy at the more involved side. This might be explained by the initial forefoot contact in 64% of the hemiplegic patients, which caused considerable energy absorption at the ankle joint during loading response shown (Fig. 2). 4.4. Asymmetry of joint and segment angles All RMS differences between both sides increased significantly. Detailed analysis of peak joint angles allows for further discussion in this section. At the ankle joint, peak dorsiflexion asymmetry increased due to the greater excursion on the healthy or less involved side. This might be explained by contractures of the involved sides that limit
further joint excursion. As a result the calf muscles of the involved side had to operate at a shorter length. This reduces the force generation abilities of the contractile elements of the calf muscles [16], which might explain the reduced power generation at the involved side. At the knee joint, the asymmetry of peak knee flexion during loading response increased significantly. In hemiplegics the healthy side showed increased knee flexion at loading response during running which might have allowed for more energy absorption, which can be clearly seen at the average power curves in Fig. 2. However due to considerable individual differences seen in the high standard deviations (Table 1), knee absorption power asymmetry was not significantly increased. Hip extension asymmetry significantly increased in hemi and diplegic patients. All asymmetric diplegic patients and 87% of the hemiplegics showed less hip extension on the involved side. The cause might be either contractures or spasticity of the hip flexors on the more involved side. Spasticity per definition is velocitydependent [17], and the increase during running is expected. The reduced hip extension at the involved side occurred at the end of the stance phase; therefore it can be linked to less power absorption of the involved hip at the same time (Fig. 2). Pelvic anterior and posterior tilts were significantly increased. In particular the anterior tilt is suggested to be a cause of hip flexor (psoas) dysfunction which demonstrates a typical double and single bump pelvic pattern for diplegic and hemiplegic patients, respectively [18]. An increased anterior tilt at the involved side was found in 68% and 70% of hemi- and asymmetric diplegic patients, respectively. This increased magnitude of the pelvis might serve as another proximal power input joint as a way to propelling the swing limb forward [19] when more power was required during running. 4.5. Difference between hemi and diplegic patients The MANOVA revealed a significant multivariate difference between hemiplegic and diplegic patients. This is not surprising, since in hemiplegic patients only one leg is affected. However univariate ANOVAs corrected for multiple testing to a = 0.0023 did not reveal any significant differences in any of the gait parameters. However for six parameters a trend at a = 0.05 significance level could be observed that asymmetries were greater for hemiplegic patients. These parameters were duration of stance, ankle dorsiflexion, RMS of hip and knee angles, knee absorption and ankle absorption energies. The reason for the missing significance in univariate statistical testing might be that 10/24 diplegic patients were diagnosed to be asymmetric and their asymmetry curves profiles were similar to hemiplegic patients (Figs. 1 and 2). In this study asymmetry was determined by clinical observation, and to our best knowledge there is no valid criteria based on instrumented gait analysis to distinguish between asymmetric and symmetric diplegic patients. All significant differences in asymmetry between walking and running were seen in both patients groups, and there were no significant interactions between groups. Therefore the results of this study suggest that running increase the asymmetry in the same way in hemi- and diplegic patients. In Sections 4.2 and 4.3, the increase in asymmetry during running could be related to the involved or more involved side and thus running might reveal additional information about the impairment of the more involved side. On the other hand compensation mechanisms of the healthy side with the goal to increase gait symmetry might not work during running, since higher loads must be generated by the healthy side to enable adequate running speeds. In clinical gait analysis, compensation mechanisms to increase symmetry are the most difficult gait deviations to assess, e.g. increased knee flexion during stance phase of the less involved
H. Bo¨hm, L. Do¨derlein / Gait & Posture 35 (2012) 322–327
side. Therefore the increased asymmetry in running might be particularly useful to uncover these compensation mechanisms. 5. Conclusion In the literature it has been suggested that gait deviations associated with CP maybe better tolerated during running [11]. However this study demonstrated that gait asymmetry increased significantly from walking to running. The increase of asymmetry could be mainly associated to impairments (spasticity, contractures or weaknesses) of the involved or more involved side. Asymmetries masked during walking might appear during running since running might create higher loads on the musculoskeletal system. Therefore analysis of asymmetry of walking and running gives a more comprehensive assessment of the gait pathology for surgical decision making. Running therapy with the aim to establish a symmetric gait pattern might be more effective than walking therapy because it enables the therapist to more easily identify origin and magnitude of asymmetries. Conflict of interest statement All authors do not have any financial and personal relationships with other people or organizations that inappropriately influence the work performed. References [1] Patterson KK, Gage WH, Brooks D, Black SE, McIlroy WE. Evaluation of gait symmetry after stroke: a comparison of current methods and recommendations for standardization. Gait Posture 2010;31(2):241–6. [2] Ko PS, Jameson PG, Chang TL, Sponseller PD. Transverse-plane pelvic asymmetry in patients with cerebral palsy and scoliosis. J Pediatr Orthop 2011;31(3):277–83.
327
[3] Burkett B, Smeathers J, Barker T. Walking and running inter-limb asymmetry for Paralympic trans-femoral amputees: a biomechanical analysis. Prosthet Orthot Int 2003;27(1):36–47. [4] Sadeghi H, Allard P, Prince F, Labelle H. Symmetry and limb dominance in ablebodied gait: a review. Gait Posture 2000;12:34–45. [5] Lin PP, Henderson RC. Bone mineralization in the affected extremities of children with spastic hemiplegia. Dev Med Child Neurol 1996;38(9): 782–6. [6] Riad J, Finnbogason T, Brostro¨m E. Leg length discrepancy in spastic hemiplegic cerebral palsy: a magnetic resonance imaging study. J Pediatr Orthop 2010;30(8):846–50. [7] Salazar-Torres JJ, McDowell BC, Kerr C, Cosgrove AP. Pelvic kinematics and their relationship to gait type in hemiplegic cerebral palsy. Gait Posture 2011;33(4):620–4. [8] Arampatzis A, Bru¨ggemann GP, Metzler V. The effect of speed on leg stiffness and joint kinetics in human running. J Biomech 1999;32(12):1349–53. [9] Isakov E, Burger H, Kranjik K, Gregoric M, Marinek C. Influence of speed on gait parameters and on symmetry in trans-tibial amputees. Prosthet Orthot Int 1996;20:153–8. [10] Prince F, Allard P, Therrien RG, Mc Fayden BJ. Running gait impulse asymmetries in below-knee amputees. Prosthet Orthot Int 1992;16:19–24. [11] Davids JR, Bagley AM, Bryan M. Kinematic and kinetic analysis of running in children with cerebral palsy. Dev Med Child Neurol 1998;40(8):528–35. [12] Lieberman DE, Venkadesan M, Werbel WA, Daoud AI, D’Andrea S, Davis IS, et al. Foot strike patterns and collision forces in habitually barefoot versus shod runners. Nature 2010;463(7280):531–5. [13] Hof AL. Scaling gait data to body size. Gait Posture 1996;4:222–3. [14] MacWilliams BA, Cowley M, Nicholson DE. Foot kinematics and kinetics during adolescent gait. Gait Posture 2003;17(3):214–24. [15] Hof AL, Gazendam MG, Sinke WE. The condition for dynamic stability. J Biomech 2005;40:451–7. [16] Bo¨hm H, Cole GK, Bru¨ggemann GP, Ruder H. Contribution of muscle series elasticity to maximum performance in drop jumping. J Appl Biomech 2006;22:3–13. [17] Engsberg JR, Ross SA, Olree KS, Park TS. Ankle spasticity and strength in children with spastic diplegic cerebral palsy. Dev Med Child Neurol 2000;42(1):42–7. [18] Gage JR, Schwartz MH, Koop SE, Novacheck TF, editors. Clinics in developmental medicine no 180–181, the identification and treatment of gait problems in cerebral palsy. 2nd edition, London: MacKeith Press; 2009. 451 p.. [19] Miller F. Physical therapy of cerebral palsy. NY: Springer Science + Business Media LLC; 2007, 296 p..