

COPYRIGHT © 2002 BY THE JOURNAL OF BONE AND JOINT SURGERY, INCORPORATED

Ankle and Knee Coupling in

Patients with Spastic Diplegia:
Effects of Gastrocnemius-
Soleus Lengthening
BY ADRIAN BADDAR, MD, KEVIN GRANATA, PHD, DIANE L. DAMIANO, PT, PHD,
DAVID V. CARMINES, PHD, JOHN S. BLANCO, MD, AND MARK F. ABEL, MD
Investigation performed at the Motion Analysis and Motor Performance Laboratory,
Kluge Children’s Rehabilitation Center and Research Institute, University of Virginia, Charlottesville, Virginia

Background: Empirical observations of subjects with an equinus gait have suggested that there is coupled mo-
tion between the ankle and knee such that, during single-limb stance, the ankle moves into equinus as the knee
extends. Since the gastrocnemius-soleus muscle-tendon unit spans both joints, we hypothesized that this muscle-
tendon unit may be responsible for the coupling and that lengthening of the gastrocnemius-soleus muscle alone
would result in greater ankle dorsiflexion as well as greater knee extension in single-limb stance, effectively un-
coupling these joints. The concept that gastrocnemius-soleus lengthening may promote knee extension is
counter to the popular notion that crouch gait may result if the hamstrings are not lengthened concomitantly.
Methods: A retrospective review identified thirty-four subjects with specific kinematic characteristics of equinus
gait, and their gait was compared with that of normal children. Of the thirty-four subjects, eleven (twenty-two
limbs) subsequently underwent isolated midcalf lengthening of the gastrocnemius and soleus muscles with use
of a recession technique. Gait analysis including joint kinematics and joint kinetics, electromyography, and
physical examination were performed to test the hypothesis.
Results: We found that, unlike the normal subjects, the patients with an equinus gait pattern had a positive cor-
relation (r = 0.7) between ankle and knee motion during single-limb stance. As hypothesized, ankle plantar flex-
ion occurred while the knee moved into extension during single-limb stance. Calculations of the lengths of the
gastrocnemius-soleus muscle-tendon units showed them to be short throughout the gait cycle (p < 0.0001). Af-
ter gastrocnemius-soleus recession, peak ankle dorsiflexion (p < 0.001) and peak ankle power (p < 0.001)
shifted to occur later in stance than they did in the preoperative gait cycle. Furthermore, the magnitude of peak
power increased (p < 0.001) in late stance despite the added length of the gastrocnemius-soleus muscle-tendon
unit. The electromyographic amplitude of the gastrocnemius-soleus was reduced during loading (p < 0.02), and
this finding, together with the kinetic changes, suggested that muscle tension was reduced. Changes at the
knee were less pronounced but included greater knee extension at foot contact (p < 0.01). No increase in the
knee flexion angle or extension moment occurred in midstance after the surgery.
Conclusions: Patients with an equinus gait pattern function with a shortened gastrocnemius-soleus muscle-
tendon unit, and this results in coupled motion between the ankle and knee during single-limb stance. Length-
ening, with use of a recession technique, shifted ankle power generation and dorsiflexion to a later time in
stance with no tendency to increase midstance knee flexion. Knee extension did increase at foot contact, but
excessive midstance knee flexion persisted and was likely due to concomitant contracture of the hamstrings.

S
pastic diplegia is the most common pattern of motor
impairment in patients with cerebral palsy1,2. In these pa-
tients, motor impairment is due to a number of deficits,
including poor muscle control, weakness, impaired balance,
hypertonicity, and spasticity. As a consequence, muscle-tendon
units frequently become contracted over time, contributing
to malalignment of the extremity during gait. Muscle-tendon
lengthening is a common treatment to address the shortened
muscle-tendon units and improve gait function.
Equinus gait, one of the more common abnormal pat-
terns of gait of patients with spastic diplegia, is characterized
by forefoot strike to initiate the cycle and premature plantar
flexion in early stance to midstance3. Equinus gait is typically
associated with increased stance-phase knee flexion or crouch.
 
THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG A N K L E A N D KN E E C O U P L I N G I N PA T I E N T S W I T H S P A S T I C
VO L U M E 84-A · N U M B E R 5 · M AY 2002 D I P L E G I A : E F F E C T S O F G A S T RO C N E M I U S -S O L E U S L E N G T H E N I N G

Consequently, surgical management often involves single-stage,

simultaneous lengthening of both the gastrocnemius-soleus
muscle-tendon unit and the distal hamstrings to address both
equinus and crouch4. In fact, pervasive coexisting hamstring
spasticity is thought to exacerbate crouch following isolated
lengthening of the gastrocnemius-soleus muscle as the ankle
becomes more dorsiflexed4-7. Alternatively, increased knee flex-
ion in stance could conceivably result from shortening (static
contracture) of the gastrocnemius-soleus muscle-tendon unit
since the gastrocnemius-soleus spans both joints. For this to
be the case, knee extension would have to be coupled with
ankle plantar flexion during the gait cycle. If this were true,
lengthening of the gastrocnemius-soleus could produce both
ankle dorsiflexion and greater knee extension, provided that
hamstring spasticity and contracture were not overwhelming.
The effects of muscle-tendon surgery on equinus and
crouch have been evaluated in several studies8-10, but the re-
sults have been confounded by the performance of multiple,
simultaneous muscle-tendon lengthenings, including both
the hamstrings and the gastrocnemius-soleus muscle. Re-
cently, we reported the results of lengthening of only the
gastrocnemius-soleus muscle-tendon unit in a small subset
of patients with spastic diplegia3. In this group, increased
ankle dorsiflexion was not accompanied by increased knee
flexion, or crouch, during gait; instead, knee extension dur-
ing stance increased. On the basis of these observations, we
hypothesized that shortening of the gastrocnemius-soleus
muscle-tendon unit necessitated a coupling between the an-
kle and knee such that knee flexion occurred with ankle dor-
siflexion and knee extension occurred with ankle plantar
flexion during stance in patients with spastic diplegia and an
equinus gait. Furthermore, we reasoned that isolated reces-
sion of the gastrocnemius-soleus muscle-tendon unit would
uncouple the knee and ankle so that knee extension could
occur with ankle dorsiflexion.
The purpose of this investigation was to evaluate equinus
gait relative to normal gait. Specifically, the length and excur-
sion of the gastrocnemius-soleus muscle-tendon unit in relation
to knee and ankle kinematics were studied. We hypothesized
Fig. 1
that (1) patients with equinus gait and contracture of the
Comparison of joint kinematics and gastrocnemius-soleus muscle-
gastrocnemius-soleus muscle-tendon unit have mechanical
tendon lengths between the patients with an equinus gait and the
coupling between the ankle and knee, and (2) lengthening of
normal subjects.
the gastrocnemius-soleus muscle-tendon unit would improve
stance-phase ankle dorsiflexion while simultaneously increasing
knee extension, effectively uncoupling these two joints. Because ture reversal into plantar flexion during single-limb stance.
the moment of the gastrocnemius-soleus muscle-tendon unit is Thus, the slope of the sagittal ankle kinematics during single-
greater for the ankle than for the knee, we expected the greatest limb stance was negative (downsloping), in contrast to the
effects at the ankle following the operation. normal, positive slope (Fig. 1). These criteria excluded chil-
dren with midfoot breakdown (excessive mobility and sub-
Methods luxation of the talonavicular and calcaneocuboid joints).
Subjects Patients were also excluded if they had had any prior muscle-
hildren with spastic diplegia and an equinus gait who had tendon surgery, botulinum injections within the last six
C been evaluated in our laboratory from December 1995
through July 2000 were selected from the laboratory data-
months, or other surgery such as rhizotomy that may affect
gait. Thirty-four children (seventeen girls and seventeen boys),
base. Equinus gait was identified by peak dorsiflexion during ranging in age from 2.6 to 15.2 years and with a mean age
loading (in the first 20% of the gait cycle) followed by prema- (and standard deviation) of 7.2 ± 3.4 years, met the criteria
 
THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG
VO L U M E 84-A · N U M B E R 5 · M AY 2002
for an equinus gait pattern. Two of the subjects required the
use of a posterior walker, and the remainder walked without
aids.
Of the thirty-four children, eleven (seven girls and
four boys) underwent bilateral gastrocnemius-soleus reces-
sion at the midpart of the calf at the study institution. These
eleven children were studied to evaluate the effects of this
operation on knee and ankle kinematics. The age range of
this group was 3.6 to 11.2 years, with a mean of 6.1 ± 2.0
years, at the time of surgery. Concomitant surgical proce-
dures were not considered exclusionary if the muscles affect-
ing knee motion were not involved. Of the eleven patients,
seven underwent gastrocnemius-soleus recession alone; one,
simultaneous lengthening of the posterior tibialis for treat-
ment of a varus deformity; one, simultaneous lengthening of
the adductor longus (without gracilis release); one, simulta-
neous external rotational osteotomy of the tibia; and one, si-
multaneous plantar fascia release.
Kinematic data collected from a group of seventy-eight
age-matched children without a physical disability or move-
ment disorder were used for comparison. These children, who
were family members of the hospital staff or healthy siblings of
the subjects, ranged in age from three to thirteen years, with a
mean age of 6.9 ± 2.8 years.
The study protocol was reviewed by the institution’s Hu-
man Investigation Committee, and all subjects or their parents
gave consent for participation.
Evaluation
All of the subjects underwent a clinical examination and gait
analysis with a six-camera, three-dimensional motion-analysis
system (VICON, Oxford Metrics, Oxford, England). Fifteen
reflective markers were placed by a trained physical therapist
to mark the lower extremities and pelvis of each subject. All
subjects walked barefoot along a 10-m carpeted walkway at a
freely selected speed. Force-plates under the walkway recorded
ground reaction forces during walking trials, and joint mo-
ments were expressed as internal moments to counter the
ground reaction force. Trials were recorded until each individ-
ual had made contact with a force-plate with each foot at least
three times. The data from successful trials for each limb were
averaged and were reported at 2% intervals of the gait cycle.
Pertinent data, including the popliteal angle, maximum passive
knee extension, and maximum dorsiflexion with the knee both
flexed and extended, were obtained by clinical examination.
Electromyographic data were collected from the seven sub-
jects (twelve extremities) undergoing isolated gastrocnemius-
soleus recession. The electromyographic activity of the rectus
femoris (quadriceps), hamstrings, anterior tibialis, and gas-
trocnemius muscles was monitored with surface electrodes
placed over the point of maximum bulk of each muscle. Signals
were collected with use of an eight-channel telemetric system
(Noraxon, Scottsdale, Arizona), full-wave rectified, filtered, and
averaged over three successful trials. Electromyographic sig-
nals were normalized to peak amplitude for between-subject
comparisons.
A N K L E A N D KN E E C O U P L I N G I N PA T I E N T S W I T H S P A S T I C
D I P L E G I A : E F F E C T S O F G A S T RO C N E M I U S -S O L E U S L E N G T H E N I N G
Operative Protocol
Gastrocnemius-soleus recession was performed in eleven pa-
tients who exhibited static contractures and who would be fol-
lowed at the testing institution. To perform this procedure, a 3
to 4-cm posterior midline incision was made in the leg at the
level of the gastrocnemius-soleus muscle-tendon junction.
With the broad aponeurotic expansion of the tendon exposed,
a transverse incision was made through both the gastrocne-
mius and the soleus fascia. The foot was then dorsiflexed to at
least a neutral position with the knee held in extension. A cast
was applied and worn for three weeks postoperatively; then a
flexible (posterior leaf-spring) ankle-foot orthosis was used
during the day until six months postoperatively, to maintain
stretch on the calf muscle-tendon units. Clinical examination
and gait analysis were performed within the two months prior
to the surgery and again after sufficient recovery had taken
place. Follow-up motion data were obtained from nine to
twenty-three months (mean, fifteen months) postoperatively;
thus, we are reporting the early changes after surgery. At the
time of follow-up, the children’s ages ranged from 4.6 to 12.4
years, with a mean age of 7.2 ± 2.1 years.
Analysis
To assess ankle-knee coupling, the thirty-four patients (fifty-six
limbs) with an equinus gait were compared with the seventy-
eight normal subjects (156 limbs). Loss of limb markers or
midfoot breakdown precluded use of data from twelve limbs in
the equinus group. The comparison of preoperative and post-
operative values was based on the data obtained from the eleven
subjects (twenty-two limbs) undergoing surgery at the study in-
stitution. On the basis of the kinematic data, gastrocnemius-
soleus muscle length was calculated for each subject with use
of the two-dimensional, sagittal-plane musculoskeletal model
shown in Figure 211. Lengths were calculated as the distance
from the origin to the insertion and were measured at 2% inter-
vals throughout the gait cycle. The origin and insertion were
estimated as the points of attachment along the femur and cal-
caneus, respectively. The angle of the calcaneal tuberosity rela-
tive to the plane of the forefoot was estimated on the basis of
anatomic models12. All lengths were normalized to tibial length.
Motion coupling between the ankle and the knee was
determined for each subject by the correlation of ankle and
knee kinematic data during single-limb stance with use of
Pearson r procedures. A positive correlation signified that
greater ankle plantar flexion tended to be accompanied by in-
creased knee extension, and a negative relationship indicated a
link between greater ankle dorsiflexion and increased knee ex-
tension. Statistical analyses were performed on the kinematic
and passive-range-of-motion data with use of factorial (sub-
ject group) or repeated-measures (for comparisons between
preoperative and postoperative values) multiple analysis of
variance. P values of <0.05 were considered significant. When
p was >0.05, a power analysis was performed to determine the
effect size necessary to detect a change with a beta of 0.20.
Data displayed graphically represent the group mean calcu-
lated every 2% of the gait cycle, while means for intergroup
 
THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG A N K L E A N D KN E E C O U P L I N G I N PA T I E N T S W I T H S P A S T I C
VO L U M E 84-A · N U M B E R 5 · M AY 2002 D I P L E G I A : E F F E C T S O F G A S T RO C N E M I U S -S O L E U S L E N G T H E N I N G

in the normal group), but it occurred a mean of 10% later in

the swing phase (that is, at 82% of the cycle) in the patients
with an equinus gait (p < 0.0001). Overall, knee excursion was
reduced in the group with an equinus gait because of incom-
plete knee extension.

Ankle-Knee Couple
A negative correlation (r = –0.7) between knee and ankle mo-
tion was found in the normal group, whereas the group with
an equinus gait had a similarly high correlation but in a posi-
tive direction (r = 0.7). Thus, in the normal subjects the trend

Fig. 2
The model employed to determine gastrocnemius-soleus muscle-
tendon length. The results are expressed as a percentage of tibial
length. Lg = length of the gastrocnemius muscle-tendon unit, Lt = tibial
length, Xα = vector moment arm of the Achilles tendon insertion from
the ankle joint center, Xκ = vector moment arm of the gastrocnemius
origin from the knee joint center, and Ri = vector rotation matrix about
the ankle (i = α) and knee (i = β).

comparison were obtained by considering specific points dur-

ing intervals of the gait cycle.

Results
Kinematics of Normal and Equinus Gait
he average sagittal kinematics of knee and ankle motion
T in the normal group and the group with an equinus gait
are shown in Figure 1. In the equinus gait, foot contact was
initiated by landing on the forefoot. During loading, the mean
peak dorsiflexion occurred at 15% of the cycle, compared with
42% in the normal group (p < 0.001), and it reached only 5.6°,
compared with 13° in the normal group (p < 0.0001). In the
equinus gait, plantar flexion rather than dorsiflexion oc-
curred throughout single-limb stance. Thus, the slope of the
ankle kinematics during single-limb stance was −0.32°/% of
the gait cycle in the patients compared with 0.20°/% of the gait
cycle in the normal subjects (p < 0.0001). The timing of maxi-
mum plantar flexion, which occurred at push-off, was similar
for the two groups. It reached a mean of –24° in the group
with an equinus gait, which was two times greater than the
mean of –12° in the normal group (p < 0.0001). After push-
off, the ankle steadily dorsiflexed in both groups, but swing-
phase dorsiflexion was limited to a mean of –4.3° in the group
with an equinus gait compared with a mean of 5.7° in the nor-
mal subjects (p < 0.001).
The group with an equinus gait began the cycle at foot
contact with five times more knee flexion (mean, 40°) than the
normal group did (mean, 8°) (p < 0.0001), and the knee re-
mained more flexed throughout stance phase (Fig. 1). At mid-
stance, the knee reached its maximum extension at a mean of
16° of flexion in the group with an equinus gait compared Fig. 3
with a mean of 5° of flexion in the normal group (p < 0.0001). Changes in joint kinematics and gastrocnemius-soleus muscle-tendon
During swing phase, peak knee flexion was similar in the two length following gastrocnemius-soleus recession in eleven patients, as
groups (mean, 60° in the group with an equinus gait and 61° compared with preoperative and normal values.
 
THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG
VO L U M E 84-A · N U M B E R 5 · M AY 2002
during single-limb stance was for the knee to extend while the
ankle was dorsiflexing, whereas the subjects with an equinus
gait had an opposite trend, with the knee extending while the
ankle was plantar flexing (Fig. 1).
Gastrocnemius-Soleus Length
The length curves demonstrated that, in the patients with an
equinus gait, the gastrocnemius-soleus muscle-tendon unit
was operating in a shortened position throughout the gait
cycle (Fig. 1). The mean length of the muscle-tendon unit was
118% (normalized to tibial length) compared with 121% (p <
0.0001) in the normal subjects; however, the dynamic range
was comparable, being 9% and 10% for the patients and the
normal subjects, respectively. The muscle-tendon unit nor-
mally reaches a maximum length of 125% at 44% of the gait
cycle, which is at the end of single-limb stance, when simulta-
neous maximum ankle dorsiflexion and maximum knee exten-
sion occur. In contrast, in the group with an equinus gait, the
mean maximum length reached only 121% (p < 0.0001) and
occurred much earlier, at 22% of the gait cycle (p < 0.0001).
Furthermore, the mean minimum length of the gastrocnemius-
soleus muscle-tendon unit was 112% in the patients with an
equinus gait, compared with 115% in the normal subjects.
Postoperative Results
Spatial-Temporal Data
Following recession of the gastrocnemius-soleus in the eleven
patients, the mean stride length increased 11%, from 0.71 to
Fig. 4
Moment and power changes at the ankle and knee following gastrocnemius-soleus muscle-tendon recession.
A N K L E A N D KN E E C O U P L I N G I N PA T I E N T S W I T H S P A S T I C
D I P L E G I A : E F F E C T S O F G A S T RO C N E M I U S -S O L E U S L E N G T H E N I N G
0.79 m (p < 0.04). The mean cadence remained unchanged, at
125 steps/min, and the change in mean velocity (0.81 m/sec
preoperatively compared with 0.88 m/sec postoperatively) did
not reach significance (p = 0.07) with the number of subjects
evaluated (an effect size of 0.14 m/sec was needed for β = 0.20).
Passive Range of Motion
Preoperatively, the patients undergoing the gastrocnemius-
soleus recession were high-functioning community ambula-
tors who were not dependent on walking aids (gross motor
function classification13 type I or II). All patients had full pas-
sive knee extension, and the mean pelvic tilt was 5.6° anteri-
orly. The popliteal angle (mean, 135° ± 20°) did not change
significantly following the operation. However, the mean pas-
sive ankle dorsiflexion increased from 5.1° to 12.0° (p < 0.004)
with the knee flexed and increased from 5.5° of plantar flexion
to 2.1° of dorsiflexion with the knee extended (p < 0.001).
Kinematics
Following gastrocnemius-soleus recession, the mean peak ankle
dorsiflexion increased from 7.1° to 15° (p < 0.03) and, more
importantly, occurred later in the gait cycle, at a mean of 34%
of the gait cycle (closer to push-off and thus more like normal
gait) compared with 14% of the cycle preoperatively (p <
0.001) (Fig. 3). Also, mean dorsiflexion of the ankle at foot
contact increased from −3° preoperatively to 6° postoperatively
(p < 0.001). Mean maximum plantar flexion during push-off
was significantly reduced, from –23° to –6.4° (p < 0.001), while
 
THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG
VO L U M E 84-A · N U M B E R 5 · M AY 2002
the mean ankle dorsiflexion during midswing was increased
from –4° to 5° (p < 0.0001). At foot contact, the knee demon-
strated improved mean extension, from 40° to 32° (p < 0.01).
Midstance knee extension did not change significantly (from a
mean of 17° to a mean of 13°; p = 0.2, with an effect size of 7°
necessary for β = 0.20), while midswing knee flexion was re-
duced (from a mean of 60° to a mean of 51°, p < 0.04). Total
knee excursion did not change significantly (p = 0.6).
Postoperative Gastrocnemius-Soleus Length
The mean maximum length of the gastrocnemius-soleus in-
creased from 121% preoperatively to 124% postoperatively (p <
0.002), becoming similar to normal. The timing of the maxi-
mum length also shifted from 20% of the gait cycle to 33% (p <
0.04), but it still occurred significantly earlier in the gait cycle
than it did in the normal subjects, in whom it occurred at 44%
of the gait cycle (p < 0.004). The minimum length of the gas-
trocnemius-soleus muscle changed from 112% preoperatively
to 117% postoperatively (p < 0.0004) so that the overall excur-
sion of the muscle-tendon unit was reduced from 9% to 7%.
Ankle-Knee Couple
The positive correlation between knee and ankle kinematic
data during single-limb stance was lost after surgery (r = 0.05)
(Fig. 3). However, the normal negative correlation was not
achieved. The gastrocnemius-soleus length curve in single-limb
stance was relatively flat and shifted into the dorsiflexion
range, with greater length occurring later in stance. Thus, the
Fig. 5
Electromyographic (EMG) changes following gastrocnemius-soleus muscle-tendon recession.
A N K L E A N D KN E E C O U P L I N G I N PA T I E N T S W I T H S P A S T I C
D I P L E G I A : E F F E C T S O F G A S T RO C N E M I U S -S O L E U S L E N G T H E N I N G
operation reduced the ankle-knee coupling but did not restore
normal knee and ankle interactions.
Moment and Power
As a result of the operation, the ankle plantar flexion moment
during loading was reduced whereas the moment at push-off
was increased (Fig. 4). Therefore, the mean peak ankle mo-
ment was shifted from 12% to 46% of the gait cycle (p <
0.005). Also, the mean peak ankle plantar flexion power was
shifted to later in stance and had greater magnitude postoper-
atively (p < 0.001).
The mean peak knee extension moment increased sig-
nificantly, from 0.16 to 0.53 N-m/kg (p < 0.001), after loading
and was associated with a mean increase in the peak knee
power generation from 0.34 to 0.47 W/kg (p < 0.02) to ini-
tiate single-limb stance. Also, the knee absorption of power
was greater at push-off postoperatively (–0.28 compared with
–0.81 W/kg preoperatively) (p < 0.03).
Electromyographic Changes
The pattern of muscle activation did not change greatly after
surgery (Fig. 5). The burst in gastrocnemius activity at loading
was not as apparent postoperatively (p < 0.02) and was associ-
ated with less ankle power absorption. Furthermore, the re-
duced loading activity was followed by reduced ankle moment
and power generation in early stance (Fig. 4). Gastrocnemius
activity was slightly elevated during the remainder of stance,
corresponding to the time of progressively increasing ankle
 
THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG
VO L U M E 84-A · N U M B E R 5 · M AY 2002
dorsiflexion, although this elevation did not reach significance
(p = 0.07). The only change noted in the electromyographic
activity of the quadriceps following gastrocnemius-soleus re-
cession was in response to loading, when a significantly higher
mean amplitude was found postoperatively (p < 0.005). This
increased activity corresponded with the increased extension
moment. The electromyographic findings in the hamstrings
and anterior tibialis did not change significantly as a result of
the operation. Although the tibialis anterior had less relative
activity in late stance and swing postoperatively, significance
was not found because of high variability.
Additional Results
Postoperative hip and pelvic motion in the sagittal plane was
not seen to differ significantly from that seen preoperatively.
Discussion
he equinus gait pattern, as defined in this study, is com-
T mon in patients with spastic diplegia. As shown, the pat-
tern is characterized by ankle dorsiflexion during loading,
followed by progressive plantar flexion in stance while the knee
joint is simultaneously extending. Because of this apparent
motion couple between the ankle and knee, we investigated
whether the abnormal motion at the knee might be imposed
by constraints of the shortened gastrocnemius-soleus muscle-
tendon unit that spans both joints. We believed that, if this
were the case, then isolated lengthening of the gastrocne-
mius-soleus muscle-tendon unit could lead to improvements
in both knee and ankle motion as well as gait function. Isolated
lengthening of the gastrocnemius-soleus muscle-tendon unit in
patients with spastic diplegia has been discouraged because of
the fear that the pervasive influence of spastic hamstrings
would lead to excessive knee flexion (crouch) in stance4,5,7. With
this in mind, we selected patients for isolated gastrocnemius-
soleus recession when they could passively achieve full knee ex-
tension and yet had a contracture of the gastrocnemius muscle
that prevented passive movement into dorsiflexion with the
knee extended. Because the soleus muscle does not cross the
knee joint, the effect on knee motion may have been achieved
without dividing the soleus fascia. However, in practice our
surgical objective was to obtain normal dorsiflexion and this
required recessing the fascia of both the gastrocnemius and the
soleus. We employed a recession technique rather than a
lengthening of the Achilles tendon as the latter procedure has
been shown experimentally to have more deleterious effects on
total force-generating capacity14. Furthermore, the recession
technique is relatively less invasive than z-lengthening of the
Achilles tendon and thus presumably can be performed in a
more uniform manner.
Our results showed that, although the gastrocnemius-
soleus muscle-tendon unit was shorter than normal throughout
the gait cycle in patients with equinus gait, lengthening of
the gastrocnemius-soleus unit affected mainly ankle kine-
matics. However, this procedure did alter the deviant pattern
of coupled motion between the ankle and knee. Following
the operative lengthening, the ankle excursion shifted in the
A N K L E A N D KN E E C O U P L I N G I N PA T I E N T S W I T H S P A S T I C
D I P L E G I A : E F F E C T S O F G A S T RO C N E M I U S -S O L E U S L E N G T H E N I N G
dorsiflexion direction, with maximum dorsiflexion observed
later in single-limb stance. Moreover, the peak dorsiflexion
occurred while the knee was extending, which indicated un-
coupling. At foot contact, the knee was more extended, but
the degree of knee extension and the slope of the knee curve
during midstance did not change significantly with the num-
ber of subjects tested. Nevertheless, we did not find the in-
crease in either knee extension moment or magnitude of
knee flexion in midstance that was predicted by others4,5,7.
Thus, there was no effect that would promote crouch over
the term of this study. In fact, the mean knee moment re-
mained anterior to the knee in midstance, promoting ex-
tension rather than crouching. However, progressive crouch
is a risk whenever a patient demonstrates persistent knee
flexion during single-limb stance, especially when a patient
has more muscle weakness and is dependent on a walker.
Adding length to the gastrocnemius-soleus unit is not
necessarily beneficial if power generation is disturbed. How-
ever, despite reduction in total ankle and muscle length excur-
sion, the timing of maximum ankle power generation shifted
to later in stance and the magnitude increased. This finding
suggests that preoperatively the gastrocnemius-soleus muscle-
tendon unit was in such a shortened state that effective power
generation was reduced in late stance. Lengthening, as de-
scribed, restores length-tension relationships to favor greater
power generation in late stance, when the body center is far-
ther in front of the stance limb. Consequently greater power
transfer may be achieved to promote longer strides.
The gastrocnemius-soleus recession had no obvious ad-
verse functional consequences, but the task that we studied was
relatively simple. It is possible that deficits would have been
more obvious if we had assessed running or fast walking be-
cause these activities require earlier power coupling between
the ankle and knee15. Furthermore, during these more inten-
sive activities, energy is stored in the elasticity of the tendon
and postoperative changes in the tendon properties may not be
favorable. More work in this area is needed to ascertain the im-
pact of muscle-tendon surgery on more intensive activities.
Although the moments and powers at the knee were rel-
atively low compared with those at the ankle, knee extension
moment and power generation in early stance were increased
postoperatively and persistently high knee power absorption
was found at push-off. These findings coupled with the elec-
tromyographic data suggest that the greater extension mo-
ment during loading still triggered spastic contraction of the
quadriceps muscle to account for the early power generation.
In late stance, the increased absorption at the knee (due to a
power couple with the ankle) may induce a stretch in the rec-
tus femoris (quadriceps) muscle and account for the reduced
knee flexion in swing.
The electromyographic data help to clarify the effects of
the operation and were consistent with the kinetic data. The
added length of the gastrocnemius-soleus muscle reduced the
average amplitude of the muscle’s signal during loading as well
as the ankle moment and power in early stance. Thus, adding
length to the muscle-tendon unit seemed to alter the position
 
THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG
VO L U M E 84-A · N U M B E R 5 · M AY 2002
of the foot at contact and thereby attenuated the stretch re-
sponse. Later in stance, as this muscle was reaching its maxi-
mum length, mean amplitudes increased. Quadriceps activity
remained high during loading because the extension moment
and eccentric loading persisted. Anterior tibialis activity was
quite variable, but the mean signal in late stance and early
swing seemed to decrease after the operation. This relative de-
crease may indicate that alleviation of the equinus reduced the
coactivation of the anterior tibialis necessary for ankle stability
in stance and reduced the activation needed for ankle dorsiflex-
ion in swing. The net effect of lengthening the gastrocnemius-
soleus muscle was to increase dorsiflexion during swing.
The model of gastrocnemius-soleus length used in this
study assumed that the anatomy was similar and proportional
across individuals. Furthermore it assumed that the foot was
acting as a solid lever when in fact midfoot motion, especially
during forefoot loading, is common in people with cerebral
palsy. We were careful to evaluate video data and to exclude
subjects with excessive midfoot subluxation, but still our
model may have overestimated the maximum length of the
gastrocnemius-soleus muscle-tendon unit in the patients, as
some forefoot dorsiflexion must have occurred during load-
ing. This may partially account for the fact that the mean total
muscle-tendon excursion in the patients approached normal.
None of our patients were treated with Botox (botuli-
num) because they were studied before Botox was used rou-
tinely at our institution. Furthermore, this medication is ideal
for dynamic muscle-tendon deformities, before the fibrous
component of the muscle-tendon contracture is too advanced.
However, if the muscle-tendon contracture is not too great,
then Botox treatment may be a useful first option. Previously
published studies of Botox injection into the gastrocnemius-
soleus have focused exclusively on ankle kinematics16,17; thus its
effect on ankle-knee couple remains speculative.
In conclusion, in subjects with an equinus gait pattern,
shortening of the gastrocnemius-soleus muscle-tendon unit re-
sulted in some coupled motion between the knee and ankle
joints during stance. As hypothesized, since the gastrocnemius-
soleus moment arm for the ankle was greater than that for the
knee, lengthening of the gastrocnemius-soleus muscle-tendon
unit resulted in greater changes at the ankle. The data suggest
that, if passive-range-of-motion and kinematic criteria are met
as described here, lengthening of the gastrocnemius-soleus
muscle-tendon unit through a recession technique will pro-
duce greater ankle dorsiflexion during stance and swing as well
as decreased knee flexion at foot contact and will not adversely
increase knee flexion, or crouch, in midstance. The outcome
References
1. Blair E, Stanley F. Issues in the classification and epidemiology of cerebral
palsy. Mental Retard Devel Disab Res Rev. 1997;3:184-93.
2. Hagberg B, Hagberg G, Olow I, van Wendt L. The changing panorama of
cerebral palsy in Sweden. VII. Prevalence and origin in the birth year period
1987-90. Acta Paediatr. 1996;85:954-60.
3. Abel MH, Damiano DL, Pannunzio M, Bush J. Muscle-tendon surgery in di-
plegic cerebral palsy: functional and mechanical changes. J Pediatr Orthop.
1999;19:366-75.
A N K L E A N D KN E E C O U P L I N G I N PA T I E N T S W I T H S P A S T I C
D I P L E G I A : E F F E C T S O F G A S T RO C N E M I U S -S O L E U S L E N G T H E N I N G
of the operation also included a delay in maximum ankle
power generation during the gait cycle without reducing the
magnitude.
The findings of this study also confirm that when a
muscle-tendon unit is shown to operate in a shortened position
throughout the gait cycle, lengthening can enhance gait
function3. The electromyographic changes following the opera-
tion suggest that the gastrocnemius-soleus and quadriceps
muscles are especially susceptible to stretch activation with sud-
den eccentric lengthening, as occurs in the loading phase, and
the rectus femoris (quadriceps) is also potentially susceptible
after push-off 18. Following the gastrocnemius-soleus lengthen-
ing, the electromyographic response was attenuated during
loading to account for the increased dorsiflexion. The persis-
tently increased knee flexion to initiate stance may reflect over-
activity or pathological shortening of the hamstrings. Since the
lengths of the hamstrings were not assessed in this study, one
can only speculate about whether simultaneous lengthening
would have been more beneficial. The reduced joint excursions
and persistently abnormal joint kinematics following the oper-
ation underscore that other neuromuscular deficits, including
poor muscle control, weakness, and abnormal cocontraction,
also play important roles.

NOTE: The authors appreciate the work of Felix Cheung, who helped to develop and to apply the
screening algorithms used for patient selection.
Adrian Baddar, MD
Kevin Granata, PhD
David V. Carmines, PhD
John S. Blanco, MD
Department of Orthopaedic Surgery, P.O. Box 800159, University of Vir-
ginia, Charlottesville, VA 22908
Diane L. Damiano, PT, PhD
Human Performance Laboratory, Barnes Jewish Hospital, 4555 Forest
Park Parkway, St. Louis, MO 63108
Mark F. Abel, MD
Kluge Children’s Rehabilitation Center and Research Institute, 2270 Ivy
Road, Charlottesville, VA 22903
In support of their research or preparation of this manuscript, one or
more of the authors received grants or outside funding from the Ortho-
paedic Research and Education Fund. None of the authors received pay-
ments or other benefits or a commitment or agreement to provide such
benefits from a commercial entity. No commercial entity paid or
directed, or agreed to pay or direct, any benefits to any research fund,
foundation, educational institution, or other charitable or nonprofit
organization with which the authors are affiliated or associated.
4. Gage JR. Gait analysis in cerebral palsy. New York: Cambridge University
Press; 1991. Diplegia and quadriplegia; p 151-72.
5. Sutherland DH, Cooper L. The pathomechanics of progressive crouch gait in
spastic diplegia. Orthop Clin North Am. 1978;9:143-54.
6. Rang M. Cerebral palsy. In: Morrissy RT, editor. Lovell and Winter’s pediatric
orthopaedics. 3rd ed. Philadelphia: JB Lippincott; 1990. p 465-506.
7. Bleck EE. Orthopaedic management in cerebral palsy. Philadelphia: JB Lippin-
cott; 1987. Spastic diplegia; p 282-391.
 
THE JOUR NAL OF BONE & JOINT SURGER Y · JBJS.ORG A N K L E A N D KN E E C O U P L I N G I N PA T I E N T S W I T H S P A S T I C
VO L U M E 84-A · N U M B E R 5 · M AY 2002 D I P L E G I A : E F F E C T S O F G A S T RO C N E M I U S -S O L E U S L E N G T H E N I N G

8. Etnyre B, Chambers CS, Scarborough NH, Cain TE. Preoperative and postop-
erative assessment of surgical intervention for equinus gait in children with
cerebral palsy. J Pediatr Orthop. 1993;13:24-31.
9. Nene AV, Evans GA, Patrick JH. Simultaneous multiple operations for spas-
tic diplegia. Outcome and functional assessment of walking in 18 patients.
J Bone Joint Surg Br. 1993;75:488-94.
10. Rose SA, DeLuca PA, Davis RB 3rd, Ounpuu S, Gage JR. Kinematic and ki-
netic evaluation of the ankle after lengthening of the gastrocnemius fascia in
children with cerebral palsy. J Pediatr Orthop. 1993;13:727-32.
11. Frigo C, Pedotti A. Determination of muscle length during locomotion. In:
Asmussen E, Jorgensen K, editors. Biomechanics VI-A. Baltimore: University
Park Press; 2001. p 355-60.
12. Ettema GJ. Gastrocnemius muscle length in relation to knee and ankle joint
angles: verification of a geometric model and some applications. Anat Rec.
1997;247:1-8.
13. Palisano R, Rosenbaum P, Walter S, Russel D, Wood E, Galuppi B. Develop-
ment and reliability of a system to classify gross motor function in children
with cerebral palsy. Dev Med Child Neurol. 1997;39:214-23.
14. Delp SL, Zajac FE. Force- and moment-generating capacity of lower-extremity
muscle before and after tendon lengthening. Clin Orthop. 1992;284:247-59.
15. Lacquaniti E, Grasso R, Zago M. Motor patterns in walking. News Physiol
Sci. 1999;14:168-74.
16. Sutherland DH, Kaufman KR, Wyatt MP, Chambers HG, Mubarak SJ. Double-
blind study of botulinum A toxin injections into the gastrocnemius muscle in
patients with cerebral palsy. Gait Posture. 1999;10:1-9.
17. Crenna P. Spasticity and ‘spastic’ gait in children with cerebral palsy. Neuro-
sci Biobehav Rev. 1998;22:571-8.
18. Ubhi T, Bhakta BB, Ives HL, Allgar V, Roussounis SH. Randomised double
blind placebo controlled trial of the effects of botulinum toxin on walking in
cerebral palsy. Arch Dis Child. 2000;83:481-7.

Bone and Joint Decade

-

Support the
Bone and Joint
Decade!

For more information, please contact:

The Bone and Joint Decade Secretariat
SE-221 85 Lund, Sweden
Phone: +46 46 17 71 61
Fax: +46 46 17 71 67
E-Mail: bjd@ort.lu.se

w w w. b o n e a n d j o i n t d e c a d e . o r g