Subjects
Seven male and eleven female recreational athletes with no history
of lower-limb injury, disease, or trauma who participated in sports
that were associated with a high prevalence of injury to the anterior cruciate
ligament (for example, netball and basketball) were recruited to
participate in the study. (Netball is somewhat similar to basketball
and is played predominantly in the Commonwealth countries by females.)
The subjects had a mean age (and standard deviation) of 22.6 ± 2.5 years, a mean height of 1.70 ± 0.07 m, and a mean body mass of 65.5 ± 12.0 kg. To ensure that the selection criteria
were satisfied, each subject completed an informed consent form
and a questionnaire regarding his or her history of injury. In addition,
all testing was conducted in accordance with the National Health
and Medical Research Council Statement on Human Experimentation9.
Experimental Protocol
Each subject was required to accelerate forward for approximately
three paces, leap from the nondominant limb, and then abruptly decelerate
by landing on the dominant (test) limb in single-limb stance on a
force platform. Data were collected for a minimum of five successful
trials (that is, landing with the foot centrally located on the
force platform) under two test conditions: (1) catching a leather netball
(Mitre, India) at chest height during landing (catch), and (2) refraining
from use of any pronounced upper-limb motion upon landing (no catch).
These two conditions were randomly presented to the subjects, and
therefore the subjects did not know whether they would be required
to catch a ball during landing until the moment that the ball was
released by the thrower. Abrupt landing was selected as the deceleration
task in the present study, as such landings have been implicated
as a typical noncontact mechanism for injury of the anterior cruciate
ligament. Before performing the deceleration task, each subject
completed a standardized five-minute cycling warm-up on a wind-braked
cycle ergometer (Monark, Varberg, Sweden) (set at a workload of
50 to 100W) to ensure that all subjects had a similar warm-up.
Data Collection
Film Data
The landing action of each subject was filmed in the plane of
progression with use of a LOCAM (model 51; Redlake Imaging, Morgan
Hill, California) 16-mm pin-registered high-speed motion-picture
camera (200 Hz, 1/600-second exposure time) leveled on a tripod
0.95 m above the ground and 3.5 m from the subject to allow for
later analysis of the landing technique. The camera placement and alignment
were designed to minimize errors of perspective. A 1-m horizontal
scaling ruler, placed directly over the force platform, was filmed
before each set of trials to enable later conversion of the film
images to the actual distance in meters. The high-speed camera was
time-synchronized with the electromyographic and force data by marking
the film with an ultrabright current-limited light-emitting diode
system placed in the camera's field of view.
Ground-Reaction Force Data
Following familiarization with the task, each subject performed
the test. Three orthogonal components and the point of application
of the ground-reaction forces generated during landing were recorded
over four seconds (1000 Hz) as the subject landed on a multicomponent
force plate (type 9281B; Kistler Instrumente AG, Winterthur, Switzerland)
interfaced with a multichannel charge amplifier (type 9865A; Kistler
Instrumente AG). The force plate was secured on four steel mountings
embedded on a concrete base and covered with a granulated rubber
sports surface so that the landing surface was flush with a wooden
walkway (7.26 by 2.42 by 0.11 m). Each subject's foot placement on
landing was manually measured (in millimeters) from the point of
the location of the anterior part of the subject's shoe corresponding
to the second toe and from the point of the midline of the heel
with respect to the short and long axes of the force plate. These
foot-placement data were used in conjunction with the kinematic
and force-plate data to locate the x and y coordinates of the center
of pressure relative to the subject's landed foot10.
The four vertical, two anteroposterior, and two mediolateral channels
were summed and scaled to obtain force-time curves as input to later
determine the tibiofemoral shear forces and the peak resultant ground-reaction
force.
Muscle Activity Data
The skin-surface sites of the six superficial muscles crossing
the knee joint (the rectus femoris, vastus lateralis, vastus medialis,
semimembranosus, biceps femoris, and medial head of the gastrocnemius)
were initially prepared by shaving, abrading, and swabbing with
diluted ethanol to reduce skin impedance (less than 6 kW), measured
with the Artifact Eliminator (model CE01; CardioMetrics, Taylors
Lakes, Victoria, Australia), at the site of each electrode. Adhesive-backed
bipolar silver-silver chloride disposable infant-monitoring surface electrodes
(3M, Morden, Manitoba, Canada) then were placed over the relevant
muscle bellies (with an inter-detection-surface spacing of 10 mm),
and, after confirming correct placement of the electrodes by means
of muscle palpation and a clear electromyographic signal, the electrode
wires were taped to the subject's skin to minimize movement artifact.
A reference electrode was placed on the lateral femoral epicondyle.
The electromyographic signals were relayed from the electrodes to
a battery-powered transmitter (Telemyo; Noraxon, Scottsdale, Arizona),
with eight channels and a mass of 0.96 kg, strapped firmly to the
subject's lower back, to the Telemyo receiver. The analogue output
for the muscles from the receiver (±5 V for
full scale) were sampled at 1000 Hz (bandwidth, 0 to 340 Hz) with
use of Bioware software (version 3.06 for Windows 95; Kistler) and
stored for later analysis.
Data Analysis
Kinematic Data
After processing of the film, the two-dimensional coordinates
for the adhesive skin-marker landmarks placed on the lateral aspect
of each subject's foot, ankle, knee, hip, and shoulder were manually
digitized (200 Hz) with use of a sonic digitizer (GP9; Science Accessories,
Stratford, Connecticut) interfaced with a personal computer. The
landmarks were selected to enable later computation of foot, leg,
thigh, and trunk motion during landing. Three representative trials
per subject were analyzed, commencing with the frame representing
the initial contact between the shoe on the subject's test limb and
the force plate and continuing until fifteen frames after the frame
corresponding to the generation of the peak resultant ground-reaction
force. After digitizing, a fourth-order zero-phase-shift Butterworth
digital low-pass filter was applied to the data11 in
order to filter out any high-frequency noise (a cutoff frequency
of 11 Hz, determined with use of the residual analysis method11). The processed digitized film data
then were used to determine the kinematic variables of interest
during the landing task-namely, the hip and knee flexion angles
and the trunk angle relative to the right-hand horizontal. Variables
that were required to derive the joint-reaction forces and the moments of
force about the knee joint also were calculated; these included
the linear acceleration of the foot and leg mass centers (m/sec-2),
the distance from the segmental center of mass to the point of application
of the reaction forces (m), and the angular acceleration of the
foot and leg in the sagittal plane (deg/sec-2).
Kinetic Data
The magnitude and timing of the peak of the summed vertical and
anteroposterior ground-reaction forces and the time of initial contact
and peak resultant ground-reaction forces were determined for each
landing trial. All values were recorded both in newtons and as normalized
for body weight. The joint-reaction forces and sagittal planar net moments
of force for the knee then were calculated with use of newtonian
equations of motion and inverse dynamics. The input for the analyses
was derived from the kinematic and ground-reaction force data, combined
with estimates of the mass and moment of inertia of each segment.
The tibiofemoral shear force (FS), which acts parallel to the orientation
of the tibial plateau, then was calculated, with use of the method
of Kuster et al.12, as the sum
of the shear component of the patellar tendon force (FP) and the
shear component of the resultant joint-reaction force (R) (Fig. 1Fig. 1). The
data pertaining to knee joint geometry-that is, the angle of the
patellar tendon force (FP) relative to the knee flexion angle (b)
and the angle of the tibial plateau relative to the long axis of
the tibia (a) (Fig. 1Fig. 1)-used as input in these calculations
were obtained from an article by Nisell13.
The time to the peak tibiofemoral shear force was used to represent
the time of maximal anterior loading of the knee joint during the
landing task.
Muscle-Activation Patterns
To assess the temporal characteristics of the muscle bursts,
the raw electromyographic data were visually inspected and any signal
offset was removed with use of signal-processing software (PROG)14. The full-wave rectified data then
were filtered with use of a fourth-order zero-phase-shift Butterworth
low-pass filter11 (a cutoff frequency
of 20 Hz), and the resultant linear envelopes were screened with
a threshold detector. Cutoff frequencies ranging from 10 to 25 Hz
(in 1-Hz increments) were initially tried, and 20 Hz was selected
as it produced a smoothed representation of the raw electromyographic
signals that closely resembled the shape of the muscle tension curves
while retaining critical temporal components of the signal. Muscle
burst onsets and offsets were deemed to have occurred when fourteen
consecutive samples (a 1000-Hz sampling rate) of the linear envelope
exceeded and passed back under, respectively, a threshold of 7%
of the maximum amplitude of the linear envelope of the muscle burst of
interest. A 7% threshold was chosen after thresholds ranging from
3% to 15% were tested in trials, and the output was compared with
the muscle burst onsets and offsets manually derived from the filtered
electromyographic data and the linear envelopes. This threshold
value also has been used to calculate muscle burst onsets and offsets
in a previous study in which lower-limb muscle synchrony was examined7. Each signal then was visually inspected
to confirm the validity of the calculated results and to minimize
the probability of a type-I error. The temporal characteristics
(the duration, onset, and peak activity) of each muscle burst immediately
prior to landing, for each of the six muscles, were analyzed relative
to the timing of initial contact, the peak resultant ground-reaction
force, and the peak tibiofemoral shear force. These variables were
chosen to provide information pertaining to the effects of upper-limb
motion on lower-limb muscle-activation patterns during the deceleration
task. The muscle-activation patterns and kinematic variables were assessed
relative to the times of initial contact, the peak resultant ground-reaction
force, and the peak tibiofemoral shear force, as these times have
been identified as the most critical in the landing task7.
Statistical Analysis
The means and standard deviations for the kinematic, kinetic,
and electromyographic dependent variables were calculated for each
of the two landing conditions. Normality was confirmed for the data
with use of a Kolmogorov-Smirnov test with the Lilliefors correction,
and equal variance was confirmed with a Levene median test. The
p value to reject normality and/or equal variance was set at p £
0.05. The dependent means were analyzed with use of paired t tests.
The main purpose of this design was to establish whether upper-limb
motion significantly (p £ 0.05) affected synchrony of the lower-limb
muscles during the dynamic deceleration task.
Kinetic Data
No significant differences between the catch and no-catch test
conditions were found with respect to any of the kinetic variables
displayed at landing (Table ITable I)-that is, the subjects generated
peak forces at landing of a similar magnitude and at a consistent time
after initial contact, irrespective of whether they were required
to catch a ball.
Kinematic Data
No significant difference between the catch and no-catch test
conditions was detected with respect to knee angle at either initial
contact or peak resultant ground-reaction force, suggesting that
catching a ball upon landing did not alter knee-joint kinematics
throughout the landing phase. In contrast, paired t tests indicated
significant differences between the catch and no-catch conditions
with respect to both the mean hip angle and the mean trunk angle
at initial contact and at peak resultant ground-reaction force (Table IITable II).
Although significant (p £ 0.05), the difference in the means between
the two conditions at both initial contact and peak resultant ground-reaction
force ranged from only 3.5° to 4.5°. As the error inherent in measuring
joint angles is often considered to be as much as 5°, the difference
between the two conditions with respect to the hip and trunk angles
in the present study was not considered to be meaningful. Furthermore,
in view of the findings of one of us (J.R.S.) and colleagues15, this small change in hip angle
was not expected to substantially alter the ability of the hamstring
muscles to restrain anterior tibial translation.
Muscle-Activation Patterns
The burst onset time of the rectus femoris muscle and the burst
peak time of the gastrocnemius muscle, relative to the time of initial
contact, were significantly earlier in the catch condition than
in the no-catch condition (Fig. 2Fig. 2) (Table IIITable III).
Furthermore, the burst onset time of the rectus femoris muscle,
relative to the time of the peak tibiofemoral shear force, was significantly
earlier for the catch condition than for the no-catch condition.
However, the burst onset time of the biceps femoris muscle, relative
to the time of the peak tibiofemoral shear force, was significantly
later for the catch condition than for the no-catch condition (Fig. 3Fig. 3).
Compared with that in the no-catch condition, the upper-limb
motion involved in catching a ball resulted in significantly earlier
onset of rectus femoris muscle activity relative to the time of
the initial contact and relative to the time of the peak tibiofemoral
shear force. With the exception of vastus medialis onset relative
to the timing of the peak tibiofemoral shear force, a similar trend
was noted for earlier burst onset of the vastus muscles relative to
the time of initial contact and relative to the time of the peak
tibiofemoral shear force. This finding suggests that upper-limb
motion caused earlier activation of the anterior thigh muscles.
Although not significant, the power of these trends for the vastus muscles
was low (5% to 10%), and therefore they warrant further investigation.
Miyatsu et al.8 reported that
the moment of ball release by a subject who performed a throwing
action before landing from a jump from a box was accompanied by increased
vastus medialis activity and suppressed semimembranosus activity.
They suggested that the upper-limb action of throwing a ball resulted
in overcontraction of the quadriceps muscles relative to the hamstring
muscles. Our study supports this theory, as we found that earlier
quadriceps muscle contraction may promote anterior tibial translation, particularly
if the hamstring muscles are not activated sufficiently to generate
an antagonistic posterior tibial-drawer force.
Miyatsu et al.8 did not examine
the effects of upper-limb motion on biceps femoris or gastrocnemius
activity, so comparisons with our study are limited. However, they
noted that subjects who were required to release a ball when they
landed from a jump from a box had suppressed semimembranosus activity
just before landing. We also noted a later biceps femoris onset
relative to the timing of the peak tibiofemoral shear force in the
catch test condition, suggesting that upper-limb motion significantly
(p £ 0.05) delayed the onset of biceps femoris muscle activity. Despite
the difference between our study and that of Miyatsu et al.8 with respect to the landing techniques,
the findings of both studies suggested that motion of the upper-limbs
delayed or suppressed the activation of the muscles that act as
synergists to the anterior cruciate ligament.
The activity of the gastrocnemius muscle was found to reach a
peak significantly earlier in the catch test condition than in the
no-catch test condition. However, because of the large window of
time proposed for electromechanical delay of the lower-limb muscles
(20 to 100 msec)16, it could not
be established whether this earlier peak positively or negatively
affected the onset of gastrocnemius mechanical force generation
relative to the timing of the peak tibiofemoral shear force. As
the gastrocnemius muscle can act as a synergist to the anterior
cruciate ligament by imparting a posterior drawer on the proximal
part of the tibia, additional research is warranted to establish
the precise electromechanical delay associated with the gastrocnemius
muscle so that the effects of upper-limb motion on gastrocnemius
muscle activity can be more clearly interpreted.
The time-interval between the onset of biceps femoris muscle
activity and the onset of rectus femoris muscle activity in the
no-catch condition was longer than that in the catch condition (Fig. 3Fig. 3). It has
been postulated that the increased interval between the biceps femoris
and rectus femoris onset times in the no-catch condition provides greater
protection to the anterior cruciate ligament by allowing more time
for a posterior tibial-drawer force to be generated by the biceps
femoris before the onset of the rectus-femoris-induced anterior
tibial translation17. Therefore,
the significant alterations in the biceps femoris and rectus femoris
onset times in the catch condition may predispose the anterior cruciate
ligament to less protection from the hamstring muscles during landing.
Reflex contraction of the hamstring muscles occurs too slowly
to protect the knee effectively from injury in dynamic landing tasks18,19. All six muscles examined in
the present study consistently showed onset times before initial
contact, suggesting that the activity of the lower-limb muscles
was preprogrammed before landing20.
As subjects did not know whether the upper-limb motion of catching
was required until the ball was released by the thrower, the significant
differences observed in lower-limb muscle-activation patterns between
the catch and no-catch conditions suggest that upper-limb motion
was responsible for altering the preprogrammed activity of the lower-limb
muscles. The mechanism by which this was possible is currently unknown,
but, as the changes appear to predispose the anterior cruciate ligament
to increased potential for injury, additional investigation is warranted.
Our results suggest that upper-limb motion is not accompanied
by any significant change in knee-joint angle at the time of either
initial contact or the peak resultant ground-reaction force. Therefore, any
changes in lower-limb muscle-activation patterns in the present
study were not considered to have resulted from changes in lower-limb
or trunk kinematics during the landing task. These results are in
contrast to the findings of Miyatsu et al.8,
who reported that subjects who threw a ball while jumping from a
40-cm-high box had less knee flexion than subjects who jumped without
throwing a ball. This extended knee position was accompanied by
suppressed hamstring activity. Trunk and hip angles were not reported
by Miyatsu et al.; however, they suggested that forearm extension
may have been responsible for the decreased knee flexion angles
and the suppressed hamstring muscle recruitment at the time of landing.
The findings in the study by Miyatsu et al. suggest a link between upper-limb
motion and an increased potential for knee-joint injury, as the
hamstring muscles act synergistically with the anterior cruciate
ligament in restraining anterior tibial translation.
We detected no significant differences between the catch and
no-catch conditions with respect to either the timing or the magnitude
of the forces generated by the subjects upon landing. Therefore,
we concluded that the upper-limb motion involved in catching a ball
did not alter the kinetics of landing.
Although the upper-limb motion involved in catching a ball did
not alter the kinetics or joint angles displayed at landing, it
resulted in significantly later hamstring muscle activity and significantly earlier
quadriceps muscle activity (p £ 0.05). We concluded that upper-limb
motion involved in catching a ball during an abrupt decelerative
landing can increase the potential for anterior cruciate ligament
injury by limiting the time available for a posterior tibial-drawer
force to be generated by the hamstring muscles before the onset
of the quadriceps-muscle-induced anterior tibial translation. The exact
mechanism by which upper-limb motion can alter lower-limb muscle-activation
patterns remains a question for further investigation.