Clinical Biomechanics 21 (2006) 849859

www.elsevier.com/locate/clinbiomech

Muscular compensation and lesion of the anterior cruciate

ligament: Contribution of the soleus muscle during
recovery from a forward fall
P. Colne a, P. Thoumie

a,b,c,*

a
INSERM UMR S 731, 33 boulevard de Picpus, 75012 Paris, France
Universite Pierre et Marie Curie-Paris 6, 33 boulevard de Picpus, 75012 Paris, France
Service de Reeducation Neuro-orthopedique, Hopital Rothschild APHP, 33 boulevard de Picpus, 75012 Paris, France
b

Received 9 November 2005; accepted 4 April 2006

Abstract
Background. Knee stability following an anterior cruciate ligament lesion has been widely studied. Only recent studies focused on the
contribution of the soleus muscle. Our purpose was to characterize the dynamic and muscular activity of balance recovery in healthy
subjects and patients with an anterior cruciate ligament rupture. The role of the soleus was investigated in the ipsilateral compensation
developed to stabilize the knee and in the contralateral compensation to recover balance.
Methods. Twelve anterior cruciate ligament decient patients, ten anterior cruciate ligament repaired patients and 14 control subjects
were recorded during a forward fall involving stepping to recover balance.
Findings. The dynamic of the centre of gravity remained normal when compared to the control group regardless of the treatment,
suggesting an adapted compensation to knee instability in this situation. A bilateral increase in soleus activity was related to an increased
duration in the balance recovery process in all patients. Patients used one of two strategies to recover balance regardless of the treatment:
reducing the step length, involving an early recruitment of the soleus before heel contact, or anticipating braking with a similar step
length requiring a predominant activity of the hamstrings.
Interpretations. These results suggest that bilateral activity of the soleus is involved to compensate for instability and highlight the
contribution of the soleus to rehabilitation after an anterior cruciate ligament lesion, not only as a compensatory muscle acting at
the knee level but also at a higher level in the bilateral control of stance.
 2006 Elsevier Ltd. All rights reserved.
Keywords: Anterior cruciate ligament; Neuromuscular control; Soleus muscle; Balance recovery

1. Introduction
The control of postural balance depends on various systems (nervous system, locomotor apparatus). A lesion in
any of these can lead to a perturbation of postural reactions. In clinical practice, pathology may associate several
lesions that perturb the control of equilibrium. This is the

Corresponding author.
E-mail address: philippe.thoumie@rth.aphp.fr (P. Thoumie).

0268-0033/$ - see front matter  2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.clinbiomech.2006.04.002

case following a lesion of the anterior cruciate ligament

(ACL) when there persists a risk of knee instability (Noyes
et al., 1983), linked not only to the suppression of the
mechanical properties of the ligament but also to the loss
of the proprioceptive control of the joint exerted by the
mechanoreceptors of the ACL (Adachi et al., 2002; Johansson et al., 1991).
After an ACL injury, some people (called copers) are
able to return to their daily and sporting activities, stabilizing their knee without requiring surgical repair, whereas
others (called non-copers) cannot (Noyes et al., 1983;
Rudolph et al., 2001). For these patients, surgical repair

850

P. Colne, P. Thoumie / Clinical Biomechanics 21 (2006) 849859

of the ligament restores the mechanical stability of the

knee. However, whether the proprioceptive aerences after
an ACL reconstruction can be regenerated remains controversial (Bonm et al., 2003; Iwasa et al., 2000).
In all cases, an understanding of the muscular synergies
required to regain a stable knee after an ACL lesion is crucial in devising appropriate rehabilitation. Many studies
(Berchuck et al., 1990; Alkjaer et al., 2003; Beard et al.,
1996; Boerboom et al., 2001; Kvist, 2004; Rudolph et al.,
2001; Torry et al., 2004) have sought to determine which
compensations are applied with an ACL lesion (repaired
or not) especially during the stance phase of gait. These
studies have shown various neuromuscular, kinematic
and kinetic adaptations, but have not identied a common
knee stabilisation strategy in subjects who compensate well
and recover normal knee function. The study of muscular
synergies stabilizing the knee after an ACL lesion has
focussed primarily on the hamstrings, the quadriceps and
the gastrocnemius muscles (Solomonow et al., 1987; Baratta et al., 1988; Lass et al., 1991), which are directly
responsible for mediating the motor function and the active
stability of the knee. Few studies have investigated soleus
muscle activity after an ACL lesion (Ciccotti et al., 1994;
Rudolph and Snyder-Mackler, 2004; Rudolph et al., 2001).
Two recent studies (Sherbondy et al., 2003; Elias et al.,
2003), however, suggest that one function of the soleus in
a closed kinetic chain could be to help protect the ACL
by pulling back the proximal part of the tibia and preventing its anterior displacement. However, the experimental
protocol of these studies did not correspond to a real situation of loading on the knee. Because the soleus is involved
in controlling upright posture, the kinesiological study of
the soleus must be paired with a global evaluation of balance; several authors have highlighted a bilateral perturbation of the unilateral stance after a ligament lesion
(Lysholm et al., 1998).
Since compensations of an ACL lesion are found when
the knee is unstable (especially in a closed kinetic chain
when the stresses are higher), the study of situations other
than normal gait have been proposed (Alkjaer et al., 2002;
Rudolph and Snyder-Mackler, 2004) to analyze knee instability after an ACL lesion.
And so, the study of balance recovery could constitute
an interesting paradigm to characterize the adaptation of
the motor command when faced with a situation of instability. Preventing a fall when starting from an inclined position, an experimental paradigm initially described by Do
et al. (1982), implies the use of an adapted strategy which
aims to prevent the fall of the centre of gravity by taking
several steps. Several authors have described the contribution of the soleus in this process and their relationship with
the dynamics of the centre of gravity (Do et al., 1982, 1999;
Thoumie and Do, 1996).
The aim of this work was to study the dynamics of balance recovery and the muscular activities after a forward
fall in patients presenting an ACL lesion, and in control
subjects. Our hypotheses were that changes in the activa-

tion of the soleus could be associated with a dierent balance recovery strategy after an ACL lesion (repaired or
not), and that surgical restoration of the stability of the
knee should be associated with a modication of the strategy used by the ACLD subjects. Moreover, a bilateral
study of the soleus would also help distinguish between
the compensation directly linked to the lesion in ipsilateral
muscular activity and that of contralateral activity which
aims to control general balance.
2. Methods
2.1. Patients
Three groups of subjects took part in the study: a control group comprising 14 healthy subjects with no pathology of the knee, a group of 12 ACL decient knee patients
(ACLD) having an unilateral lesion of the ACL conrmed
clinically (Lachman test) and by magnetic resonance imaging, a group of 10 ACL knee surgically reconstructed
patients (ACLR) with an unilateral lesion of the ACL conrmed during surgery. Knee joint pain, eusion and limitation of the full knee joint range of motion were exclusion
criteria of the study.
Two of the patients were regular participant in high level
sport activities and surgically treated 2.5 months after the
lesion. All others were recreational sport participants, only
participating occasionally in sport activities like skiing and
jogging during holidays. They were surgically treated only
if they experienced instability of the knee during daily life
since their lesion. All reconstructed patients were treated
by the technique of Kenneth Jones except one reconstructed by the technique of Mac Intosh.
Most of the patients having a relatively poor level of
functional activities they were assessed before the test by
a questionnaire relevant to their capacities for walking,
going up and down stairs and jogging. At the time of the
test all patients armed to feel no discomfort in daily life
activities. Laxity of the knee joint was assessed clinically
and qualitatively by the same evaluator manually performing the Lachman test. All ACLD patients showed signicant drawer signs by comparison to ACLR subjects and
control group. The force of the quadriceps assessed by
the maximal voluntary isometric contraction manually
resisted (break test) and by squatting ability was considered
as normal.
The balance recovery test was performed at least three
months after the initial lesion (0.253 years) or surgical
treatment (0.453.5 years). A comparison of the time
elapsed between the test and the lesion or the surgery
(respectively 22(21) months in ACLD and 11(12) months
in ACLR) did not show statistical dierence with regard
to the treatment (Wilcoxon test, P = 0.34). All patients
underwent 6 weeks of rehabilitation programs from various physiotherapists in private practices or rehabilitation
centres. The three groups of subjects were homogeneous
in terms of height [1.73(0.1) m in CTL, 1.7(0.09) m in

P. Colne, P. Thoumie / Clinical Biomechanics 21 (2006) 849859

ACLD and 1.78(0.09) m in ACLR], age [29(11) years in

CTL, 38(10) years in ACLD and 27(8) years in ACLR],
and weight [73(10) kg in CTL, 74(11) kg in ACLD and
76(13) kg in ACLR]. All subjects rst gave their informed
consent.
2.2. Material
We used the experimental paradigm described by Do
et al. (1982). In this paradigm (Fig. 1), the subject stands
on a force plate in a forward inclined posture (15). The
body is straight, the arms hang alongside the body and
the eyes look straight ahead. The subject is held by a
restraining device composed of an abdominal belt and a
horizontal steel cable connected to an electromagnet
mounted on a dynamometer. The device is released, without the subjects knowledge, causing the subject to fall forward. The subject is instructed to take a few steps to
recover balance.
A 60 120 cm AMTI force plate (AMTI, Watertown,
MA, USA) recorded the ground reaction force. Data were
recorded with a sampling rate of 500 Hz. Recordings
started 50 ms before the release of the restraining device
and continued for 850 ms afterward. Each trial was analyzed individually. Then, the values of each parameter were
averaged for each subject, and the total mean value was
calculated for all the subjects of the same group.
The surface EMG activity of ve muscles: soleus, tibialis, hamstrings medial (semi tendinosus) and lateral (femoris biceps), rectus femoris (quadriceps) was recorded in the
moving limb then in the stance limb for the control group.
A special attention was paid to characterize the activity of
the soleus. Electrodes were placed at the medial part of the
shank near the tibial bone to avoid cross-talk with
gastrocnemius.
For the patient group, the same muscles were recorded
on the injured side. When balance was recovered by stepping with the injured side, the activity of the soleus and tibialis of the contralateral healthy stance side was also

851

recorded, since these muscles were likely to participate in

controlling the fall (Michel and Do, 2002).
Muscular activity was recorded using a bipolar detection
mode. The skin was rst lightly abraded and cleaned to
reduce impedance. Then, the surface electrodes (diameter:
1 cm) were axed to the skin over the relevant muscle bellies, spaced 2 cm centre to centre and aligned parallel to the
underlying muscle bre direction. The lead wires were
axed to the adjacent segments to prevent interference of
the EMG signal. A reference electrode was axed to the
subjects wrist. The EMG signal was amplied twice: rst
by a preamplier (gain 1000, dierential amplier of great
impedance 10 MX) contained in a case (550 g, 8 channels)
and attached to the subjects belt. The preamplier (LPMCNRS, Orsay, France) was connected to a second amplier
by a 4 m-long cable. The gain of the second amplier was
adjustable from 1000 to 50 000. The bandwidth of the
amplication chain was between 3 Hz and 2000 Hz, covering the range of the EMG signal. The analogue signals were
collected then converted into digital signals by a CED 1042
card controlled by a PC. The sampling rate was 1000 Hz.
2.3. Protocol and data processing
Each of the subjects (control and patient) performed 20
trials (ten starting with the right lower limb and ten with
the left one).
Software calculated the coordinates of the centre of foot
pressure and the dynamics of the centre of gravity from the
ground reaction force. The vertical acceleration of the subjects centre of gravity (z00 G) was calculated by dividing the
dierence between the vertical component of the ground
reaction force and the subjects weight by the subjects
mass (z00 G = (RZ-P)/m). The vertical velocity (z 0 G) and
position (zG) of the CG were calculated by successive integration. The displacement of the centre of foot pressure
was used to calculate the length of the rst step taken to
recover balance.
Each of the EMG signals recorded was processed individually. The latencies of beginning and end of the various
EMG bursts were marked for each trial, in order to calculate their mean values as well as the mean duration of each
EMG burst (dierence between the start and end latencies
for each subject). The mean value for each group was calculated using the mean values obtained for each subject.
Mechanical and EMG data obtained for each group of
subjects were compared 2 to 2 using a non-parametric
Mann and Whitney test (Statview software for PC). The
threshold of signicance dierence used was P < 0.01.
3. Results
3.1. Mechanical aspect

Fig. 1. Experimental setup for studying balance recovery after a forward

fall.

Results were analysed separately in patients and control

groups. Since the laxity of the knee was dierent in the
patient group according to treatment, data were analysed

852

P. Colne, P. Thoumie / Clinical Biomechanics 21 (2006) 849859

rst according to the treatment. However, since no statistical dierence was found in the patient group with regard to
the treatment, the results were then compared for all combined patients.
3.1.1. Control group
The control subjects show a reproductive behaviour
identical to that previously described by Do et al. (1982).
Balance recovery includes rst a two-foot stance reaction
phase from release (t0) to toe-o (TO) with a duration of
263(SD: 24) ms. It is followed by a step execution phase
from toe-o to heel contact (HC) with a swing phase duration of 211(32) ms, with the contralateral foot remaining in
stance. Then the ground acceptance phase of the swinging
foot occurs, preceding execution of the second step.
Vertical biomechanical events of balance recovery rst
show (Fig. 2) a negative value of the vertical acceleration
of the CG (z00 G1), which corresponds to the fall of the
CG following the release of the restraining device. Then,
the reaction phase occurs corresponding to the reverse of
z00 G reaching a positive value (z00 G2) before toe-o. The
latency of z00 G2 is 187(15) ms and its amplitude is
3.17(1.03) m s2. Step execution begins with toe-o, ending
with foot contact of the swinging limb.

Balance recovery ends when the rear foot leaves the

ground (executing the second step), at which time the
length of the step, whose amplitude is 0.754(0.06) m, and
the duration of balance recovery which is 592(43) ms, are
measured. The vertical velocity of the CG (z 0 G) shows a
negative peak (z 0 Gmin) corresponding to the braking of
balance recovery (brake peak, BP). This appears after heel
contact. The amplitude of the vertical velocity of the CG at
heel contact (z 0 GHC) is 0.225(0.07) m s1. The height of
the CG (zG) measured at heel contact is slightly lower than
its initial value before the release of the restraining device
the dierence being 0.008(0.021) m.
3.1.2. Comparison of data in ACLD and ACLR subjects
The values of the patients data (ACLD and ACLR) are
displayed in Table 1. A comparison of both groups of
patients (ACLD, ACLR) does not show any dierence
linked to the treatment. This is true for both the temporal-spatial parameters of balance recovery and the subjects
kinematic parameters. The entire balance recovery duration and the duration of its various phases are not statistically dierent in either group of patients. Nor is there any
statistical dierence in the dynamic of the CG or the length
of the rst step in these two groups.

Fig. 2. Recordings of biomechanical parameters and EMG activities from the right and left soleus during a balance recovery step (right limb moving).
Single trial for a control subject: z00 G, z 0 G, zG are respectively the vertical acceleration, velocity and position of the centre of the gravity (U for upward)
z 0 Gmin is the negative peak of the vertical velocity and corresponds to the braking of the balance recovery (BP peak of braking speed), xP and yP the
antero-posterior and lateral displacement of the foot pressure (f for forward and s for stance, L for the length of the step), SOLm and SOLs, raw EMG
activity of the soleus from the moving and the stance limb.

P. Colne, P. Thoumie / Clinical Biomechanics 21 (2006) 849859

853

Table 1
Mean values of the temporo-spatial components of the balance recovery step after a forward fall and the biomechanical parameters of the dynamic of the
centre of gravity per group with respect to the treatment (ACLD, ACLR) and respective P values
Parameters

ACLD

Swinging limb

Uninjured side

Injured side

Uninjured side

ACLR
Injured side

P value

Balance recovery duration (ms)

Toe-o latency (ms)
Swing phase duration (ms)
Z00 G1 latency (ms)
Z00 G1 amplitude (m s2)
Z00 G2 latency (ms)
Z00 G2 amplitude (m s2)
Z 0 G amplitude at HC (m s1)
DZG at HC (m)
Step length (m)

665(50)
291(45)
234(24)
55(9)
0.99(0.33)
208(18)
3.27(0.88)
0.246(0.08)
0.004(0.018)
0.616(0.169)

657(53)
301(92)
234(25)
54(10)
0.98(0.25)
199(28)
3.46(1.05)
0.253(0.077)
0.003(0.014)
0.596(0.169)

654(64)
285(39)
245(24)
49(8)
0.93(0.29)
201(38)
2.60(0.75)
0.185(0.069)
0.001(0.016)
0.613(0.180)

672(81)
296(45)
245(31)
53(12)
0.92(0.30)
204(39)
2.62(1.09)
0.175(0.086)
0.000(0.023)
0.596(0.166)

NS
NS
NS
NS
NS
NS
NS
NS
NS
NS

Z00 G1: negative peak of the vertical acceleration of the centre of gravity (CG) before toe-o of the swing limb.
Z00 G2: positive peak of the vertical acceleration of the reaction phase (before toe-o).
Z 0 G at HC: vertical velocity of the CG at heel contact (end of the swing phase).
DZG at HC, variation of the height of the CG at heel contact.

3.1.3. Comparison of data in patient and control groups

regardless of the treatment
Data for the control group and the patient group are displayed in Table 2. A comparison of the dynamics of the CG
shows that both the latency of z00 G2, comprised between
208(30) ms and 195(35) ms in both subgroups of patients,
and its amplitude, comprised between 3.20(1.13) m s2
and 2.98(1.16) m s2 in both subgroups of patients, are similar in the patient and control groups [latency 187(1.03) ms,
amplitude 3.17(1.03) m s2]. The vertical velocity of the
centre of gravity at TO and HC as well as the amplitude
of the peak of braking (z 0 Gmin) are identical in the patient
and control groups. In the same way, the height of the CG at
heel contact of the swinging foot is not statistically dierent
from that of the control [mean values of the fall of the CG
comprised between 0.004(0.020) m and 0.001(0.016) m in
the patient group versus 0.008 m (0.021) in the control].
For all patients, the braking time with respect to heel con-

tact (db = tz 0 Gmin-tHC) shows two dierent behaviours

when the balance recovery step is performed with the
injured side (Fig. 3). Some subjects brake before heel contact [db < 0: mean value 16(18) ms] while others brake
after heel contact [db > 0: mean value 20(17) ms versus
26(20) ms in control]. Ten subjects (6 ACLD and 4 ACLR)
brake before (B, braking) and 12 subjects (6 ACLD and 6
ACLR) brake after (NB, non-braking). The distribution
of subjects in both groups shows no link with either the
treatment (rehabilitation alone or surgery) or with the time
elapsed since the injury or surgery [respectively 19(16) and
13(11) months in the B and NB groups, Wilcoxon test
P = 0.71, NS].
The length of the step performed by the patients (Fig. 3)
follows the same two-group distribution regardless of the
swing limb. The length of the step is not statistically dierent
from that of the control group for the group which brakes
before [uninjured side 0.686(0.135) m and injured side

Table 2
Mean values of the temporo-spatial components of the balance recovery step and the biomechanical parameters of the dynamic of the centre of gravity
after a forward fall, per group, with respect to the time of braking
Parameters

Control

Braking patients

Non-braking patients

Swinging limb

Normal

Uninjured side

Injured side

Uninjured side

Injured side

Balance recovery duration (ms)

Toe-o latency (ms)
Heel-o latency (ms)
Swing phase duration (ms)
db (ms)
Z00 G1 latency (ms)
Z00 G1 amplitude (m s2)
Z00 G2 latency (ms)
Z00 G2 amplitude (m s2)
Z 0 G amplitude at HC (m s1)
DZG at HC (m)
Step length (m)
Vs (m s1)

592(43)
263(24)
474(46)
211(32)
26(20)
57(9)
1.21(0.27)
187(15)
3.17(1.03)
0.225(0.07)
0.008(0.021)
0.754(0.068)
3.6(0.5)

667(41)**
290(43)
536(40)*
247(20)*
9(10)
51(8)
0.94(0.35)
209(32)
3.08(0.89)
0.241(0.057)
0(0.018)
0.686(0.135)
2.8(0.7)**

668(44)**
294(41)
536(57)*
242(26)*
16(17)**,
52(10)
0.87(0.29)
208(30)
3.20(1.13)
0.212(0.09)
0.001(0.016)
0.664(0.147)
2.8(0.6)**

654(67)**
289(34)
522(44)*
233(26)*
11(17)
54(9)
0.99(0.28)
202(34)
2.87(0.89)
0.199(0.093)
0.004(0.018)
0.555(0.18)**,
2.4(0.7)**

660(82)**
282(47)
519(58)*
236(30)*
20(17)
54(11)
1.02(0.24)
195(35)
2.98(1.16)
0.221(0.091)
0.004(0.02)
0.539(0.16)**,
2.3(0.6)**

db is the braking time (=tz 0 Gmin-tHC). Vs is the velocity of the step (=ratio of the length of the step to the swing phase duration).
P values: *P < 0.05, **P < 0.01, signicantly dierent from control; P < 0.05, P < 0.01, signicantly between subgroups.

854

P. Colne, P. Thoumie / Clinical Biomechanics 21 (2006) 849859

Fig. 3. Recordings of some of the biomechanical parameters of the balance recovery step and some of the EMG activities for a control subject (CTL), an
ACL subject who brakes before heel contact (B) and an ACL subject who brakes after heel contact (NB). Single trial for each subject. The swinging limb
was the left leg for the CTL and the injured leg for the ACL. Vertical solid line shows the time of release (t0) and the dotted lines correspond to the time of
toe o (TO, start of the step execution phase) and heel contact (HC, end of the swing phase). z00 G and z 0 G are the vertical acceleration and velocity (U,
upward), L on xP graph is the length of the rst step executed to recover balance (F, forward and L, step length, BP peak of braking speed). SOLm and
HAMm are respectively the soleus and hamstrings muscle activities of the moving limb.

0.664(0.147) m versus 0.754(0.068) m in control] while it is

signicantly shorter in the group which brakes after
[uninjured side 0.555(0.178) m; injured side: 0.539(0.161) m
versus 0.754(0.068) m in control, P < 0.01].
The balance recovery duration measured at the end of
the rst step is always greater in the patient group with mean
values comprised between 660(67) ms and 668(44) ms versus
592(43) ms in control. The dierence is signicant (P < 0.01)
regardless of the moving limb or time of braking (Table 2).
If we consider the duration of the various phases of balance
recovery, the swing phase, comprised between TO and HC,
is the only one whose duration is signicantly increased
[respectively 294(41) ms and 282(47) ms in the B and NB
groups versus 263(24) ms in control, the level of signicance
is lower P < 0.05].
The mean velocity of step execution (Vs, ratio of the
length of the step to the swing phase duration) appears
to be slower in the entire ACL group with mean values
comprised between 2.3(0.6) m s1 and 2.8(0.6) m s1 versus
3.6(0.5) m s1 in control. The dierence is signicant for
the B and NB groups (P < 0.01).
3.2. EMG aspect
Results were analysed separately in patients and control
groups. Although the mechanical pattern of balance recov-

ery does not show any dierence between the ACLR and
ACLD subjects, EMG data were rst analysed with respect
to the treatment, then for the entire group.
3.2.1. Control group
On the moving limb, each muscle shows two bursts of
EMG (Figs. 3 and 4). The soleus is the rst muscle to
become active. Its activity starts 65(11) ms after the release
of the restraining device and ceases 189(31) ms after [duration 124(31) ms]. This burst occurs before the positive peak
of vertical acceleration of the CG and ends a short while
afterwards. This activity corresponds to the reaction to
the fall occurring just after release (Thoumie and Do,
1996). The second burst of the soleus appears after TO
[433(63) ms after the release] and occurs 63(43) ms before
heel contact, preparing to stabilize the swinging limb during
the ground acceptance phase, with the knee and ankle in
exion at the time of heel contact. The activity of the soleus
continues during the unilateral stance which follows.
The rst burst of the tibialis occurs 195(35) ms after
release and ends 333(42) ms afterwards with a 138(39) ms
mean duration, starting before TO and ending shortly
afterwards. It allows lifting the foot from the ground and
swinging. The second burst of the tibialis starts
435(47) ms after the release and occurs 64(29) ms before
HC. It prepares to stabilize the ankle at heel contact.

P. Colne, P. Thoumie / Clinical Biomechanics 21 (2006) 849859

ms

Solm ipsi

Tam ipsi

Sols contra

ms

800

855

Tas contra

800

***

700

***

***

700

NS
600

600

HC 500

500

400

400

300

300

***
200

200

100

100
0

0
CTL

NB

CTL

ms

CTL

NB

NB

CTL

NB

-100

-100

Quam ipsi
800

HamMm ipsi

HamLm ipsi

***

***

700

**

NS

600

HC 500
400
300
200
100
0
CTL

NB

CTL

NB

CTL

NB

-100

Fig. 4. Mean values and standard deviations of EMG activities during balance recovery after a forward fall in control and patients regardless of the
treatment. EMG burst of the moving limb (injured side for the patient). From bottom to top: rst EMG burst duration, onset of the second EMG burst
and time of heel contact (HC). From left to right on the diagrams: control (CTL), braking (B) and non-braking (NB) groups. Right side of the gure
corresponds to the mean values of the moving limb, left side corresponds to that of the stance limb. HC is the time of heel contact, Solm, Tam, Quam,
HamM and HamL are respectively the Soleus, Tibialis, Quadriceps, Hamstrings medial and lateral muscle of the moving limb, IPSI correspond to the limb
ipsilateral to the lesion. Sols and Tas are the Soleus and Tibialis muscle of the stance limb, CONTRA corresponds to the limb contralateral to the lesion.
***P < 0.001, **P < 0.01.

The rst burst of the hamstrings medial (HamM) and lateral (HamL), respectively, lasts 114(51) ms and 127(40) ms.
It occurs respectively 70(12) ms and 78(25) ms after the
release and ends afterward [respectively 183(55) ms and
206(58) ms]. These bursts occur before the positive peak
of the vertical acceleration and end shortly afterwards. They
act again as a posterior muscular activity reacting to the forward fall. The second burst of these muscles occurs
378(63) ms and 411(69) ms, respectively, after the release
during the swing phase. These bursts start before HC
[116(39) ms for HamM and 89(41) ms for HamL] to control
the extension of the knee during the swing phase, and then
to stabilize the lower limb during the ground acceptance
phase while the hip and knee are in exion at the time of
heel contact. These muscular activities occur during the unilateral stance which follows.
Quadriceps shows a rst burst occurring 178(47) ms
after release and ending 336(37) ms afterward [duration
159(51) ms]. This burst occurs around TO, starting a little
before and ending afterward. This burst makes it possible

to extend the knee during the swing phase. The second

burst of this muscle occurs 449(37) ms after the release
and is located before HC 50(26) ms. It contributes to stabilize the knee at the time of heel contact then during the unilateral stance.
On the stance limb, all muscles recorded show only one
burst of EMG. Activity of the soleus occurs 73(17) ms after
the release and ends 486(55) ms afterward [duration
413(55) ms]. This activity is located before the positive
peak of the vertical acceleration and ceases at about the
time of heel contact of the swinging limb. This activity
comprises an initial part corresponding to the bilateral
stance reaction to the fall which is followed by a later part
in unilateral stance to stabilize the stance limb and to control the forward body progression (Michel and Do, 2002).
The tibialis shows varying degrees of activity depending
on the subject. It starts 97(10) ms after the release and ends
470(51) ms after [duration 373(51) ms]. The bursts of
the hamstring medial and lateral, respectively, last
202(67) ms and 203(83) ms. They become active 70(10) ms

856

P. Colne, P. Thoumie / Clinical Biomechanics 21 (2006) 849859

and 75(13) ms after release and end shortly after TO,

respectively 273(67) ms and 278(79) ms after release. This
activity is shorter than that of soleus and occurs primarily
in reaction to the fall. The quadriceps is active 151(56) ms
after the release, a little later than the hamstrings, and ends
443(58) ms afterwards. It serves primarily to support the
unilateral stance.
3.2.2. Comparison of data in ACLD and ACLR subjects
When balance recovery is performed with the injured
limb moving, the respective duration of the rst EMG
burst is not dierent in the two groups of patients. Some
dierences are observed in certain parameters of latency,
but they do not aect the duration of the EMG activities
and do not characterize any particular behaviour between
the ACLD and ACLR groups.
Only the second burst of the soleus and hamstrings medial occurs earlier in the ACLD group [respectively
408(55) ms in the ACLD group and 431(56) ms in the
ACLR for soleus and 360(59) ms in ACLD and
382(52) ms in ACLR for hamstrings]. The dierence is signicant (P < 0.02 for the soleus and P < 0.01 for the
HamM) but it does not aect the fact that the time between
the beginning of the second burst of EMG and the time of
HC [respectively 94(33) ms for the soleus and 106(62) ms
for the hamstrings medial in the ACLD group and
67(28) ms for soleus and 77(45) ms for hamstring medial
in the ACLR group] increases statistically for these two
muscles as for the others in both groups of patients in comparison with the control.
In the uninjured contralateral stance limb, the duration
of soleus activity [448(13) ms in ACLD and 480(81) ms in
ACLR] does not dier between the two groups of patients,
but is signicantly increased in these groups compared to
the control. Although the duration of the tibialis is longer
in the ACLR group, it is longer in both groups of patients
with respect to the control.
In conclusion, respective latencies of the second burst of
the soleus and hamstrings medial are the only parameters
which dierentiate muscular activity in ACLD and ACLR
patients when the injured limb is the moving one.
When balance recovery is performed with the injured
stance limb, the duration of EMG activities does not dier
between the two groups of patients except for the hamstrings lateral in the ACLD group, where the EMG duration is longer because of later termination. The increase
in duration is signicant in all muscles recorded in both
groups of patients in comparison with the control, except
for the quadriceps.
3.2.3. Comparison of data in patient and control groups
regardless of the treatment
In both groups of patients, we observe early compensations with respect to the release of the restraining device, as
well as compensations preceding heel contact of the swinging limb, characterizing a unique pattern for each of the
two groups: B and NB (Fig. 4).

When balance recovery is performed with the injured

limb moving, the rst burst of the soleus of injured limb
shows the same duration in both groups of patients [respectively 125(35) ms and 144(4) ms for the B and NB groups]
and control [124(32) ms]. The second burst of the soleus is
earlier in the NB group with respect to release [respectively 399(56) ms in the NB group versus 450(43) ms in
the B group, and 433(63) ms in control, P < 0.001 for both
groups] and also earlier with respect to HC [respectively
115(48) ms in the NB group versus 82(52) ms in the B
group and 63(43) ms in control, P < 0.001 for both groups].
Note that there is the same number of ACLD and ACLR
subjects in the NB group and that the respective latency of
the second burst of the soleus in these two sub-groups does
not dier with respect to release [respectively 396(60) ms
and 401(53) ms].
The activity of the tibialis does not dier between the
three groups. The rst burst of the hamstrings (HamM
and HamL) shows the same duration in all three groups.
The time elapsed between the beginning of the second burst
of the hamstrings and HC is signicantly longer in the
patients than in the control [respectively 160(31) ms and
144(33) ms for HamM and HamL in the B group and
142(45) ms and 128(37) ms for HamM and HamL in the
B group versus 116(39) ms for HamM and 89(41) ms for
HamL in the control, P < 0.001 for all muscles]. This time
is even higher for HamM in the B group than in the NB
group (P < 0.01).
The rst burst of the quadriceps does not dier in the
three groups. The time elapsed between the second burst
of this muscle and HC is higher in both ACL groups with
no statistical dierence.
On the contralateral uninjured stance limb, the duration
of the bursts of the soleus [466(40) ms in the B group and
466(81) ms in the NB group] does not dier between the
two groups of patients. The same is true for the tibialis
[duration 435(92) ms in the B group and 424(74) ms in
the NB group]. This duration is higher (P < 0.001) than
in the control for both muscles [respective duration in
control 413(55) ms for the soleus and 373(51) ms for the
tibialis].
When balance recovery is performed with the injured
stance limb, the duration of the bursts of all recorded muscles in the injured stance limb is higher in both ACL groups
compared to the control. The dierence is signicant with
respect to the control for all muscles of both groups ACL
(P < 0.001) except for the quadriceps in the NB group.
4. Discussion
The aim of the study was to characterize the muscular
activities and dynamics of balance recovery after an ACL
lesion, and in particular, to identify the role of the soleus
based on treatment (rehabilitation alone or surgery). Our
results show that while the activity of the soleus is dierent
after an ACL lesion, it corresponds more to a common strategy in the subgroups of patients and cannot be explained

P. Colne, P. Thoumie / Clinical Biomechanics 21 (2006) 849859

solely by the passive mechanical instability of the knee

related to the lesion of the ligament corrected by surgery.
The following discussion considers the modications
observed in the mechanical pattern of the balance recovery
motor program, the related modications of EMG and the
dynamic stabilization of the knee after an ACL lesion.
4.1. Modications to the mechanical balance recovery
pattern after an ACL lesion
Our results show that patients who have an ACL injury
are able to recover balance without any discomfort that
was suggested by the absence of modication in the dynamics of the centre of gravity: the position and falling velocity
of the centre of gravity are similar at the end of the balance
recovery process in all the groups and subgroups.
The strategy shared by both groups is characterized by a
signicant increase in the balance recovery duration regardless of the treatment or the moving limb, and regardless of
the time since injury or surgery. This change is relevant to
the prevalent increase in the swing phase duration which
constitutes the essential parameter permitting the regulation
of balance recovery (Do et al., 1999). Our study describes a
new clinical situation where we observed an increase in the
swing phase duration to control the moving velocity of the
whole body and to protect the injured knee during contact
with the ground after swinging. This result should be compared to the observations made by Alkjaer et al. (2002)
who described a lengthening of the duration of a forward
lunge movement in ACLD subjects. The initial part of the
movement, corresponding to the ground acceptance phase
of the foot as the knee was exing, was increased in non-copers and copers, suggesting to the author that the lengthening
of the motor program could allow ACLD subjects to react
in case of instability of the knee.
Besides this common increase in the duration of balance
recovery in all ACL subjects, a modication of the time of
braking and of the length of the step could be seen allowing
patients to control the impact of the ground acceptance of
the foot using one of these two strategies: one group of subjects brakes before ground contact of the foot when swinging with the injured limb, without reducing the length of
the step. An other group brakes after ground contact of
the foot, like the control, bilaterally shortening the step
to recover balance. A similar shortening strategy was found
in subjects having sustained a unilateral chronic deaerentation (abolition of the Achilleous reex), whereas an
experimental blocking of this reex resulted in the execution of a normal step length at the expense of a lowering
of the centre of gravity (Thoumie and Do, 1996). These
authors suggested that the strategy used by the subjects
with chronic deaerentation could result from a motor
relearning acquired over time. In our study, the results
do not seem to depend on a motor relearning during the
time elapsed since the initial lesion or surgical repair, since
both ACL groups (B and NB) are comparable with respect
to this parameter.

857

In the same way, although we did not quantitatively

evaluate the force of the quadriceps and soleus muscles,
modications of balance recovery do not seem to be related
to a motor decit since patients and control do not show
any dierence in the positive peak of vertical acceleration,
showing evidence of the bilateral stance motor response. In
similar situations, other authors (Rudolph and SnyderMackler, 2004; Lewek et al., 2003) showed that a dierence
between ACLD copers and non-copers was not simply
related to a decit in quadriceps strength but rather to
the chronology of muscular activities. Furthermore, Williams et al. (2004) recently demonstrated the existence of
a quadriceps motor control decit after a lesion of the
ACL.
In our study, the mechanical pattern of ACLD and
ACLR patients does not dier, suggesting that passive
mechanical stability restoration of the knee after surgery
does not inuence the parameters we have studied. Then
the dierences observed in the entire patient group cannot
be interpreted in terms of passive mechanical stability
alone. Snyder-Mackler et al. (1997) has shown that anterior
laxity of the knee in ACLD subjects was poorly correlated
to the measurements of their functional capacities. A
decrease of the proprioceptive sensitivity of the knee is
known after a lesion to the ACL (Corrigan et al., 1992),
and the loss of the aerences resulting from the ACL injury
is a common decit in our ACLD and ACLR patients, this
would be responsible for the modications of balance
recovery although the absence of a proprioceptive sensitivity evaluation of the knee in our study does not allow us to
arm it. Regardless of their nature, the two strategies used
by patients are equally eective in maintaining postural
balance since they prevent the fall of the centre of gravity
that was observed after experimental blocking of muscular
aerences (Thoumie and Do, 1996).
4.2. Modications of EMG activity after ACL lesion
Our results show the same general pattern of muscular
activities in patients and in control during balance recovery. The main dierences between the control and patients
concern the latencies and durations of ipsilateral hamstrings and bilateral soleus muscle activation when the
injured limb is moving and all ipsilateral muscles recorded
when the injured limb is in stance. Two kinds of modication are observed after an ACL lesion: a global increase in
durations related to an increase in the duration of the balance recovery process and an earlier activation of some
muscles with respect to the release of the restraining device
and relative to the time of heel contact, showing evidence
of anticipation.
The increase in muscular activity duration when the
injured limb is the stance one is linked to an increase in
the balance recovery duration and probably to an increase
in lower limb stiness and in knee stability, especially during the unilateral stance. Hortobagyi and DeVita (2000)
have shown an increase in muscular activity and in lower

858

P. Colne, P. Thoumie / Clinical Biomechanics 21 (2006) 849859

limb stiness when elderly subjects step down. They conclude that this is a strategy to compensate for their reduced
motor control capacities.
We observed an increase in the contralateral activity
duration of both recorded muscles in the uninjured limb.
This suggests that muscular activity of the stance limb
helps protect the injured knee during the weight acceptance
phase of the fall after swinging with the injured limb, by
slowing down the forward progression of the body, thereby
reducing the impact of ground acceptance. This hypothesis
is supported by the results of other authors (Michel and
Do, 2002) who concluded that the function of the soleus
is to control the forward body progression in balance
recovery and gait initiation.
In all ACL groups, preparation for the ground acceptance phase of the injured moving limb is characterized by
an increase in the time elapsed between the burst of hamstrings and heel contact. These results resemble the changes
in the hamstrings activity of ACL subjects (facilitation, timing of activation, duration, time of beginning and peak) previously reported in normal or ascent gait (Kalund et al.,
1990; Lass et al., 1991; Beard et al., 1994; Kvist and Gillquist, 2001; Rudolph et al., 2001; Boerboom et al., 2001).
In the patients (B and NB), the early activation of the hamstrings compared to HC would signicantly reduce the step
execution velocity in order to stabilize the knee at heel contact of the ground and thus to protect it. Early activation of
the hamstrings with respect to heel contact also suggests an
anticipation in the control of the forward translation of the
tibia attributed to these muscles when the knee is extended in
an open kinetic chain (Solomonow et al., 1987; Baratta et al.,
1988). The early activation of the hamstrings medial is characteristic of the strategy of the B group compared to the NB
group. Since patients of the B group are the only ones to
brake before heel contact, the hamstrings medial could play
a prevalent role in anticipating the stabilization of the knee
at heel contact for this group. This hypothesis can be related
to that of Alkjaer et al. (2002) who showed that activity of
the hamstrings medial (semi tendinosus) was greater as the
knee extends in a forward lunge in ACLD coper subjects
compared with non-copers and with the control.
The early activation of the second burst of the soleus is
characteristic primarily of the behaviour of the ACLD and
NB groups. It does not seem to be related solely to the passive mechanical instability of the knee since there is the
same number of ACLD and ACLR subjects in the NB
group and that the respective time of activation of the
soleus does not dier in these two subgroups. Various
authors (Sutherland et al., 1980; Ciccotti et al., 1994) have
highlighted the role of the soleus in stabilizing the knee
during the unilateral stance phase of gait and Ciccotti
et al. (1994) reports that the activity of the soleus is
increased at this phase in ACLD subjects.
Recent studies (Elias et al., 2003; Sherbondy et al., 2003)
reported that the soleus (solicited in a mock closed kinetic
chain situation) could act as an agonist of the ACL by pulling backwards the proximal part of the tibia. This action of

the soleus is not inuenced by the position of the knee but

is increased by ankle exion (Sherbondy et al., 2003). In
our study, the moving limb is in exion at the knee and
ankle when it contacts the ground. Lower activity of the
soleus in ACLD non-copers is one of the results described
by Rudolph and Snyder-Mackler (2004) during the injured
limb ground acceptance phase after stepping, whereas it is
not in copers compared to the control. Rudolph et al.
(2001) has also described in non-copers a signicant higher
activation of the soleus during weight acceptance of gait,
that is conicting.
Without any real test permitting to classify the subjects
in copers or non-copers, it is dicult to conclude in our
study if the subjects were copers or not. Indeed, it can be
noted that ACL patients in the study of Rudolph were
all participants in high level sport activities, which is not
the case for our patients. At the time of the test our patients
have all returned to normal daily life activities (including
recreational jogging for some) without experiencing instability of the knee, which suggests they developed an
adapted strategy to stabilize the knee. In our study, the second burst of the soleus activity was signicantly earlier for
both the ACLD and NB groups what suggests that these
groups developed a strategy to stabilize the knee in a closed
kinetic chain. A previous study (Chmielewski et al., 2002)
supports the hypothesis that training to control the unilateral stance of the injured limb modies the timing of the
muscular activities stabilizing the knee and improves the
functional capacities of ACLD non-copers, highlighting
the modulation of quadriceps activity by soleus and biceps
femoris. In the same way other authors (Torry et al., 2004)
demonstrated the existence of two dierent compensatory
strategies in walking in ACLD subjects who have all recovered a good level of sports activities (potentially qualied
as copers by the author) and stressed that stabilization of
the knee can depend on dierent strategies in the same category of patients, which also seems to be in our study. Our
results are in line with these previous studies and highlight
the role of soleus according dierent strategies to stabilize
the knee whatever the treatment proposed after an ACL
lesion.
5. Conclusion
This study shows that a lesion of the ACL always results
in a change to the balance recovery motor program. The
lack of a truly characteristic pattern after rehabilitation
or surgery suggests that the changes could be the expression of a motor program resulting from the loss of the proprioceptive aerences from the ACL. Our results highlight
the role of the soleus in compensating for the injured knee
and its bilateral activation in recovering balance regardless
of the strategy used. They also suggest that the goals of
rehabilitation cannot be limited to the injured knee after
an ACL lesion (surgically repaired or not) but should
include strengthening of the soleus and improving its
activation.

P. Colne, P. Thoumie / Clinical Biomechanics 21 (2006) 849859

Acknowledgment
We thank Ms K. De Haan for reviewing the English in
this article.
References
Adachi, N., Ochi, M., Uchio, Y., Iwasa, J., Ryoke, K., Kuriwaka, M.,
2002. Mechanoreceptors in the anterior cruciate ligament contribute to
the joint position sense. Acta Orthop. Scand. 73, 330334.
Alkjaer, T., Simonsen, E.B., Peter Magnusson, S.P., Aagaard, H.,
Dyhre-Poulsen, P., 2002. Dierences in the movement pattern of a
forward lunge in two types of anterior cruciate ligament decient
patients: copers and non-copers. Clin. Biomech. (Bristol, Avon) 17,
586593.
Alkjaer, T., Simonsen, E.B., Jorgensen, U., Dyhre-Poulsen, P., 2003.
Evaluation of the walking pattern in two types of patients with
anterior cruciate ligament deciency: copers and non-copers. Eur. J.
Appl. Physiol. 89, 301308.
Baratta, R., Solomonow, M., Zhou, B.H., Letson, D., Chuinard, R.,
DAmbrosia, R., 1988. Muscular coactivation. The role of the
antagonist musculature in maintaining knee stability. Am. J. Sports
Med. 16, 113122.
Beard, D.J., Dodd, C.A., Trundle, H.R., Simpson, A.H., 1994. Proprioception enhancement for anterior cruciate ligament deciency. A
prospective randomised trial of two physiotherapy regimes. J. Bone
Joint Surg. Br. 76, 654659.
Beard, D., Soundarapandian, R., OConnor, J., Dodd, C.A.F., 1996. Gait
and electromyographic analysis of anterior cruciate ligament decient
subjects. Gait Posture, 8388.
Berchuck, M., Andriacchi, T.P., Bach, B.R., Reider, B., 1990. Gait
adaptations by patients who have a decient anterior cruciate
ligament. J. Bone Joint Surg. Am. 72, 871877.
Boerboom, A.L., Hof, A.L., Halbertsma, J.P., van Raaij, J.J., Schenk, W.,
Diercks, R.L., van Horn, J.R., 2001. Atypical hamstrings electromyographic activity as a compensatory mechanism in anterior cruciate
ligament deciency. Knee Surg. Sports Traumatol. Arthrosc. 9, 211
216.
Bonm, T.R., Jansen Paccola, C.A., Barela, J.A., 2003. Proprioceptive
and behavior impairments in individuals with anterior cruciate
ligament reconstructed knees. Arch. Phys. Med. Rehabil. 84, 1217
1223.
Chmielewski, T.L., Rudolph, K.S., Snyder-Mackler, L., 2002. Development of dynamic knee stability after acute ACL injury. J. Electromyogr. Kinesiol. 12, 267274.
Ciccotti, M.G., Kerlan, R.K., Perry, J., Pink, M., 1994. An electromyographic analysis of the knee during functional activities. II. The
anterior cruciate ligament-decient and -reconstructed proles. Am. J.
Sports Med. 22, 651658.
Corrigan, J.P., Cashman, W.F., Brady, M.P., 1992. Proprioception in the
cruciate decient knee. J. Bone Joint Surg. Br. 74, 247250.
Do, M.C., Breniere, Y., Brenguier, P., 1982. A biomechanical study
of balance recovery during the fall forward. J. Biomech. 15,
933939.
Do, M.C., Schneider, C., Chong, R.K., 1999. Factors inuencing the
quick onset of stepping following postural perturbation. J. Biomech.
32, 795802.
Elias, J.J., Faust, A.F., Chu, Y.H., Chao, E.Y., Cosgarea, A.J., 2003. The
soleus muscle acts as an agonist for the anterior cruciate ligament. An
in vitro experimental study. Am. J. Sports Med. 31, 241246.
Hortobagyi, T., DeVita, P., 2000. Muscle pre- and coactivity during
downward stepping are associated with leg stiness in aging.
J. Electromyogr. Kinesiol. 10, 117126.

859

Iwasa, J., Ochi, M., Adachi, N., Tobita, M., Katsube, K., Uchio, Y., 2000.
Proprioceptive improvement in knees with anterior cruciate ligament
reconstruction. Clin. Orthop., 168176.
Johansson, H., Sjolander, P., Sojka, P., 1991. Receptors in the knee joint
ligaments and their role in the biomechanics of the joint. Crit. Rev.
Biomed. Eng. 18, 341368.
Kalund, S., Sinkjaer, T., Arendt-Nielsen, L., Simonsen, O., 1990. Altered
timing of hamstring muscle action in anterior cruciate ligament
decient patients. Am. J. Sports Med. 18, 245248.
Kvist, J., 2004. Sagittal plane translation during level walking in poorfunctioning and well-functioning patients with anterior cruciate
ligament deciency. Am. J. Sports Med. 32, 12501255.
Kvist, J., Gillquist, J., 2001. Anterior positioning of tibia during motion
after anterior cruciate ligament injury. Med. Sci. Sports Exerc. 33,
10631072.
Lass, P., Kaalund, S., leFevre, S., Arendt-Nielsen, L., Sinkjaer, T.,
Simonsen, O., 1991. Muscle coordination following rupture of the
anterior cruciate ligament. Electromyographic studies of 14 patients.
Acta Orthop. Scand. 62, 914.
Lewek, M.D., Chmielewski, T.L., Risberg, M.A., Snyder-Mackler, L.,
2003. Dynamic knee stability after anterior cruciate ligament rupture.
Exerc. Sport Sci. Rev. 31, 195200.
Lysholm, M., Ledin, T., Odkvist, L.M., Good, L., 1998. Postural
controla comparison between patients with chronic anterior cruciate
ligament insuciency and healthy individuals. Scand. J. Med. Sci.
Sports 8, 432438.
Michel, V., Do, M.C., 2002. Are stance ankle plantar exor muscles
necessary to generate propulsive force during human gait initiation?
Neurosci. Lett. 325, 139143.
Noyes, F.R., Mooar, P.A., Matthews, D.S., Butler, D.L., 1983. The
symptomatic anterior cruciate-decient knee. Part I: The long-term
functional disability in athletically active individuals. J. Bone Joint
Surg. Am. 65, 154162.
Rudolph, K.S., Snyder-Mackler, L., 2004. Eect of dynamic stability on a
step task in ACL decient individuals. J. Electromyogr. Kinesiol. 14,
565575.
Rudolph, K.S., Axe, M.J., Buchanan, T.S., Scholz, J.P., Snyder-Mackler,
L., 2001. Dynamic stability in the anterior cruciate ligament decient
knee. Knee Surg. Sports Traumatol. Arthrosc. 9, 6271.
Sherbondy, P.S., Queale, W.S., McFarland, E.G., Mizuno, Y., Cosgarea,
A.J., 2003. Soleus and gastrocnemius muscle loading decreases
anterior tibial translation in anterior cruciate ligament intact and
decient knees. J. Knee Surg. 16, 152158.
Snyder-Mackler, L., Fitzgerald, G.K., Bartolozzi 3rd., A.R., Ciccotti,
M.G., 1997. The relationship between passive joint laxity and
functional outcome after anterior cruciate ligament injury. Am. J.
Sports Med. 25, 191195.
Solomonow, M., Baratta, R., Zhou, B.H., Shoji, H., Bose, W., Beck, C.,
DAmbrosia, R., 1987. The synergistic action of the anterior cruciate
ligament and thigh muscles in maintaining joint stability. Am. J. Sports
Med. 15, 207213.
Sutherland, D.H., Cooper, L., Daniel, D., 1980. The role of the ankle
plantar exors in normal walking. J. Bone Joint Surg. Am. 62,
354363.
Thoumie, P., Do, M.C., 1996. Changes in motor activity and biomechanics during balance recovery following cutaneous and muscular
deaerentation. Exp. Brain Res. 110, 289297.
Torry, M.R., Decker, M.J., Ellis, H.B., Shelburne, K.B., Sterett, W.I.,
Steadman, J.R., 2004. Mechanisms of compensating for anterior
cruciate ligament deciency during gait. Med. Sci. Sports Exerc. 36,
140314012.
Williams, G.N., Barrance, P.J., Snyder-Mackler, L., Buchanan, T.S.,
2004. Altered quadriceps control in people with anterior cruciate
ligament deciency. Med. Sci. Sports Exerc. 36, 10891097.