Available online at www.sciencedirect.

com

Gait & Posture 28 (2008) 150–156

www.elsevier.com/locate/gaitpost

Effects of plantarflexion on pelvis and lower limb kinematics

R. Brunner a,*, T. Dreher c, J. Romkes b, C. Frigo d
a
Neuro-Orthopaedic Department, University Children’s Hospital, Roemergasse 8, 4058 Basel, Switzerland
b
Laboratory for Movement Analysis, University Children’s Hospital, Basel, Switzerland
c
Department of Orthopaedic Surgery, Heidelberg, Germany
d
Department of Bioengineering, Politecnico, Milano, Italy
Received 15 April 2007; received in revised form 4 November 2007; accepted 18 November 2007

Abstract

Modelling the effect of soleus and gastrocnemius contractions against the floor resistance in a forward dynamics simulation revealed that
hip flexion, internal rotation and adduction together with external pelvic rotation could be attributed to a direct, but distant effect of triceps
surae contraction. Knee flexion smoothed out the effect. To validate this clinically relevant biomechanical observation, ankle plantar flexion
was correlated with hip and pelvic rotation retrospectively in children with spastic cerebral palsy.
In 49 children with spastic hemiplegia, plantar flexion showed a significant correlation with increased pelvic retraction and hip internal
rotation. In contrast, in 47 children with spastic diplegia no significant effect of the triceps surae on hip and pelvis kinematics was found.
Bilateral hip and knee flexion in diplegia appeared to prevent the proximal effect of the triceps surae seen in the hemiplegics. In diplegia
triceps surae overactivity did not appear to be a significant cause of internal rotation gait.
# 2007 Elsevier B.V. All rights reserved.

Keywords: Soleus gastrocnemius function; Biomechanical modelling; Hip internal rotation; Pelvic retraction; Spastic plantarflexion

1. Introduction remained inconclusive [8]. Delp et al. suggested possible

Internally rotated gait is not fully understood. In
hemiplegic and diplegic cerebral palsy (CP) gait is often
characterised by ankle equinus, hip internal rotation and pelvic
retraction (external rotation) [1,2]. Pelvic retraction is
commonly seen as a compensation for the increased internal
hip rotation [3]. These proximal deviations, however, may
occur even when there is only calf muscle involvement.
Interestingly, plantarflexion together with hip internal rotation
and pelvic retraction is also seen in patients with non-neurolo-
gical conditions, such as congenital absence of the anterior
cruciate ligament or habitual toe walking. This suggests that
the underlying mechanism is not necessarily neurological.
Results after surgical correction of femoral torsion for
internally rotated gait are sometimes disappointing [4–7]
which suggest a multifactorial origin of this problem. The
gait pattern itself has been accused but the study results
mechanisms, such as hip flexion deformity that increases
internally rotating moments, or compensatory internal
rotation for weak hip abductors [9,10].
Characteristically in equinus gait, the triceps surae
muscle shows an onset of activation in late swing and a
prolonged, enhanced activity during stance [11,12]. The
effect of such overactivity on hip and pelvic position was
modelled for various positions of the leg in space using a
forward dynamics simulation. It was hypothesised that
overactive plantarflexors would result in hip flexion/internal
rotation and pelvic retraction. The clinical relevance of the
results from this theoretical approach was then tested in
patients with hemiplegic and diplegic CP.
2. Methods
2.1. Modelling

2.1.1. General structure of the model

* Corresponding author. Tel.: +41 61 685 55 24; fax: +41 61 685 50 12. The model was composed of 13 geometrical solids representing
E-mail address: reinaldbrunner@dplanet.ch (R. Brunner). the main anatomical segments of the human body (Table 1). A

0966-6362/$ – see front matter # 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.gaitpost.2007.11.013
 R. Brunner et al. / Gait & Posture 28 (2008) 150–156 151

Table 1 model because of their dynamic coupling. The displacement of the

Antropometric parameters of the human body model segments was computed by double integration (Kutta–Merson
Mass Mass Size numerical method) starting from the initial kinematic condition
ratio (%) (Kg) (m) and taking into account the ground reaction forces as an external
Whole body 100 45 h = 1.60 constraint. The ground reaction forces resulted from a ‘non-COM
Head 7.3 3.28 r = 0.10 penetration’ constraint between foot and ground. As a consequence
of the movement the spring representing the activated muscle
Trunk and pelvis 50.7 22.81
shortened and the force vanished to zero. The viscous damping
Trunk 34.4 15.70 l = 0.18 w = 0.30 h = 0.30
Pelvis 15.8 7.11 l = 0.12 w = 0.24 h = 0.16 coefficient of all the spring/damper elements was set to 1 Ns/m to
smooth out the movement and to achieve a new steady state without
Upper limb 4.9 2.20 oscillations.
Upper arm 2.6 1.17 r = 0.04 l = 0.28
Forearm and hand 2.3 1.03 r = 0.035 l = 0.30
Lower limb 16.1 7.24 2.1.3. Simulation conditions
Thigh 10.3 4.63 r = 0.06 l = 0.35 As a first condition gravity was excluded from our computation,
Shank 4.4 1.98 r = 0.045 l = 0.36 so that internal muscle forces were not required to keep the system
Foot 1.5 0.67 l = 0.24 w = 0.08 h = 0.06
in its static equilibrium. This assumption was justified by two
Mass ratio: percentage of segment mass in relation to body mass; h, height; reasons: (1) the induced acceleration depends on the force pertur-
r, radius; l, length; w, width. bation only, and so it is the same either if the system was in
equilibrium under gravity (which requires internal forces) or if it
was completely unloaded, (2) to compute the muscle forces in the
hypothetical subject, 160 cm tall, 45 kg of body mass, was con- presence of gravity (or any other loading condition) would require
sidered, and size and mass of anatomical segments were defined to assume a criterion of sharing the internal forces among the
according to Clauser et al. [13]. The mass density was uniform, so various muscles, because the system was mathematically undeter-
that centre of mass and central moments of inertia could be defined mined.
by simple geometrical calculations. A second condition was that all initial velocities and accelera-
A simulation software was used to implement the model and tions were set to zero so that the effect of the force perturbation was
to calculate the forward dynamics (Working Model 3D, MSC- interpreted as a deviation with respect to the previous, unperturbed
Software). The solid segments were connected by means of trajectory.
rotational constraints: spherical hinges at the shoulders and at The initial configuration of the model was defined to corre-
the hips (3 degrees of freedom), and cylindrical hinges at the spond with standing up-right posture, and to the conventional
elbows, knees and ankles (revolute joints, 1 degree of freedom). descriptors of stance phase [14]. The corresponding joint angles
Between the head and trunk and between the trunk and pelvis the were set to normal values for the sub-phases of stance phase and
upper segment was constrained to slide along an axis that was initial swing [14]. An exaggerated plantarflexion was simulated
hinged to the lower segment by a spherical joint. The relative producing forefoot ground contact to evaluate the effect of an
position of the upper and lower segments was controlled by four equinus ankle.
spring/damper elements. At time T = 0 the natural length of the spring representing the
The knee joint axes were oriented horizontally and parallel to soleus or the gastrocnemius was changed (reduced) by 50 mm. The
the coronal plane with a valgus of 58. The shank longitudinal axis stiffness of the spring was 200 N/m, and the resulting initial force
was vertical in standing position. The ankle axis was horizontal and 10 N. When the two muscles were activated simultaneously, the
perpendicular to the longitudinal foot axis which was externally stiffness of each one was reduced to 100 N/m to make the effect
rotated by 158. comparable to the single muscle contraction. The mechanical work,
DL = 1/2KDL2 corresponded to 0.25 J both in single and in simul-
2.1.2. Muscle actuators taneous muscle contraction.
The muscles were represented by mechanical spring/damper Similarly the effect of soleus and gastrocnemius was tested for a
elements. The two components of gastrocnemius were represented hypothetical configuration reproducing a typical crouch gait pos-
by just one of these elements, running from the thigh above the knee ture.
to the rear foot. The soleus was represented by another spring/ To assess the effects of the different muscles’ contractions,
damper element running from the shank to the same attachment pelvis and thigh orientation, knee and ankle joint angles, and foot
point of gastrocnemius on the rear foot. The origin and insertion of inclination in relation to the ground plane were analysed after
both gastrocnemius and soleus lay in the sagittal plane (soleus complete recoil of the spring (end of muscle contraction).
origin 0.02 m posterior to the longitudinal shank axis, 0.14 m distal
to the knee joint centre, gastrocnemius origin 0.06 m posterior to 2.2. Clinical study
the longitudinal thigh axis, 0.023 m proximal to the knee joint
centre; insertion of Achilles tendon 0.08 m posterior to, and, 0.06 m The hypothesis of a functionally short triceps resulting in hip
distal from the ankle joint centre). flexion/internal rotation and pelvic retraction was tested retrospec-
The activation of the muscles was simulated by shortening the tively in patients with hemiplegic or diplegic CP. Inclusion criteria
natural length of the spring. In this way the initial force abruptly to consider the affected leg were
rose to F = KDL (where K was the elastic constant and DL was the
change between the actual spring length and the new length). This  hemiplegic or diplegic CP,
force perturbation induced accelerations in all segments of the  a full instrumented standardised gait analysis,
152 R. Brunner et al. / Gait & Posture 28 (2008) 150–156
 equinus deformity (at least 58 of plantar flexion at initial contact
as seen in 3 days gait analysis), and
 abnormal medial gastrocnemius activity in terminal swing from
dynamic EMG raw data (the medial gastrocnemius was assumed
to be representative of triceps surae).
Fig. 1. Effect of single and simultaneous contraction of soleus and gastro-
cnemius when the starting position is the one depicted in the first column
(gait phases: IC = initial contact, LR = loading response, MS = mid-stance,
LS = late stance, PS = Preswing, TO = toe off). The first column of pictures
shows the position before muscle contraction according to the position of
segments in the gait phases, the following columns the resulting position of
the body segments after contraction of the muscle mentioned in the heading.
Only the lateral view is shown.
Exclusion criteria were
 previous Botulinum-toxin A injections within 6 months,
 previous casting of the lower limbs within 12 months and
 any previous surgical procedures.
All children were independent walkers. Forty-seven children had
spastic diplegia (80 involved legs, age 11.4  3.8 years, body mass
37.2  13.9 kg), 49 hemiplegia (49 involved legs, age 12.8  5.6
years, body mass 43.7  17.4 kg). All patients/parents gave written
consent for research purposes in accordance with local ethical
committee requirements.
2.2.1. Gait analysis
All patients underwent a full instrumented gait analysis between
the years 2000 and 2005, using a motion capture system (six camera
VICON 460 system, Oxford Metrics Ltd., UK), two force plates
(Kistler Instrumente AG, Winterthur, Switzerland) and an eight
channel surface EMG system (Zebris, Tübingen, Germany). The
patients walked at their self-selected speed. The Helen Hayes
Marker set [15] was used and at least six trials were recorded.
Anthropometric data were recorded for appropriate scaling. Sur-
face EMG was recorded simultaneously. Bipolar Ag/AgCl surface
electrode pairs (electrode diameter 10 mm and an inter-electrode
spacing of 22 mm) were placed bilaterally over the medial gastro-
cnemius, tibialis anterior, rectus femoris, and semitendinosus
muscles. For electrode placement, the SENIAM [16] recommenda-
tions for surface EMG were followed. The ground electrode was
placed over the tibial tuberosity. The EMG signals were band-pass
filtered (10–700 Hz) and collected at a sampling rate of 2500 Hz.
All data were expressed in percentage of gait cycle using the
Polygon software (Oxford Metrics Ltd., UK).
2.2.2. Data evaluation
Hip rotation, pelvic rotation and plantarflexion (mean values
over the whole stance phase and during mid-stance) were evaluated.
Mid-stance was studied because an increase of the plantar flexor
moment compared to normals is typically located during this phase.
Mean and standard error of the mean (S.E.M.) were calculated for
each parameter and each patient group (hemiplegics and diplegics)
and compared.
2.2.3. Statistical analysis
Spearman Rank Order Correlation coefficients were calculated
between the mean ankle plantarflexion and the pelvic rotation as
well as the mean hip rotation in stance and mid-stance, respectively,
for hemiplegics and for diplegics. Differences between the two
groups that could also influence the correlation between the equinus
deformity and the internal rotation gait pattern were identified by
multiple unpaired t-tests. The level of significance was set at
p = 0.05 for all statistical tests.
3. Results
3.1. Modelling
When the soleus contracted alone unilaterally, strong
plantarflexion and knee extension occurred (Fig. 1, Table 2).
The hip was continuously adducted (maximum at late
 R. Brunner et al. / Gait & Posture 28 (2008) 150–156 153

Table 2
Results from modelling
Leg position
After contraction Standing Initial contact Loading response Mid-stance Late stance Preswing Toe off
Soleus
Pelvic obliquity: up (up +) 5.91 2.3 3.71 5.73 9.7 7.24 4.48
Pelvic anterior tilt (ant. +) 3.57 3.5 4.03 2.23 0.842 5.06 3.92
Pelvic retraction (ext. rot. +) 15.2 12.9 14.5 19.4 11.2 6.76 10.4
Hip adduction (add. +) 6.75 1.91 3.83 6.93 11.8 8.04 5.58
Hip extension (ext. +) 12.73 8.7 10.13 11.33 7.092 9.26 13.92
Hip internal rotation (int. +) 8.04 9.23 20.22 15.7 3.97 5.37 13.89
Knee extension (ext. +) 0 4.94 4.96 0 2.02 10 17.87
Ankle plantarflexion (plantar +) 35.7 30.9 33.79 37.8 39.46 40.2 36.8
Gastrocnemius
Pelvic obliquity: up (up +) 1.36 0.249 0.398 1.01 1.54 0.78 0.969
Pelvic anterior tilt (ant. +) 3.68 3.24 3.32 3.3 3.07 2.57 1.82
Pelvic retraction (ext. rot. +) 14 11.8 11.9 13.1 10.7 10.9 8.04
Hip adduction (add. +) 2.86 3.119 3.438 5.51 5.54 3.08 3.069
Hip extension (ext. +) 26.18 19.24 21.62 24.1 23.97 25.07 19.4
Hip internal rotation (int. +) 18.41 29.7 27.7 33.5 28.4 18.66 14.35
Knee extension (ext. +) 32.3 21.66 26.54 35.8 37.18 36 27
Ankle plantarflexion (plantar +) 9.46 16.1 11.79 8.45 7.53 9.95 13.84
Soleus and gastroc.
Pelvic obliquity: up (up +) 6 2.03 3.05 6.94 9.4 5.96 4.17
Pelvic anterior tilt (ant. +) 3.6 3.93 4.43 2.57 1.89 1.38 0.515
Pelvic retraction (ext. rot. +) 15.7 15.5 17.2 18 15.7 11.9 5.58
Hip adduction (add. +) 6.88 1.63 3.98 8.04 11.7 9.66 7.57
Hip extension (ext. +) 13.17 14.23 19.93 11.36 18.89 29.28 20.875
Hip internal rotation (int. +) 7.34 20.33 27.3 14.49 22.33 18.83 13.21
Knee extension (ext. +) 0.127 3.6 13.04 0 17.58 37.2 26.6
Ankle plantarflexion (plantar +) 36.6 33.6 33.19 36.4 38.76 39.7 36.8

stance) and internally rotated (maximum at loading

response) and flexed (maximum at loading response), but
externally rotated and extended at Preswing and toe off. At
the pelvis it produced an elevation in all phases of gait. For
initial contact, loading response, mid-stance and late stance
there was pelvic retraction and anterior tilt while the
opposite was found at Preswing and toe off.
Gastrocnemius contraction resulted only in plantarflexion
(less than with soleus), knee and hip flexion, internal rotation
and adduction with a maximum for flexion and rotation
slightly later than for soleus (Fig. 1, Table 2). The pelvis was
anteriorly tilted and elevated. The same was true for
combined contraction of soleus and gastrocnemius (Fig. 1,
Table 2).
Knee flexion before simulated contraction partly
removed the proximal effects of the soleus and gastro-
cnemius contraction (Fig. 2).

3.2. Clinical study

Increased ankle plantarflexion over the whole stance

phase and during mid-stance correlated with increased
pelvic retraction in hemiplegics (r = 0.45, p = 0.001). The
correlation between the mean ankle plantarflexion and
internal hip rotation during stance (and mid-stance) was
significant (r = 0.35, p = 0.009) for hemiplegics. In
Fig. 2. Model in crouch position. First column: initial condition; other
columns: configurations attained at the end of activation of soleus, gastro-
cnemius, and both muscles simultaneously. The same position is shown in
three different views for three-dimensional presentation.
154 R. Brunner et al. / Gait & Posture 28 (2008) 150–156

Fig. 3. Sagittal and transverse kinematics comparing both groups; mean values of all involved legs of the hemiplegic (solid black line), the diplegic patients
(dashed black line) as well as the age matched-norm children (solid grey line) in respect to the gait cycle (x-axis) are displayed. Significant deviations ( p < 0.05)
between the hemiplegic and diplegic children are shown by the hatched bar below the curves. Positive values on the y-axis represent hip flexion (a), knee flexion
(b), dorsiflexion (c) and internal rotation (d + e). The vertical lines represent the transition from stance to swing phase.

contrast, mean plantarflexion in stance showed neither
significant correlation with internal hip rotation nor with
pelvic retraction in diplegics, as well as for the evaluation
during mid-stance.
Hemiplegics showed significantly less anterior pelvic tilt
compared to diplegics during the first 25% of the gait cycle
( p < 0.050). Increased hip flexion during stance was only
found in diplegics. Knee flexion at initial contact
( p < 0.001) and during loading response ( p = 0.013) was
significantly higher in diplegics. In comparison to hemi-
plegics, who reached normal knee extension during mid-
stance and terminal stance, there was insufficient knee
extension in diplegics. There were no significant differences
in plantar flexion during the whole stance phase ( p = 0.443)
between the two groups. The mean plantar flexion at initial
contact in hemiplegics was 48 higher than in diplegics, but
this difference was not significant ( p = 0.186). Hence,
influences on the pelvic and hip rotation effect are not to be
expected. Differences in kinematics between the two groups
are displayed in Fig. 3. Whereas no significant differences
could be found for hip rotation ( p = 0.843), the hemiplegic
group showed significantly more pelvic retraction
( p = 0.002). An increased internal foot progression angle
was found in diplegics ( p = 0.001), whereas hemiplegics
showed an almost normal foot progression angle. The ankle
plantar flexor moment and power absorption ( p = 0.040)/
generation ( p < 0.001) was significantly greater in hemi-
plegics.
4. Discussion
In CP hip internal rotation is often combined with pelvic
retraction, and hip and pelvic motions are interpreted as
 R. Brunner et al. / Gait & Posture 28 (2008) 150–156
compensating for each other. Excessive plantarflexion is a
frequent concomitant deviation. These deviations at hip and
pelvis are seen, as expected, in Type 4 hemiplegic patients
but also in Type 2 ones with no neurological involvement at
hip level [17]. The combination of plantarflexion, hip
internal rotation and pelvic retraction can also be found in
other, non-neurological conditions. Correcting the increased
femoral anteversion, adductor and tensor fasciae lata
spasticity in CP surgically is not always effective in
controlling hip internal rotation during gait. Discrepancies
between the intraoperative amount of femoral derotation and
the functional outcome have been reported [3,6]. Hence
additional causes for hip internal rotation during gait need to
be considered.
The causes of functional hip internal rotation in CP may
be difficult to determine as gait patterns are subject to
compensations and secondary effects. A biomechanical
model may help: in the present study the model simulated
soleus and gastrocnemius function and included floor
resistance and dynamic coupling between anatomical
segments.
Our model allowed testing the effect of the triceps surae
components under various conditions of load and positions
of flexion at the ankle, knee, hip and spine. In our simulation
results, each component of the triceps surae muscle flexed,
internally rotated and adducted the hip and displaced the
pelvis backwards and upwards. This was particularly
noticeable in early stance when triceps surae overactivity
is typical in CP [2,11,18]. In contrast, a contraction of soleus
alone in late stance had the opposite effect on the body
segments. Hence hip and pelvic motions, especially the
rotational component, were a result of triceps surae
contraction under load. This model suggests that the pattern
of internal hip rotation and pelvic retraction does not
necessarily occur as a compensation, which has been often
assumed in CP. However, at knee level, the soleus extended
the joint whilst gastrocnemius flexed it. This is in keeping
with the concept of soleus controlling stance stability. This
model suggests that effects at the hip and knee are related to
the function of the triceps surae and are not directly
dependent on neuromuscular control.
The hemiplegic patients analysed in this study typically
showed a combination of triceps surae functional short-
ening, hip internal rotation and flexion and pelvic retraction.
There was a moderate correlation for plantarflexion and hip
internal rotation as well as pelvic retraction supporting the
concept of a distant triceps surae effect in the model. The
distant effect of the soleus on the hip and pelvis may also
explain failures after corrective osteotomies for hip internal
rotation without adductor and hamstrings involvement
[19,20]. Thus functional shortening of the triceps mechan-
ism may be another important factor for hip internal rotation
and pelvic retraction in hemiplegia.
In contrast, no correlation between triceps surae activity
and hip or pelvic motion was found in the diplegic patients,
although the amount of plantar flexion did not differ from
155
that in hemiplegics. There may be several reasons for this.
First, there may be other factors still unknown that cause hip
internal rotation, adduction and abnormal pelvic motion in
diplegics. Secondly, there may be other biomechanical
factors that partially or completely prevent the effect of the
triceps surae on the hip and pelvis. In this study, one possible
explanation is the increased flexion of knee and hip in the
diplegics, compared to the hemiplegics. This hypothesis is
supported by the results of the model, where flexion erased
the effect of triceps surae overactivity. Thirdly, bilateral
involvement may remove part of these effects as the opposite
effect occurs on the contralateral side. Fourthly, diplegics
showed significant foot malalignment with reduced moment
and power production/absorption at the ankle. This may be
due to the triceps acting out of plane.
In conclusion, our model showed that excessive ankle
plantarflexion was responsible for flexion, internal rotation
and adduction at the hip, as well as pelvic retraction and
upward motion of the pelvis. The model suggests that these
effects were not compensatory movements. These effects
contributed to the hip and pelvic deviations in the
hemiplegic patients, whereas other factors, such as knee
and hip flexion and foot malalignment may have prevented
this effect in diplegics. The model also showed that soleus
contraction produced knee extension whereas gastrocne-
mius contraction resulted in knee flexion. This may be of
relevance clinically. Correction of triceps surae function
requires careful consideration when internally rotated gait is
corrected in hemiplegia.
Acknowledgements
The authors like to acknowledge the Foundation for
Movement Disorders (Stiftung fuer Bewegungsstoerungen)
and the research project HINT@Lecco for their financial
support of this study.
Conflict of interest: None.
References
[1] Aiona MD, Sussman MD. Treatment of spastic diplegia in patients
with cerebral palsy: Part II. Pediatr Orthop B 2004;13:13–38.
[2] Gage J. The treatment of gait problems in cerebral palsy. Section 3:
gait pathology in cerebral palsy. London: Mac Keith; 2004, ISBN 1
898683 37 9.
[3] Saraph V, Zwick EB, Zwick G, Dreier M, Steinwender G, Linhart W.
Effect of derotation osteotomy of the femur on hip and pelvis rotations in
hemiplegic and diplegic children. J Pediatr Orthop B 2002;11:159–66.
[4] Ounpuu S, DeLuca P, Davis R, Romness M. Long-term effects of
femoral derotation osteotomies: an evaluation using three-dimensional
gait analysis. J Pediatr Orthop 2002;22:139–45.
[5] Pirpiris M, Trivett A, Baker R, Rodda J, Nattrass GR, Graham HK.
Femoral derotation osteotomy in spastic diplegia. Proximal or distal? J
Bone Joint Surg Br 2003;85:265–72.
[6] Dreher T, Wolf S, Braatz F, Patikas D, Döderlein L. Internal rotation
gait in spastic diplegia—critical considerations for the femoral dero-
tation osteotomy. Gait Posture 2007;26:25–31.
156 R. Brunner et al. / Gait & Posture 28 (2008) 150–156
[7] Kim H, Aiona M, Sussman M. Recurrence after femoral derotational
osteotomy in cerebral palsy. J Pediatr Orthop 2005;25:739–43.
[8] Brunner R, Meier G. Retrograde movement of the knee-joint in
hemiparetic gait: an explanation for the development of typical gait
pattern? Gait Posture 1998;8:74.
[9] Delp SL, Hess WE, Hungerford DS, Jones LC. Variation of rotation
moment arms with hip flexion. J Biomech 1999;32:493–501.
[10] Arnold AS, Komattu AV, Delp SL. Internal rotation gait: a compen-
satory mechanism to restore abduction capacity decreased by bone
deformity. Dev Med Child Neurol 1997;39:40–4.
[11] Perry J. Gait analysis. Thorofare: Slack; 1992.
[12] Chambers HG, Rose J. Dynamic electromyography. In: Gage J, editor.
The treatment of gait problems in cerebral palsy. London: Mac Keith;
2004. ISBN 1 898683 37 9.
[13] Clauser CE, McConville JT, Young JW. Weight, volume, and centre of
mass of segments of the human body, AMRL Technical Report (TR-
69-70). Wright-Patterson Air Force Base, OH, 1969.
[14] Perry J. Gait analysis. Chapter 2: phases of gait. Thorofare: Slack;
1992.
[15] Kabada MP, Ramakrishnan HK, Wootten ME. Measurement of lower
extremity kinematics during level walking. J Orthop Res 1990;8:
383–92.
[16] Hermens HJ, Freriks B, Merletti R, Stegeman D, Blok J, Rau G et al.,
European Recommendations for Surface Electromyography
(SENIAM). Enschede Roessingh Research and Development 1999,
CD-rom.
[17] Winters Jr TF, Gage JR, Hicks R. Gait patterns in spastic hemiplegia in
children and young adults. J Bone Joint Surg 1987;69-A:437–41.
[18] Brunner R, Romkes J. Abnormal EMG muscle activity during gait in
patients without neurological disorders. Gait Posture 2007 Nov 13
[Epub ahead of print].
[19] Arnold AS, Delp SL. Rotational moment arms of the medial ham-
strings and adductors vary with femoral geometry and limb position:
implications for the treatment of internally rotated gait. J Biomech
2001;34:437–47.
[20] Arnold AS, Asakawa DJ, Delp SL. Do the hamstrings and adductors
contribute to excessive internal rotation of the hip in persons with
cerebral palsy? Gait Posture 2000;11:181–90.