Biomechanical Measures of Neuromuscular
Control and Valgus Loading of the Knee
Predict Anterior Cruciate Ligament Injury
Risk in Female Athletes
A Prospective Study
Timothy E. Hewett,*†‡PhD, Gregory D. Myer,†MS, Kevin R. Ford,†MS,
Robert S. Heidt, Jr,§MD, Angelo J. Colosimo,‡MD, Scott G. McLean,||PhD,
Antonie J. van den Bogert,||PhD, Mark V. Paterno,†MS, PT, and Paul Succop,¶PhD
From the †Cincinnati Children’s Hospital Research Foundation, Sports Medicine Biodynamics
Center, Division of Molecular Cardiovascular Biology, Cincinnati Children’s Hospital Medical Center,
Cincinnati, Ohio, the ‡University of Cincinnati, College of Medicine, Department of Orthopaedic
Surgery, Sports Medicine, Cincinnati, Ohio, the §Wellington Orthopaedic and Sports Medicine
Center, Cincinnati, Ohio, the llDepartment of Biomedical Engineering and Orthopaedic Research
Center, Cleveland Clinic Foundation, Cleveland, Ohio, and the ¶Department of Environmental
Health, University of Cincinnati, Cincinnati, Ohio
Background: Female athletes participating in high-risk sports suffer anterior cruciate ligament injury at a 4- to 6-fold greater rate
than do male athletes.
Hypothesis: Prescreened female athletes with subsequent anterior cruciate ligament injury will demonstrate decreased neuro-
muscular control and increased valgus joint loading, predicting anterior cruciate ligament injury risk.
Study Design: Cohort study; Level of evidence, 2.
Methods: There were 205 female athletes in the high-risk sports of soccer, basketball, and volleyball prospectively measured
for neuromuscular control using 3-dimensional kinematics (joint angles) and joint loads using kinetics (joint moments) during a
jump-landing task. Analysis of variance as well as linear and logistic regression were used to isolate predictors of risk in athletes
who subsequently ruptured the anterior cruciate ligament.
Results: Nine athletes had a confirmed anterior cruciate ligament rupture; these 9 had significantly different knee posture and
loading compared to the 196 who did not have anterior cruciate ligament rupture. Knee abduction angle (P < .05) at landing was
8° greater in anterior cruciate ligament–injured than in uninjured athletes. Anterior cruciate ligament–injured athletes had a 2.5
times greater knee abduction moment (P < .001) and 20% higher ground reaction force (P < .05), whereas stance time was 16%
shorter; hence, increased motion, force, and moments occurred more quickly. Knee abduction moment predicted anterior cru-
ciate ligament injury status with 73% specificity and 78% sensitivity; dynamic valgus measures showed a predictive r2of 0.88.
Conclusion:Knee motion and knee loading during a landing task are predictors of anterior cruciate ligament injury risk in female athletes.
Clinical Relevance: Female athletes with increased dynamic valgus and high abduction loads are at increased risk of anterior cru-
ciate ligament injury. The methods developed may be used to monitor neuromuscular control of the knee joint and may help develop
simpler measures of neuromuscular control that can be used to direct female athletes to more effective, targeted interventions.
Keywords: neuromuscular control; dynamic valgus; knee joint load; anterior cruciate ligament (ACL) injury; injury prevention;
The American Journal of Sports Medicine, Vol. 33, No. 4
© 2005 American Orthopaedic Society for Sports Medicine
*Address correspondence to Timothy E. Hewett, PhD, Cincinnati
Children’s Hospital Medical Center, 3333 Burnet Avenue, MLC 10001,
Cincinnati, OH 45229 (e-mail: firstname.lastname@example.org).
No potential conflict of interest declared.
Winner of the 2004 O’Donoghue Award
AJSM PreView, published on February 8, 2005 as doi:10.1177/0363546504269591
Copyright 2005 by the American Orthopaedic Society for Sports Medicine.
Hewett et alThe American Journal of Sports Medicine
Female adolescents who participate in pivoting and jump-
ing sports suffer ACL injuries at a 4- to 6-fold greater rate
than do male adolescents participating in the same sports.
Since the passage of Title IX of the Educational Assistance
Act,male participation at the high school level has increased
less than 3% (from 3.7 to 3.8 million), whereas female par-
ticipation has increased more than 9-fold, roughly dou-
bling every 10 years (from 0.3 to 2.8 million).33This geo-
metric growth in participation has led to an alarming
increase in the number of ACL injuries in female athletes.
An estimated 38 000 ACL injuries occur in girls’ and women’s
athletics in the United States annually,41at an estimated
cost per injury of approximately $17 000.22At a national level,
surgery and rehabilitation costs associated with female
ACL injuries total approximately $646 million annually.
Most ACL injuries in female athletes occur during a
noncontact episode, typically during deceleration, lateral
pivoting, or landing tasks that are often associated with
high external knee joint loads.4,6Although high knee load
tasks occur during sports in both genders, why these tasks
result in such a greater incidence of ACL injury in females
has remained unclear. However, there appear to be 3 major
etiologic contributions to the gender disparity observed in
ACL injury rates, namely, anatomical, hormonal, and neu-
romuscular.18A number of studies of ACL injury risk fac-
tors have focused on anthropometric/anatomical measures,
such as thigh length,5height, and femoral notch width.36
Although these factors may contribute to ACL injury risk,
they are in essence nonmodifiable by nature. Hormonal
factors, particularly those linked to the follicular and ovu-
latory phases of the menstrual cycle, have also been linked
to ACL injury risk.2,37,44However, the precise means by
which they may contribute to ACL injury risk and, again,
the extent to which these contributions can be modified
remain unclear. There is increasing evidence in the litera-
ture suggesting that poor or abnormal neuromuscular con-
trol of the lower limb biomechanics, and in particular the
knee joint during the execution of potential hazardous
sporting movements, is a primary contributor to the female
ACL injury mechanism.18,21,26,30Specifically, dynamic joint
stabilization is achieved via a combination of active mus-
cle force and passive ligament restraints. The ACL may
experience potentially hazardous 3-dimensional (3D) forces
during landing and twisting sports movements if the mus-
culature that controls the knee joint does not sufficiently
dissipate the associated torques and forces. Understanding
the way neuromuscular control factors may manifest in
terms of ACL injury is crucial, as it offers the greatest
potential for interventional development and application in
high–injury risk populations such as female athletes.15
Neuromuscular training studies have been conducted
previously in an attempt to reduce ACL injury risk. A
prospective study of male soccer players, for example,
showed a significant effect of balance board exercises on
ACL injury rates.7Technique and phase-oriented neuro-
muscular training corrected jump and landing techniques
and significantly reduced abduction moments at the
knee22and decreased ACL injuries in a female interven-
tion group to a rate similar to that of males.19The inci-
dence of ACL injury in women’s handball was reduced
with training designed to improve neuromuscular knee
control during cutting and landing.32The above prospec-
tive studies demonstrate that neuromuscular training has
the potential to decrease ACL injury rates in female ath-
letes. However, the efficacy and efficiency of neuromuscu-
lar training protocols could be improved considerably if
they could be designed specifically for predetermined high-
risk athletes, with defined neuromuscular control deficits
at appropriate age and developmental levels.
A successful method for screening and identifying ath-
letes at increased risk of ACL injury is currently not avail-
able. Dynamic neuromuscular control parameters are
rarely measured in athletes before participation, with
measurements typically limited to static measures of joint
stability.38If lower limb neuromuscular control parame-
ters linked directly to ACL injury could be identified, more
effective screening regimens could therefore be imple-
mented to identify those athletes who are at increased
risk.These data would also afford the development of more
effective neuromuscular training regimens that could
effectively reduce current sports-related ACL injury rates,
particularly in females.
We collected lower limb biomechanical data in female
athletes during the execution of sports movements and fol-
lowed them prospectively to determine those who suffered
noncontact ACL injury. Our hypothesis was that females
who went on to ACL injury would demonstrate consistent
neuromuscular control differences that manifest in their
lower limb biomechanics during jump-landing tasks. We
further hypothesized that these biomechanical measures
could be used to predict ACL injury risk with high sensi-
tivity and specificity.
MATERIALS AND METHODS
This investigation was a prospective controlled cohort
study.There were 205 female adolescent soccer, basketball,
and volleyball players who were prospectively screened via
3D biomechanical analyses before their seasons. Nine ACL
injuries (7 during soccer and 2 during basketball play)
occurred in 205 screened female athletes. Ruptures of the
ACL were confirmed by arthroscopic surgery (8) or MRI
(1). The subjects who did not suffer an ACL knee injury
were classified in this study as uninjured. The ACL-
injured population was similar in age (15.8 ± 1.0 vs 16.1 ±
1.7 years; P = .63), height (167.7 ± 6.8 cm vs 164.1 ± 6.0 cm;
P = .08), and weight (61.5 ± 8.3 kg vs 59.1 ± 8.1 kg; P = .39)
to uninjured controls.
Informed written consent was obtained from all subjects
and their parents and approved by the Cincinnati
Children’s Hospital Medical Center Institutional Review
Board. After the informed consent was obtained, height,
weight, and dominant leg were recorded.The dominant leg
was determined for each subject by asking which leg she
would use to kick a ball with as far as possible.
Anthropometric measures were recorded during the labo-
Vol. 33, No. 4, 2005ACL Injury Risk in Female Athletes
Knee joint flexion-extension and adduction-abduction
were quantified for each subject over a series of drop ver-
tical jump (DVJ) trials.The DVJ (Figure 1) consisted of the
subject starting on top of a box (31 cm in height) with her
feet positioned 35 cm apart (distance measured between
toe markers).12Subjects were instructed to drop directly
down off the box and immediately perform a maximum
vertical jump, raising both arms as if they were jumping
for a basketball rebound. The DVJ has been shown to
demonstrate high within-session reliability with intra-
class correlation coefficients of greater than 0.93.11Three
successful trials were recorded for each subject, with the
requirement for success being that the impact phase of the
movement occurred on 2 precisely located force platforms
(AMTI, Boston, Mass), within the field of view of a high-
speed motion analysis system (Motion Analysis Corp,
Santa Rosa, Calif). The motion analysis system consisted
of 8 high-speed (240-Hz) digital cameras (Eagle cameras,
Motion Analysis Corp) connected through an Ethernet hub
to the data collection computer (Dell Computer Corp, Los
Angeles, Calif). The 2 force platforms collected ground
reaction force (GRF) data at 1200 Hz and were time syn-
chronized with the motion analysis data. They were
embedded into the floor and positioned 8 cm apart so that
each foot would contact a different platform during the
maneuver. The first contact on the platforms (ie, the drop
from the box) was used for analysis.
Kinematic and Kinetic Analyses
Before testing, each subject was instrumented with 25
retro-reflective markers secured to specific anatomical
locations (Figure 2).A stationary trial was first taken with
each subject in a neutral (standing) position to align her
with the global laboratory coordinate system. Each sub-
ject’s local joint coordinates were aligned to her standing
position to control for intersubject variation in anatomical
alignment (ie, zero-position valgus alignment) during the
static trial.The medial knee and ankle markers were then
removed before the execution of movement trials. Raw
marker coordinates were recorded with EvaRT software
(version 3.21, Motion Analysis Corp) and transformed into
global 3D coordinates via the direct linear transformation
method1and subsequently tracked using EvaRT. Marker
trajectories were filtered through a low-pass Butterworth
digital filter at a cutoff frequency of 9 Hz and subsequently
submitted to custom software (Kintrak, version 6.2,
Motion Analysis Corp) to quantify knee flexion-extension
and adduction-abduction demonstrated during each
trial.16The data convention was such that knee flexion and
adduction were denoted as positive (see Figure 3). Vertical
GRF was used to identify the time at initial contact with
the ground (IC) and at toe-off from the jump (TO). Knee
flexion-extension and abduction-adduction angles at IC
and the maximum abduction and flexion angles demon-
strated during the stance phase (IC-TO) were subsequently
recorded for each trial.
Inverse dynamics analyses were used to calculate knee
and hip joint abduction-adduction and flexion-extension
Figure 1. Biomechanical illustration of drop vertical jump
Figure 2. Marker locations and lower extremity joint-naming
convention. External marker locations were used to gener-
ate lower limb 3-dimensional joint kinematic and kinetic data.
Hewett et al The American Journal of Sports Medicine
moments from the motion and force data.43The force data
were filtered through a low-pass Butterworth digital filter
at a cutoff frequency (50 Hz). By convention, hip and knee
adduction and flexion moments were denoted as positive
(see Figure 3). External moments are described in this
article; for example, an external knee abduction load will
tend to abduct the knee (direct the distal tibia away from
the midline of the body), and an external knee flexion load
will tend to flex the knee (Figure 4). The peak knee abduc-
tion and hip adduction moments and peak knee and hip
flexion moments during the landing phase for each knee
joint were recorded for each trial.8
Participants were enrolled in this study during the sum-
mer of 2002 and 2003 and the fall of 2002, and they were
followed through 2 fall (soccer) and 1 winter (basketball)
sports seasons. Certified athletic trainers submitted weekly
team and individual injury reports for each study partici-
pant during the sports season. Team reports included the
number of practice and competition exposures. An injury
risk exposure was defined as one athlete participating in
one practice or match. Individual injury reports detailed
type and mechanism of injury and participation time lost
due to injury. The mean period of time between initial bio-
mechanical testing and ACL injury was 5.0 months (range,
0.6-13.1 months). The definition of an ACL injury was an
ACL rupture that occurred during a game or practice of
their competitive season. Our definition of noncontact was
the absence of a direct blow to the involved lower extrem-
ity. All ACL injuries reported over the 13-month injury
surveillance period were noncontact in nature.
Statistical means and SDs for all measured variables were
calculated for each subject group. An analysis of variance
test was used to compare values between the subject
groups, and paired t tests were used to compare differ-
ences between limbs. The 2 groups consisted of 9 subse-
quent ACL-injured knees versus the total tested popula-
tion of knees of uninjured females (390 knees total; 2 non-
dominant knees excluded due to data collection error).The
comparisons between different outcome variables emanated
from different hypotheses, and therefore a Dunn-
Bonferroni adjustment was unnecessary. For measures of
relative correlation between parameters, the Pearson cor-
relation coefficient was calculated. Statistical analyses
were conducted in SPSS (SPSS for Windows, SPSS Science
Figure 3. Lower extremity joint motion and moments naming
conventions used to define lower limb 3-dimensional joint
kinematic and kinetic data obtained from individual move-
ment trials. Text refers to both joint angles and moments. The
kinematic model was assigned 8 internal rotational degrees
of freedom: 3 at the hip, 3 at the knee, and 2 at the ankle.
Joint moments were calculated about these axes.
Figure 4. Dynamic valgus was defined as the position or
motion, measured in 3 dimensions, of the distal femur
toward and distal tibia away from the midline of the body.
Dynamic valgus may have included the indicated motions
Vol. 33, No. 4, 2005ACL Injury Risk in Female Athletes
Inc, Chicago, Ill) and SAS (SAS Institute, Cary, NC). The
hypotheses were tested in linear regression and logistic
regression models. All neuromuscular, moment, and force
variables were introduced into a logistic regression model
for injury. The generalized estimating equation model was
estimated using a logit link, a binomial distribution for the
outcomes, and a general (unstructured) covariance struc-
ture, which result in a repeated-measure logistic regres-
sion analysis for correlated binary (yes-no) outcomes. An
alpha level of .05 was used to judge statistical significance
in all models.
Knee abduction angles were significantly different
between ACL-injured and uninjured groups both at initial
contact and at maximum displacement. Specifically,
female knees that went on to ACL injury had 8.4° greater
knee abduction angles at IC, P < .01, and had 7.6° greater
at maximum, P < .01 (Figure 5), than the noninjured knees
of the controls had during landing. Significant correlations
between knee abduction angle and peak vertical GRF were
observed in ACL-injured (R = .67, P < .001) but not in
uninjured athletes (P = .44).
No difference in knee flexion angle at IC was observed
between the injured and uninjured athletes. Peak knee
flexion moment (P = .27) values were similarly observed
not to be different between groups. However, maximum
knee flexion angle at landing was 10.5° less in injured
(71.9° ± 12.0°) than in uninjured (82.4° ± 8.0°) athletes
(P < .05) (Figure 5).A significant correlation between max-
imum knee flexion angle and peak force was present in
uninjured (R = 0.33, P < .001) but not in ACL-injured ath-
letes (P = .55).
Females who went on to ACL injury had a greater stance
phase peak external knee abduction moment, –45.3 ± 28.5
N.m, compared to that of uninjured females, –18.4 ± 15.6
N.m (P < .001) (Figure 6).Vertical GRF was increased 20%
in the injured cohort (1266.1 ± 149.9 N vs 1057.8 ± 289.9
N, P < .05) (Figure 7). Significant correlations existed
between knee abduction moment and angle and peak GRF
(R = 0.74 and 0.67, respectively; P < .05) in ACL-injured
females. The hip adduction moment, although not greater
than controls on average (Figure 6), was correlated to knee
abduction moments in ACL-injured subjects (R = 0.69, P <
.05). Females who went on to ACL injury had a 16% short-
er stance time (P < .01) than did noninjured athletes.
Significant correlations existed between knee abduction
moment and angle and peak GRF (R = 0.74 and 0.67,
respectively; P < .05) in ACL-injured females.
There was no difference between sagittal plane knee
flexion-extension moments and ACL injury status. There
was a correlation between knee flexion (quadriceps)
moment and peak force in uninjured (R = 0.63, P < .001)
but not in ACL-injured athletes (P = .3). Hip sagittal plane
measures showed significant differences between groups.
Peak external hip flexion moment was greater in the ACL-
injured group (147.9 ± 33.5 N.m) in comparison to the
uninjured athletes (106.8 ± 45.3 N.m; P < .01).
Significant leg-to-leg differences in knee load were
observed in injured but not in uninjured females. Side-to-
side knee abduction moment difference was 6.4 times
greater in ACL-injured versus the uninjured females
(Figure 8) (P < .001). There were 6 dominant-leg and 3
nondominant-leg injuries in the group of 9 ACL injuries.
model depicting mean knee
joint kinematics during the
drop vertical jump at initial
contact and maximal dis-
placement in the ACL-injured
and uninjured groups (n = 9
knees and n = 390 knees,
respectively). Left, coronal
plane view of knee abduction
angle at initial contact in the
ACL-injured and uninjured
groups. Center, coronal plane
view of maximum knee
abduction angle in the ACL-
injured and uninjured groups.
Right, sagittal plane view of
maximum knee flexion angle
in the ACL-injured and unin-
Hewett et al The American Journal of Sports Medicine
There was not a significant effect of leg dominance on ACL
injury status (P = .5, χ2test).
Logistic regression analysis demonstrated that knee
abduction moments and angles (IC and peak values) were
significant predictors of ACL injury status (P < .001). Knee
abduction moments, which directly contribute to lower
extremity dynamic valgus and knee joint load,35had a sen-
sitivity of 78% and a specificity of 73% for predicting ACL
injury status. A linear regression analysis using the most
highly significant predictors (knee abduction angles, knee
abduction moments, and side-to-side differences in these
measures) of ACL injury showed a predictive r2value
equal to 0.88. Knee flexion angle was removed from the
model because it did not reach the 0.2 linear regression r2
cutoff criterion to be included in the predictive model and
was removed by a backward elimination technique.
Scattergrams of measurements of peak knee abduction
moment and knee abduction angle at IC in ACL-injured
and uninjured athletes are shown in Figure 9. The data
points from ACL-injured individuals are clearly in the
high end of the range for both peak knee abduction
moment and knee abduction angle at IC when grouped
with the noninjured subjects.
Altered Neuromuscular Control
in ACL-Injured Athletes
Gender differences in ACL injury rates and in neuromus-
cular control during potentially hazardous sporting move-
ments are well documented in the literature.8,12,19,30
However, the means by which neuromuscular control dif-
ferences manifest in terms of injury risk to a large extent
had remained unclear. This study compared 3D biome-
chanical measures in female athletes and compared data
obtained from normal control athletes to data from those
who went on to suffer ACL injury. Females suffering an
ACL injury during competition demonstrated altered neu-
romuscular control characteristics compared to noninjured
athletes, as evidenced by differences in lower limb biome-
chanics during jump-landing movement tasks. Specifically,
injured subjects demonstrated significant increases in
dynamic lower extremity valgus and knee abduction load-
ing before sustaining their injuries compared to uninjured
Dynamic Lower Extremity Valgus: Measurable
Symptom and Sensitive Predictor
The link between valgus loading and resultant increases
in ACL strain has been demonstrated experimentally
through both cadaveric and in vivo research.13,23,27,28It is
therefore likely that the increases observed in valgus
measures in the injured cohort were a significant compo-
nent of the mechanism that led to ACL rupture. Knee val-
gus angles and moments were the primary predictors of
ACL injury risk. Valgus loading can increase ACL
force.27,28Physiologic valgus torques on the knee can
increase anterior tibial translation and loads on the ACL
by several-fold.13Sagittal plane variables, however, specif-
ically knee flexion and hip and knee flexion-extension
moments, were not observed to be significant predictors of
ACL injury potential. This observation is consistent with
previous research in which multiple regression analysis
incorporating flexion angles, flexion and extension
moments, and valgus torque at the knee, hip, and ankle
demonstrated that valgus torques at the knee were the
sole significant predictors of peak landing forces.22
The potential link between excessive dynamic valgus
and ACL injury risk has been suggested previously. It has
been postulated, for example, that if an athlete is not prop-
erly aligned or if an unusual foot placement at landing
occurs, he or she may be at increased risk for injury.39
Landing in dynamic valgus (Figure 4) could be proposed as
potentially injurious to the knee.10,14The current findings
Figure 6. Mean ±1 SD knee abduction and hip adduction
moment at landing. ***P < .001.
Figure 7. Vertical ground reaction force mean ±1 SD time
normalized to 100% stance phase.
Figure 8. Dominant versus nondominant differences in
measurements of components of dynamic valgus (knee
abduction angle) in injured versus uninjured female athletes.
***P < .001.
Vol. 33, No. 4, 2005 ACL Injury Risk in Female Athletes
indicate that athletes should be encouraged to avoid exces-
sive valgus alignment at landing, cutting, or decelerating
to minimize their risk of knee injury.
Current observations also suggest that the gender-based
disparity observed in ACL injury rates during sporting
movements may, to a large extent, be explained by the con-
comitant differences displayed in the coronal plane joint
motions and moments during these movements. The
observed increased knee abduction motion and moments
in females before ACL injury suggest decreased neuro-
muscular control of the lower extremity in the coronal
plane. This likely reflects differences in contraction pat-
terns or insufficient neuromuscular adaptation of the
adductors of the hip and flexors of the knee to the high
demands of sports.27,40Muscular contraction can decrease
the dynamic valgus laxity of the knee 3-fold.29Joint com-
pression through muscular co-contraction allows more of
the knee adduction load to be absorbed by articular con-
tact forces, which can protect the ligaments from high
loads (Figure 10). It is likely that more equal distribution
of forces transmitted across both the medial and lateral
compartments of the knee joint would lead to decreased
landing forces.9,22In addition, a decreased dynamic valgus
moment would decrease the risk of medial femoral condy-
lar liftoff from the tibial plateau. Biomechanical studies
have previously established the relationship between
femoral condylar liftoff and ACL injury risk.29,40
Although ACL injuries may occur too quickly for reflex-
ive muscular activation, athletes may be able to adopt or
“preprogram” safer movement patterns that reduce injury
risk during landing or pivoting or unexpected loads or per-
turbations during sports movements. Preparticipation
neuromuscular training may result in safer movement
patterns that act to reduce or eliminate high knee abduc-
tion loads. Hamstrings and quadriceps can be 40% to 80%
activated at the time that the foot touches the ground.34
Coactivation of the hamstrings and quadriceps is proposed
to protect the knee joint not only against excessive anterior
drawer but also against knee abduction and dynamic
lower extremity valgus.3If the hamstrings are under-
recruited or weak, quadriceps activation may be reduced
to provide the net flexor moment required to perform the
movement. Deficits in strength and activation of the ham-
strings may thus directly limit the potential for muscular
co-contraction to protect ligaments. This potential absence
of muscular control of the joint may lead to a “ligament-
dominant” or “quadriceps-dominant” profile in the female
athlete.21If hamstrings recruitment is high, the quadri-
ceps can be activated more while still allowing for a net
flexor moment. Similar mechanisms apply to muscular
protection against torsional loading, in which gender dif-
ferences have been identified.45Taking these facts into
consideration in conjunction with the suggested link
between valgus motion/loading and ACL injury, it is likely
that more effective prevention strategies, aimed at
improving the muscle contributions to dynamic knee sta-
bility, and in particular in the coronal plane control, can be
developed with a strong potential for success.
Need for Injury-Prediction Measures in Athletes
There is a need for effective screening programs to be put
into practice that would enable successful identification of
athletes at risk of ACL injury.17The current findings
demonstrate that decreased neuromuscular control as evi-
denced by increased dynamic valgus and external knee
Figure 9. Scattergram of measurements of components of dynamic valgus (peak knee abduction moment and knee abduction
angle at initial contact) in injured (data points shown as Xs) versus uninjured female athletes. ht, height; wt, weight.
Hewett et al The American Journal of Sports Medicine
abduction moments can predict increased ACL injury risk
in a large percentage of individuals.The question remains,
therefore, as to whether it is possible to accurately and
consistently identify those individuals who display these
potential causal factors. Current technologies afford the
potential for lower limb kinematics and parameters linked
to sporting movements to be measured with confi-
dence.4,12,22,31This suggests that adequate screening
strategies can be implemented using these technologies to
successfully identify those individuals who are at risk of
ACL injury.A potential limitation may be that 3D analysis
of sporting movements comes at high financial and time
costs, thus possibly limiting the potential for implementa-
tion of large-scale screening programs.Attempts should be
made to correlate 3D measures of lower limb biomechani-
cal data with more simple 2D measurements so that
screening on a larger scale can be implemented. There is
evidence to suggest that such correlations may be possi-
ble.12Research of this nature should be pursued in more
detail considering the large number of athletes participat-
ing in sports in which ACL injury is prolific.
Neuromuscular Adaptations With Training: Potential
Significant positive alterations in movement biomechanics
and lower extremity muscle strength and recruitment are
possible in female athletes after neuromuscular training.22
Changes due to training are typically greater in females,
as their baseline neuromuscular performance levels are
often lower compared to those of males.24Previous studies,
for example, have shown specifically that neuromuscular
training can significantly decrease peak GRF and abduc-
tion motion and moments at the knee.22Intensive neuro-
muscular training can significantly increase fat-free mass,
vertical jump height, and balance measures in
females.22,25,42Muscular power has also been shown to
increase up to 44% in females with 6 weeks of training.
Prospective studies have demonstrated that training
resulting in changes such as those mentioned above has
the potential to decrease ACL injury rates in female ath-
letes.19,32However, training regimens that specifically tar-
get the causal factors of ACL injury and reduce their
impact via neuromuscular modifications need to be devel-
oped further. The results of the current study will enable
evidence-based training protocols to be developed that aim
specifically, which aim specifically to modify neuromuscu-
lar control patterns that contribute to increased valgus
motion and valgus loading. If generalized training such as
that discussed above can reduce injury rates, then the
potential of training tailored to high-risk individuals with
identified neuromuscular control deficits may prove more
Further work also appears necessary to determine when
these training programs should be implemented.
Prospective randomized trials, for example, should be con-
ducted to determine at what age or stage of development
young athletes should begin to be trained.20This intensive
neuromuscular training may induce a “neuromuscular
spurt” in female adolescents that could dramatically
increase neuromuscular control and decrease injury risk.
This approach may decrease the number of high-risk indi-
viduals and may make subsequent training protocols more
Limitations of the Study
It is likely that ACL injury has a multifactorial etiology,
with unmeasured factors influencing outcome. Injury data
demonstrate that many physical and psychological param-
eters affect injury rates. There are several possible con-
tributing and confounding variables that were not con-
trolled for in the study design, which included school,
team, age/grade, aggressiveness, foot pronation, quadri-
ceps angle, femoral notch width, reliable menstrual status
reporting, and blood hormone levels. However, neuromus-
cular parameters appear to be a major determinant. Some
of these potential factors may be alterable but are, at a
The use of only soccer, basketball, and volleyball players
is a limitation to the generalizability of the findings of this
Figure 10. A, free body diagram of forces acting on the tibia,
showing the sagittal plane equilibrium between articular con-
tact force, hamstrings force, quadriceps force, and ACL
force. In this example, quadriceps force contributes to ACL
force, whereas hamstrings and articular contact force pro-
tect the ACL. B, free body diagram of forces acting on the
tibia, showing the frontal plane equilibrium between external
dynamic valgus load, articular contact force, quadriceps
force, medial hamstrings force, and ACL force. Under exter-
nal dynamic valgus loading, contact shifts to the lateral com-
partment. The moment balance with respect to the contact
point shows that both quadriceps and medial hamstrings
force help the ACL (and medial collateral ligament, not
shown) stabilize the joint against dynamic valgus loading.
Under a given dynamic valgus load, any reduction in these
muscle forces will cause an increase in ligament loading. C,
articular contact force; Q, quadriceps force; H, hamstrings
force; MH, medial hamstrings force; V, valgus load.
Vol. 33, No. 4, 2005 ACL Injury Risk in Female Athletes
study. However, gender differences in injury incidence
have been demonstrated in several sports, including bas-
ketball, soccer, lacrosse, team handball, and volleyball.
Differences in neuromuscular control measures probably
exist in most gender-paired sports. Therefore, the associa-
tions between ACL injury status and neuromuscular con-
trol measures and injury in female basketball, volleyball,
and soccer players should be comparable to those found in
female athletes participating in other sports.
The significant effects of neuromuscular control meas-
ures on ACL injury status in female athletes were
observed in this study. The observed differences were of a
magnitude outside of the protocol’s measurement error.
However, long-term longitudinal studies testing the
hypotheses of this study must be undertaken to better
answer these questions. Future ACL ruptures are more
likely to occur for individuals with similar kinematics and
kinetic profiles to those who have already suffered an ACL
event. The results of the present study may have been
stronger with a longer survey period.All of those who were
injured in the present study had been playing sports for
multiple years.If the high-risk athletes have had excessive
valgus for multiple years, this would mean that it may
take years before excessive valgus results in an ACL
injury. A weakness of this type of prospective study design
is that there can be a substantial change in neuromuscu-
lar status (ie, excessive valgus) over the course of a long
(multiyear) injury-tracking period. We are currently con-
ducting studies to assess the effects of time, growth, and
maturation on these neuromuscular indices.
Conclusion: Measures of Neuromuscular Control and
Joint Load Predict Injury Risk in Athletes
Previously, it was known that ACL injury risk was greater
in female than in male athletes. What was not known was
the mechanism underlying this gender disparity,and there
was no method that could identify those female athletes at
increased ACL injury risk. In the current study, measures
of neuromuscular control and knee joint load were
prospectively examined relative to ACL injury status.
These measurements were employed to help delineate
whether lower limb neuromuscular control parameters
could be used to predict ACL injury risk in female athletes.
Specifically, it appears that increased valgus motion and
valgus moments at the knee joint during the impact phase
of jump-landing tasks are key predictors of an increased
potential for ACL injury in females. Current technologies
enable these data to be readily measured noninvasively in
a large number of athletes, within a relatively short time
period. Hence, such measurements appear necessary in
future screening and prevention studies aimed at reducing
ACL injury rates, particularly in females. Future research
needs to focus on controlled, prospective longitudinal stud-
ies of defined populations of female athletes who are fol-
lowed through multiple sports seasons to correlate chang-
ing neuromuscular profiles to injury risk, which predis-
pose the athlete to ACL injury. Studies that test improved
neuromuscular training approaches are also of the highest
We acknowledge funding support from the National
Institutes of Health Grant R101-ARO49735-01A1.We also
acknowledge Ralph Buncher, PhD, for his assistance with
the experimental design and for helping to perform the
statistical analysis of the data, and Robert Mangine of
Novacare Rehabilitation, Brian Colley of Wellington
Orthopaedic and Sports Medicine Center, and Spectrum
Rehabilitation for assisting in the recruitment of schools
and athletes for participation in the studies. We thank
Richelle Carlonas of Wellington Orthopaedic and Sports
Medicine Center for her assistance with editing of the
manuscript. In addition, we thank the school districts, the
athletes, and the administrators involved in the study,
especially Stu Eversole of the Lakota School District; the
athletic trainers who participated in the study and who
tracked injuries, including Pate Cagle, ATC, Jennifer
Miller, ATC, Lisa Sweeterman, ATC, Cathy Conley, ATC,
among others; and the coaches who were so pivotal in
making this study a success, including Dan Purcell, Tara
Schafer-Kalkhoff, Tracey Kornau, Kate Berz, and Heather
1. Abdel-Aziz YI, Karara HM. Direct linear transformation from com-
parator coordinates into object space coordinates in close-range
photogrammetry. In: Symposium on Close-Range Photogrammetry.
Falls Church, Va: American Society of Photogrammetry; 1971:1-18.
2. Arendt EA, Bershadsky B, Agel J. Periodicity of noncontact anterior
cruciate ligament injuries during the menstrual cycle. J Gend Specif
3. Besier TF, Lloyd DG, Ackland TR. Muscle activation strategies at the
knee during running and cutting maneuvers. Med Sci Sports Exerc.
4. Besier TF, Lloyd DG, Cochrane JL, Ackland TR. External loading of
the knee joint during running and cutting maneuvers. Med Sci Sports
5. Beynnon B, Slauterbeck J, Padua D, et al. Update on ACL risk fac-
tors and prevention strategies in the female athlete. Paper presented
at: National Athletic Trainers’ Association 52nd Annual Meeting and
Clinical Symposia; 2001; Los Angeles, Calif.
6. Boden BP, Dean GS, Feagin JA Jr, Garrett WE Jr. Mechanisms of
anterior cruciate ligament injury. Orthopedics. 2000;23:573-578.
7. Caraffa A, Cerulli G, Projetti M, Aisa G, Rizzo A. Prevention of anteri-
or cruciate ligament injuries in soccer: a prospective controlled study
of proprioceptive training. Knee Surg Sports Traumatol Arthrosc.
8. Chappell JD, Yu B, Kirkendall DT, Garrett WE. A comparison of knee
kinetics between male and female recreational athletes in stop-jump
tasks. Am J Sports Med. 2002;30:261-267.
9. Crenshaw SJ, Pollo FE, Calton EF. Effects of lateral-wedged insoles
on kinetics at the knee. Clin Orthop. 2000;375:185-192.
10. Dufek JS, Bates BT. Biomechanical factors associated with injury
during landing in jumping sports. Sports Med. 1991;12:326-337.
11. Ford KR, Myer GD, Hewett TE. Reliability of dynamic knee motion in
female athletes. Paper presented at: American Society of
Biomechanics Annual Meeting; September 25-27, 2003; Toledo,
12. Ford KR, Myer GD, Hewett TE. Valgus knee motion during landing in
high school female and male basketball players. Med Sci Sports
13. Fukuda Y, Woo SL, Loh JC, et al. A quantitative analysis of valgus
torque on the ACL: a human cadaveric study. J Orthop Res.
Hewett et al The American Journal of Sports Medicine
14. Gerberich SG, Luhmann S, Finke C, et al. Analysis of severe injuries
associated with volleyball activities. Phys Sportsmed. 1987;15:75-79.
15. Griffin LY, American Academy of Orthopaedic Surgeons. Prevention
of Noncontact ACL Injuries. Rosemont, Ill: American Academy of
Orthopaedic Surgeons; 2001.
16. Grood ES, Suntay WJ. A joint coordinate system for the clinical
description of three-dimensional motions: application to the knee. J
Biomech Eng. 1983;105:136-144.
17. Harmon KG, Dick R. The relationship of skill level to anterior cruciate
ligament injury. Clin J Sport Med. 1998;8:260-265.
18. Hewett TE. Neuromuscular and hormonal factors associated with
knee injuries in female athletes: strategies for intervention. Sports
19. Hewett TE, Lindenfeld TN, Riccobene JV, Noyes FR. The effect of
neuromuscular training on the incidence of knee injury in female ath-
letes: a prospective study. Am J Sports Med. 1999;27:699-706.
20. Hewett TE, Myer GD, Ford KR. Puberty decreases dynamic knee sta-
bility in female athletes: a potential mechanism for increased ACL
injury risk. J Bone Joint Surg Am. 2004;86A:1601-1608.
21. Hewett TE, Paterno MV, Myer GD. Strategies for enhancing proprio-
ception and neuromuscular control of the knee. Clin Orthop.
22. Hewett TE, Stroupe AL, Nance TA, Noyes FR. Plyometric training in
female athletes: decreased impact forces and increased hamstring
torques. Am J Sports Med. 1996;24:765-773.
23. Kanamori A, Woo SL, Ma CB, et al. The forces in the anterior cruci-
ate ligament and knee kinematics during a simulated pivot shift test:
a human cadaveric study using robotic technology. Arthroscopy.
24. Kraemer WJ, Keuning M, Ratamess NA, et al. Resistance training
combined with bench-step aerobics enhances women’s health pro-
file. Med Sci Sports Exerc. 2001;33:259-269.
25. Kraemer WJ, Mazzetti SA, Nindl BC, et al. Effect of resistance train-
ing on women’s strength/power and occupational performances.
Med Sci Sports Exerc. 2001;33:1011-1025.
26. Lloyd DG. Rationale for training programs to reduce anterior cruciate
ligament injuries in Australian football. J Orthop Sports Phys Ther.
2001;31:645-654; discussion 661.
27. Lloyd DG, Buchanan TS. Strategies of muscular support of varus and
valgus isometric loads at the human knee. J Biomech. 2001;34:1257-
28. Markolf KL, Burchfield DM, Shapiro MM, Shepard MF, Finerman GA,
Slauterbeck JL. Combined knee loading states that generate high
anterior cruciate ligament forces. J Orthop Res. 1995;13:930-935.
29. Markolf KL, Graff-Redford A, Amstutz HC. In vivo knee stability: a
quantitative assessment using an instrumented clinical testing appa-
ratus. J Bone Joint Surg Am. 1978;60:664-674.
30. McLean SG, Lipfert S, van den Bogert AJ. Effect of gender and
defensive opponent on the biomechanics of sidestep cutting. Med
Sci Sports Exerc. 2004;36:1008-1016.
31. McLean SG, Neal RJ, Myers PT, Walters MR. Knee joint kinematics
during the sidestep cutting maneuver: potential for injury in women.
Med Sci Sports Exerc. 1999;31:959-968.
32. Myklebust G, Engebretsen L, Braekken IH, Skjolberg A, Olsen OE,
Bahr R. Prevention of anterior cruciate ligament injuries in female
team handball players: a prospective intervention study over three
seasons. Clin J Sport Med. 2003;13:71-78.
33. National Federation of State High School Associations. 2002 High
School Participation Survey. Indianapolis, Ind: National Federation of
State High School Associations; 2002.
34. Neptune RR, Wright IC, van den Bogert AJ. Muscle coordination and
function during cutting movements. Med Sci Sports Exerc.
35. Schipplein OD, Andriacchi TP. Interaction between active and pas-
sive knee stabilizers during level walking. J Orthop Res. 1991;9:113-
36. Scoville CR, Williams GN, Uhorchak JM, et al. Risk factors associat-
ed with anterior cruciate ligament injury. In: Proceedings of the 68th
Annual Meeting of the American Academy of Orthopaedic Surgeons.
Rosemont, Ill: American Academy of Orthopaedic Surgeons;
37. Slauterbeck JR, Hardy DM. Sex hormones and knee ligament injuries
in female athletes. Am J Med Sci. 2001;322:196-199.
38. Smith J, Laskowski E. The preparticipation physical examination:
Mayo Clinic experience with 2,739 examinations. Mayo Clin Proc.
39. Steele J, Milburn P. Ground reaction forces on landing in netball. J
Hum Mov Stud. 1987;13:399-410.
40. Tibone JE, Antich TJ, Fanton GS, Moynes DR, Perry J. Functional
analysis of anterior cruciate ligament instability. Am J Sports Med.
41. Toth AP, Cordasco FA. Anterior cruciate ligament injuries in the
female athlete. J Gend Specif Med. 2001;4:25-34.
42. Tropp H, Odenrick P. Postural control in single-limb stance. J Orthop
43. Winter DA. Biomechanics and Motor Control of Human Movement.
New York, NY: John Wiley; 1990.
44. Wojtys EM, Ashton-Miller JA, Huston LJ. A gender-related difference
in the contribution of the knee musculature to sagittal-plane shear
stiffness in subjects with similar knee laxity. J Bone Joint Surg Am.
45. Wojtys EM, Huston LJ, Schock HJ, Boylan JP, Ashton-Miller JA.
Gender differences in muscular protection of the knee in torsion in
size-matched athletes. J Bone Joint Surg Am. 2003;85:782-789.