1554527x, 2001, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1016/S0736-0266(00)90004-4 by Nat Prov Indonesia, Wiley Online Library on [24/03/2023].

See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Journal of
Orthopaedic
Research
ELSEVIER Journal of Orthopaedic Research 19 (2001) 213-220
www.elsevier.nl/locate/orthres

Ankle ligament injury risk factors: a prospective study of college

athletes
Bruce D. Beynnon Per A. Renstrom a, Denise M. Alosa, Judith F. Baumhauer ',
Pamela M. Vacek
a Department qf Orthopaedics & Rehabilitation Mc Clure Musculoskeletul Resrurch Center Stufford Hull, Uniiwsity of I Prmont, Burlington,
I'T 05405-0084 USA
Department of Orthopaedic Surgery, Unioersity of' Rochester Medical Center, 601 Elmwood Acenue. Rochester, N Y 14642 USA
Department of Medical Biostatistics, 25D Hills Agricultural Science Uniaersity of I'ermont, Burlington, VT 05405, U S A
Received 28 August 1998; accepted 24 May 2000

Abstract

Over two million individuals suffer ankle ligament trauma each year in the United States, more than half of these injuries are
severe ligament sprains; however, very little is known about the factors that predispose individuals to these injuries. The purpose of
this study was to determine the risk factors associated with ankle injury. We performed a prospective study of 118 Division I
collegiate athletes who participated in soccer, lacrosse, or field hockey. Prior to the start of the athletic season, potential ankle injury
risk factors were measured, subjects were monitored during the athletic season, and injuries documented. The number of ankle
injuries per 1000 person-days of exposure to sports was 1.6 for the men and 2.2 for the women. There were 13 injuries among the 68
women (19%) and seven injuries among the 50 men (13%), but these proportions were not significantly different. Women who played
soccer had a higher incidence of ankle injury than those who played field hockey or lacrosse. Among men, there was no relationship
between type of sport and incidence of injury. Factors associated with ankle ligament injury differ for men relative to women.
Women with increased tibia1 varum and calcaneal eversion range of motion are at greater risk of suffering ankle ligament trauma,
while men with increased talar tilt are at greater risk. Generalized joint laxity, strength, postural stability, and muscle reaction time
were unrelated to injury. 0 2001 Orthopaedic Research Society. Published by Elsevier Science Ltd. All rights reserved.

Introduction have shown that the incidence of ankle sprains was

The most common injury in recreational and athletic
activity is ankle ligament injury [9]. The cost for treat-
ment and rehabilitation of these injuries has been re-
ported to be two billion dollars a year [33]. Reducing the
incidence of ankle ligament injuries depends on identi-
fying the conditions under which such injuries occur
(e.g. extrinsic variables such as environmental condi-
tions, equipment, etc.) and individual characteristics
(e.g. intrinsic variables such as height, and ankle specific
measures) that might predispose athletes to such inju-
ries.
In a recent review of the literature, we determined
that there was little consensus with regard to what
constitutes an ankle injury risk factor [l]. Sitler et al. [31]
* Corresponding author. Tel.: + 1-802-656-4257 fax: +1-802-656-
4247.
E-mail address: beynnon@salus.med.uvm.edu (B.D. Beynnon).
0736-0266/01/$ - see front matter 0 2001 Orthopaedic Research Society. Published by Elsevier Science Ltd. All rights reserved.
PII: S 0 7 3 6 - 0 2 6 6 ( 0 0 ) 9 0 0 0 4 - 4
lower in athletes that previously suffered an ankle sprain
and wore a brace in comparison to those that did not.
Even though height and weight have not been shown to
be independent risk factors for ankle sprains [31], when
expressed as a mass moment of inertia, or height
squared multiplied by weight, it was predictive of ankle
sprains [26]. Many other intrinsic and extrinsic ankle
injury risk factors have been studied [I .23,27,28]; how-
ever, most of these investigations h a w been limited by
their retrospective design and have not gathered expo-
sure data.
The purpose of this investigation was to perform a
comprehensive, prospective investigation of the risk
factors for inversion ankle ligament sprain. Based on a
review of the literature, we identified generalized and
ankle joint laxities, anatomic alignment of the foot and
ankle, strength, postural sway, and muscle reaction time
as potential risk factors, and hypothesized that one or a
combination could be used to identify athletes at risk for
ankle ligament injury.

214 B.D. Beynnon et ul. / Journal of Orthopaedic Research 19 (2001)213-2470
Materials and methods
One hundred 18 Division I National Collegiate Athletic Associa-
tion varsity athletes (50 men and 68 women ranging between 18 and 23
years of age), who competed in either lacrosse, soccer or field hockey
participated during the 1994 and 1995 seasons. Subjects were excluded
if they had previously sustained an ankle ligament sprain. undergone
surgery of the foot or ankle, sustained trauma to the lower extremity,
or if they used ankle support. Prior to preseason training, the fol-
lowing potential ankle injury risk factors were evaluated.
Demographic injormation was recorded by each participant on a
health history survey, which included sport, gender, history of injury to
the lower extremity, height, weight, and leg dominance.
Generalized und ankle joint laxities. The modified Beighton method
[6] was used by the same investigator (DMA) to quantify generalized
joint laxity. This involved measuring the limits of motion for the fifth
MCP joints, thumbs, elbows, knees, and trunk. A score of four or
more, out of a maximum of nine, identified an athlete as having gen-
eralized joint laxity, while those who scored less were not. Laxity of
both ankles was evaluated with the anterior drawer (a valid meastire of
the anterior talofibular ligament [8,15,19,35]) and talar tilt exams
(a sensitive measure of the anterior talofibular and calcaneofibular
ligaments [18,25,19]) by the same investigator (DMA). The anterior
drawer test was performed with the subject seated, knees flexed at 90"
and the ankles at lo" of plantar flexion [30]. Anterior displacement of
the talus relative to the fibula was evaluated: no or minimal displace-
ment was assigned a grade of 0, moderate displacement (less than
4 mm) a grade 1+, and severe displacement (greater than 4 mm) a
grade 2+. The talar tilt test was performed and graded as either
positive (e.g. 20" or more of calcaneal inversion, or a 10" difference
between right and left ankles [10,11]) or negative.
Anatomic alignment oj'the,foot and ankle with the subject nonweight-
hearing was measured with a goniometer [17], by the same investigator
(DMA) to provide consistent intra-rater reliability [5,14,32]. Both sides
of the lower extremity were evaluated. The subtalar neutral position
was used as a reference, or zero point, from which measures of foot
and ankle position and range of motion were made. This was estab-
lished with the subject prone, and the examiner palpating the medial
and lateral portions of the talus with the thumb and forefinger from
the anterior aspect of the ankle, while the other hand was used to
maximally abduct and adduct the forefoot to determine the point
during range of motion at which the head of the talus was felt equally
between both fingers. Once congruency of the talus was established,
slight dorsi-flexion was applied to the fourth and fifth metatarsal heads
from the plantar aspect of the foot to maintain the subtalar joint in this
neutral position. From the SUbtdlar neutral position, forefoot-to-
rearfoot position was evaluated by measuring the angle between a line
constructed through the fifth and first metatarsal heads (forefoot) and
a line perpendicular to the mid-sagittal axis of the calcaneus (rearfoot).
If the head of the first metatarsal was in a higher plane than the head of
the fifth metatarsal. the foot was in a forefoot varus position. This was
followed by evaluation of Rearfoot position. This was assessed with
the calcaneus maintained in the subtalar neutral location with the
thumb and finger of one hand while the other hand was used to
measure the angle between the lines drawn such that they bisected the
calcaneus and calf in the midsagittal planes. Next, calcaneal inversion
range-of-motion was evaluated relative to the subtalar neutral position
by applying an inversion moment to the calcaneus until the limit of
rotation was met, and measuring the angle between the lines that bi-
sected the calcaneus and calf in the midsagittal planes. Similarly,
calcaneal eversion range-of-motion was measured relative to the sub-
talar neutral position by applying an eversion moment to the calcaneus
until the end of rotation was observed and measuring the angle be-
tween the lines that bisected the calcaneus and calf in the midsagittal
planes. Finally, ankle dorsi-flexion range of motion was measured with
the knee in extension and then 90" of flexion. The talocrural joint was
passively dorsi-flexed while the subtalar joint remained in neutral. The
angle between a line constructed parallel to the base of the foot and a
line parallel to the long axis of the fibula was measured.
Anmtomic alignment of the foot and ankle with the subject weight-
heuring was evaluated while the subject was standing with knees ex-
tended and feet placed shoulder-width apart. Anatomic foot type was
classified as either pronated, neutral, or supinated using the criterion
reported by Dahle et al. [12]. The foot was pronated if the following
1554527x, 2001, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1016/S0736-0266(00)90004-4 by Nat Prov Indonesia, Wiley Online Library on [24/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
conditions were met during relaxed stance: 3" or more of calcaneal
valgus, medial bulge of the talonavicular joint, and navicular tuber-
osity below Feiss's line [24] (the line passing through the apex of the
medial malleolus and the plantar portion of the first metatarsal head:
Fig. 1).The foot was neutral if varus/valgus position of the calcaneous
fell between 0" and 3", and the navicular tuberosity was aligned with
Feiss's line. The foot was supinated if calcaneal varus was greater than
3", there was no medial bulge of the talonavicular joint, and the na-
vicular tuberosity was aligned above Feiss's line. Next, tibia1 and
calcaneal varushalgus position measurements were made with the
subject standing with feet placed shoulder-width apart [32]. Tibia1
varuslvalgus position was evaluated by aligning one arm of the goni-
ometer with an axis that bisected the lower leg in the midsagittal plane,
aligning the other arm parallel with the floor, and measuring the in-
cluded angle. Calcaneal varus/valgus position was evaluated by
aligning one arm of the goniometer with the line that bisected the
calcaneus in the midsagittal plane, aligning the other arm with the line
that bisected the tibia in the midsagittal plane, and measuring the in-
cluded angle.
Irokinetic ankle, strength was evaluated using the Cybex 6000 dy-
namometer (Lumex, Ronkonkoma, NY [4,3,14,20]. Athletes were
prone for the plantar dorsi-flexion test, and supine with the knee flexed
at 90" for the inversion-eversion test. Plantar dorsi-flexion and
inversion-eversion peak torque values, and corresponding strength
ratios, were determined using both concentric and eccentric protocols
for both ankles. Each test was performed at a speed of 30"/s using a
comfortable range ofjoint motion. Five trials were performed for each
test and all were used in the data analysis.
Anterior-posterior center of' graoity (A-P COG) stcay angle was
evaluated with the NeuroTest system (NeuroCom International,
Clackamas OR, USA). This measure compensates for differences in
subjects' heights because the angular limits of stability are similar for
all adults, irrespective of height. The maximum A-P COG sway angle
was measured while the subjects had their eyes closed (eliminating the
contribution of the visual system to maintaining balance) and with the
force platform in the fixed and sway referencing modes. In the fixed
mode, subjects stood on the force plate and A-P COG sway angle was
controlled by the subject's somatosensory (i.e. the mechanoreceptors
about the ankle and the other joints of the lower extremity) and ves-
tibular systems. In the sway reference mode, subjects stood on the
same force plate with a servo-device engaged that matched the angular
motion of the force plate to the estimated angular sway of the subjects
center of gravity. This reduced angular displacement of the ankle joint,
thereby delivering minimal proprioceptive information from the ankle
so that the subject relied primarily on the vestibular system to maintain
postural equilibrium. The maximum A P COG sway angle was mea-
sured for the following conditions: ( I ) fixed platform with dominant
C
Fig. 1. The line passing through the apex of the medial malleolus (A)
and the plantar portion of the first metatarsal head (C) was used to
identify Feiss's line (24). The location of the navicular tuberosity (B),
relative to this line, in combination with appearance of the talona-
vicular joint and calcaneal valgns measurements were used to establish
anatomic foot type.

B.D. Beynnon et ul. I Journal of
leg stance: (2) fixed platform with nondominant leg stance; ( 3 ) sway-
referenced platform with the subject using twolegged stance; (4) sway-
referenced platform with dominant leg stance: and, ( 5 )sway-referenced
platform with nondominant leg stance. Three trials of each test con-
dition were performed, each of which required 20 s to perfom. The
maximum A-P COG sway angles were summed across the three trials
and a mean value calculated for each of the five test conditions.
hfusck reaction time, the time lag between joint perturbation and
muscle activation, was measured with the Neuro Test system and
electromyographic (EMG) signals from surface electrodes (Motion
Control, Salt Lake City, Utah) placed over the muscles of interest for
dorsi-flexion and inversion perturbation of the foot.
For dorsi-flexion perturbation, electrodes were placed on the medial
head of the gastrocnemius and on the tibialis anterior. This approach
minimized crossover between electrodes by placing them as far apart as
practical [21]. The subject was positioned so perturbation produced
dorsi-flexion rotation of both feet. The stimulus followed as a 4' dorsi-
flexion rotation at a velocity of 5O"ls, eliciting the three characteristic
EMG muscle reaction components: the short-loop gastrocnemius,
medium-loop gastrocnemius, and long-loop tibialis anterior responses.
EMG data were sampled with surface electrodes (one common and two
active) that were fixed in a plastic housing and uniaxially aligned such
that the common electrode (located in the center between the active
electrodes) was spaced 0.5 cm from the active electrodes. Analog ac-
tivity from the active electrode pairs was amplified by a differential
amplifier and sampled at a frequency of 1000 Hz from 100 ms before the
perturbation began until 400 ms after. Each subject underwent per-
turbations until 10 EMG trials were observed for each muscle on each
leg. The EMG signal was rectified and then each trial was analyzed to
identify the onset of the short-loop gastrocnemius, medium-loop gas-
trocnemius, and long-loop tibialis anterior responses using the vali-
dated technique described by Lawson et al. [Zl],and then combined to
produce average values. Inversion perturbation of the ankle required
that EMG electrodes be attached to the peroneal longus, peroneal
brevis, and tibialis anterior. The subject was then positioned such that
the perturbation produced a 4' inversion rotation of the foot at a ve-
locity of 5O"/s, and the same data acqu on parameters described for
dorsi-flexion perturbation were used. This stimulus produced the three
characteristic EMG components termed the medium-loop peroneal
longus, medium-loop peroneal brevis, and long-loop tibialis anterior
reactions. Each subject underwent the inversion perturbations until ten
EMG tracings were obtained for each muscle on each leg. These data
were combined to produce average values.
All athletes played outdoors on the same natural turf and were
exposed to the same field at the same time. Thus, on any given day all
study participants were exposed to the same conditions. Athletes were
monitored throughout the athletic season, documenting their exposure
to sports, and those whom sustained an ankle injury were evaluated
immediately by an orthopaedic surgeon (PAR). If a sprain was diag-
nosed, then it was graded as either a I, 11, or I11 [7]. A grade I injury
was defined as no loss of function, no loss of ligamentous stability
(negative anterior drawer and talar tilt tests), little or no hemorrhage,
and point tenderness. Grade 11 injuries demonstrated some loss of
function, decreased motion, a positive anterior drawer and negative
talar tilt test (the ankle mortise did not open with applied inversion
stress), hemorrhage, swelling and point tenderness. Grade I11 injuries
had nearly total loss of function, a positive anterior drawer and talar
tilt test (the ankle mortise opened with applied inversion stress), diffuse
swelling and hemorrhage, and extreme point tenderness.
Pre-season potential risk factor measurements from men and
women were compared using t-tests and Chi-square tests. Since there
were gender differences for many of the outcomes, all other analyses
were performed separately for men and women. Cox regression was
used to test the effect of each variable on relative risk, taking into
account differences in the lengths of time the athletes were at risk.
This was accomplished by computing time at risk as the total number
of games and practices in which each subject participated until injury
or, if uninjured, the end of the sports season. Since the Cox model
assumes that the effect of each variable is proportional to an un-
derlying hazard function, subjects were stratified by sport to take into
account differences in the hazard functions for different sports. For
all analyses, data from only one leg was used the injured leg for
~
subjects with injuries and a randomly selected leg for those who were
uninjured.
1554527x, 2001, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1016/S0736-0266(00)90004-4 by Nat Prov Indonesia, Wiley Online Library on [24/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Orthopaedic Reseurch 19 (2001)213-220 215
Results
The men had 4249 days of exposure to sports, while
the women had 5813 days. The number of ankle injuries
per 1000 person-days of exposure was 1.6 for men and
2.2 for women. There were 13 injuries among the 68
women (19.1%), and seven injuries among the 50 men
( 14%), but these proportions were not significantly dif-
ferent. Ten of the 13 injuries in women occurred in the
right leg, which was significantly greater than would be
expected if both legs were at equal risk ( P = 0.046). This
was not the case for men, where three of the seven in-
juries occurred in the right limb. The majority of injuries
occurred during practice rather than games. The women
sustained 11 injuries during practice and two during
games, while for the men six injuries occurred during
practice and one during a game. A majority of the in-
juries occurred during sunny weather when the turf was
dry. The women suffered nine injuries when the weather
was sunny and the playing field was dry, three injuries
when the field was wet, and one injury when it was cold.
The men sustained five injuries when the weather was
sunny and the field was dry, one injury when the field
was wet, and one injury when it was cold.
Men and women differed substantially in terms of
several pre-season risk factors; thus, data are presented
separately for each gender (Tables 1-6). The mean
height and weight of the men were 70.4 in. and 169 lbs,
respectively, and for the women, 65.1 in. and 133 Ibs. As
expected, the women were smaller and weighed less than
the men ( P < 0.05, Table 1). Women had larger dorsi-
flexion motion of the ankle, larger calcaneal inversion
range of motion, and larger calcaneal eversion range of
motion, compared to the men ( P < 0.05, Table 3). The
isokinetic strength measures were significantly lower for
the women than the men ( P < 0.05, Table 4). The A-P
COG sway angle was significantly larger among the
women for the single-leg stance nonsway referenced
evaluation ( P < 0.05, Table 5). Women had a faster
short-loop reaction time of the gastrocnemius muscle in
response to a dorsi-flexion perturbation of the foot, and
a faster reaction time of the peroneal brevis in response
to an inversion perturbation ( P < 0.05, Table 6). There
were no differences between men and women with re-
gard to: limb dominance, generalized joint laxity, ankle
joint laxity, anatomic foot type, the medium-ioop reac-
tion time of the gastrocnemius muscle and the long-loop
reaction time of anterior tibialis muscles in response to
dorsi-flexion perturbation of the foot, and the peroneal
longus and anterior tibialis reaction times in response to
inversion perturbation of the foot.
The P-values in Tables 1-6 correspond to the global
chi-square statistic from the individual Cox regressions
performed for each variable. This tests the significance
of each variable on the hazard function, controlling for
time at risk and sport. Women who played soccer had a
 1554527x, 2001, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1016/S0736-0266(00)90004-4 by Nat Prov Indonesia, Wiley Online Library on [24/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
216 B.D. Bejnnon et al. / Journal of Orthopaedic Research 19 (2001) 213-220

Table 1
Demographic data for injured and uninjured subjects"
Variable Women Men
Uninjured Injured P-value (from Uninjured Injured P-value (from
Cox regression) Cox regression)
Subjects 55 13 43 7 ~

Height (in.)b 65.1 65.8 0.5 70.4 72.3 0.6

Weight, Ibs (S.D.)h 133 (14) 128 (14) 0.24 169 (15.8) 172 (9.1) 0.38
Limb dominance
Right N ("0) 52 (96) 10 (92) 0.41 39 (91) 4 (57) 0.1 1
Left N (";I) 2 (4) 3 (8) - 4 (9) 3 (43) ~

Sport
Soccer N ("4) 12 (22) 7 (54) 0.023 23 (54) 3 (43) 0.60
Field hockey N ('Iil) 28 (51) 3 (23) - - ~

Lacrosse N ("A,) 15 (27) 3 (23) 20 (46) 4(57) ~

"The data are presented for women and men separately. The P-values are presented for the Cox-regression analysis.
'Women are significantly different in comparison to men for these specific factors P 6 0.05.

Table 2
Joint laxity data for uninjured and injured subjectsd
Women Men
Uninjured N Injured N f-value (from Uninjured N Injured N P-value (from
(UYI1) ('%I ) Cox regression) ('%I) (Yo) Cox regression)
General joint laxity
- 46 (85) 10 (77) 0.34 39 (93) 7 (100) ~

+ 8 (15) 3 (23) ~

3 (7) 0 (0) ~

Anterior drawer
0 33 (61) 12 (92) 0.1 32 (78) 5 (71) 0.81
1 21 (39) 1 (8) 9 (22) 2 (29) ~

Talar tilt
- 49 (91) I 1 (85) 0.29 40 (98) 5 (71) 0.002
+ 5 (9) 2 (15) ~ 1 (2) 2 (29) -

Generalized joint laxity was measured using the modified Beighton method, and subjects were classified as either having generalized joint laxity (+)
or not (-). In addition ankle laxity was measured with the anterior drawer and talar tilt. Data are presented for men and woment seperately. The
P-values are presented for the Cox-regression analysis.

Table 3
Measurements of alignment of the foot and ankle for injured and uninjured subjects expressed in degrees"
Women Men
Uninjured Injured P-value (from Uninjured Injured P -value (from
Cox regression) Cox regression)
Anatomic foot type
Pronated N ((%I) 24 (44) 4 (31) 0.46 15 (36) 2 (29) 0.1 1
Neutral N ("A) 25 (46) 7 (54) 20 (49) 2 (29)
Supinated N (Yn) 5 (9) 2 (15) 6 (15) 3 (42)
Calcaneal inversion 17.8 (4.6) 16.8 (5.4) 0.72 15.3 (3.6) 18.0 (5.9) 0.14
unweighted: mean (S.D.)b
Calcaneal eversion 5.2 (1.8) 6.1 (2.6) 0.038 4.3 (2.4) 5.4 (2.3) 0.56
unweighted: mean (S.D.)b
Rearfool vadvalgus 4.5 (1.5) 4.2 (1.6) 0.93 4.3 (1.8) 4.9 (0.9) 0.52
unweighted: mean (S.D.)
FF/RF relationship 3.0 (3.3) 2.7 (3.0) 0.70 3.2 (3.6) 2.4 (3.2) 0.62
unweighted: mean (S.D.)
Dorsi-flexion KE: mean (S.D.)b 4.9 (1.9) 5.3 (1.8) 0.55 4.0 (2.7) 4.1 (0.9) 0.41
Dorsi-flexion KF: mean (S.D.) 12.3 (3.3) 12.8 (3.5) 0.52 11.3 (3.7) 12.7 (3.9) 0.18
Calc varus/valgus during 7.2 (3.6) 6.1 (3.5) 0.33 6.8 (4.2) 4.3 (5.3) 0.15
weightbearing: mean (S.D.)
Tibia1 varus/valgus during 5.4 (2.2) 6.6 (2.6) 0.028 4.8 (2.5) 4.7 (2.0) 0.94
weightbearing: mean (S.D.)
a These data are presented for women and men separately. The P-values are presented for the Cox-regression analysis
'Women are significantly different in comparison to men for these specific factors ( P GO.05).
 1554527x, 2001, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1016/S0736-0266(00)90004-4 by Nat Prov Indonesia, Wiley Online Library on [24/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
B. D. Bepnon et ul. I Journal of Orthopaedic Rcseurch I9 (2001 1 213 ,720 217

Table 4
Isokinetic strength values for concentric (Con) and eccentric (Ecc) contractions of the leg muscles during plantar flexion (PF), dorsi-flexion (DF),
inversion (INV), eversion (EV) modes for uninjured and injured subjects"
Women Men
Uninjured l' Injured S P-value (from Uninjured ~Y Injured A' P value (from
(S.D.) (S.D.) Cox regression) (S.D.) (S.D.) Cox regression)
Con strength PF' 53.1 (11.3) 52.3 (12.3) 0.96 71.8 (13.7) 77.3 (13.0) 0.34
Con strength DF' 18.2 (2.6) 17.0 (2.4) 0.10 26.8 (3.8) 28.1 (4.2) 0.42
Con strength INVb 14.2 (4.2) 15.2 (5.9) 0.75 20.0 (5.3) 21.1 (5.1) 0.92
Con strength EV' 13.3 (2.7) 12.6 (3.4) 0.47 19.5 (5.3) 22.3 (5.6) 0.43
Ecc strength PFh -68.4 (18.1) -68.5 (16.3) I .OO -89.7 (20.9) -97.0 (32.5) 0.3 1
Ecc strength DF' -26.8 (4.1) -26.1 (4.0) 0.34 -39.4 (6.1) -38.9 (6.1) 0.62
Ecc strength lNVh -16.3 (4.3) -17.3 (3.7) 0.84 -23.8 (6.2) -25.7 (6.2) 0.38
Ecc strength EV' -16.2 (3.8) -15.5 (3.3) 0.31 -24.6 (5.8) -24.0 (5.4) 0.85
Con DFlCon PF 0.36 (0.09) 0.34 (0.08) 0.25 0.39 (0.09) 0.37 (0.09) 0.74
Con EVlCon INV 0.99 (0.30) 0.88 (0.24) 0.33 1.02 (0.30) 1.07 (0.26) 0.73
Ecc DFlEcc PFh 0.42 (0.I 1 ) 0.40 (0.09) 0.49 0.47 (0.15) 0.44 (0.17) 0.75
Ecc EVlEcc INV 1.03 (0.24) 0.92 (0.25) 0.17 1.08 (0.30) 0.96 (0.19) 0.21
"These data are presented for women and men separately. The P-values are presented for the Cox-regression analysis.
'Women are significantly different in comparison to men for these specific factors ( P < 0.05).

Table 5
Anterior-posterior center of gravity (A-P center of gravity) sway angle for uninjured and injured subjects expressed in degrees. A-P COG sway angle
expressed in terms of degrees of sway is presented for women and men separately"
Women Men
Uninjured X Injured A' P-value (from Uninjured X Injured A' P-value (from
(S.D.) (S.D.) Cox regression) (S.D.) (S.D.) Cox regression)
Single leg stance nonsway 3.4 (0.33) 3.4 (0.32) 0.65 3.0 (0.24) 3.6 (0.55) 0.21
referenced
Single leg stance sway 4.8 (0.79) 5.0 (0.71) 0.25 4.8 (0.77) 4.9 (1.1) 0.59
referenced
Two leg stance sway referenced 3.9 (0.4) 4.0 (0.64) 0.90 3.8 (0.39) 3.4 (0.36) 0.27
a The P-values are presented for the Cox-regression analysis.
Women are significantly different in comparison to men for these specific factors ( P 6 0.05).

Table 6
Muscle reaction times, expressed as ms, in response to dorsi-flexion and inversion perturbations of the foot for both uninjured and injured subjects>
Women Men
Uninjured .Y Injured X P-value (from Uninjured X Injured X P-value (from
(S.D.) (S.D.) Cox regression) (S.D.) (S.D.) Cox regression)
Short-loop GSTb 32.4 (2.5) 33.2 (3.1) 0.31 34.4 (3.0) 34.3 (3.5) 0.68
Med-loop GST 82.4 (12.0) 77.8 (13.7) 0.076 83.9 (1 1.6) 83.1 (15.1) 0.63
Long-loop AT 119 (15.4) 127 (17.7) 0.1 1 131 (17.2) 135 (13.4) 0.41
Inversion PBb 83.5 (12.5) 80.5 (8.8) 0.44 87.6 (11.6) 89.2 (4.8) 0.69
Inversion PL 83.0 (11.6) 83.4 (9.2) 0.56 86.2 (12.5) 86.5 (4.6) 0.94
Inversion AT 117 (23.5) 125 (29.0) 0.44 120 (27.5) 118 (26.4) 0.51
a Data are presented for dorsi-flexion perturbation (short-loop gastrocnemius, medium-loop gastrocnemius, and long-loop anterior tibialis responses)

and for inversion perturbation (inversion PB (peroneal brevis), inversion PL (peroneal longus) and inversion AT (anterior tibialis)). The P -values are
presented for the Cox-regression analysis.
Women are significantly different in comparison to men for these specific factors ( P < 0.05).

higher incidence of ankle injuries than those who played
field hockey or lacrosse ( P = 0.02, Table 1). Ankle in-
juries were more common among women with increased
tibia1 varum and calcaneal eversion range of motion
( P = 0.03 and 0.02, respectively). In spite of the fact that
the reaction times of the muscles in response to inversion
perturbation of the foot were not predictive of injury, we
observed an interesting trend of the reaction times in
response to dorsi-flexion perturbation of the foot. On
average, the medium-loop reaction of the gastrocnemius
muscle was 5 ms faster (P= 0.07) while the long-loop
reaction of the anterior tibialis was 8 ms longer ( P = 0.1)
among female athletes who sustained injuries compared
to those who did not (Figs. 2 and 3). Generalized and

118 B. D. Beynnon el al. I Journal of Orthopaedic Research 19 (2001) ,713-220
T" In T T'"
I I I I I
i
I I
(4
Fig. 2. Pre-season muscle reaction data from a female subject that was
not injured. Shown is the reaction of the tibialis anterior muscles in
response to a 4" rotation of the foot during approximately 125 ms
(Fig. I(a)).The characteristic EMG reactions of the tibialis anterior for
the left leg (L. Tibia, Fig. I(b))and the right leg (R.Tibia, Fig. 1(c))are
presented. The long-loop response (LLI)occurred at 99 and 96 ms for
the left and right muscles, respectively
ankle joint laxities, anatomic foot type, strength, and A-
P COG sway angle were similar among injured and
noninjured women.
Among the men, there was no effect of type of sport
on incidence of ankle injury (Table 1). Men whose talar
tilt exams demonstrated a rotation in excess of 20"
sustained a greater proportion of ankle ligament injuries
than those whose exams showed a rotation less than 20"
( P = 0.003, Table 2). Strength, anatomic foot type, A-P
COG sway angle, and the reaction time of the muscles in
response to dorsi-flexion and inversion perturbations of
the foot were no different among the injured than among
the uninjured men.
Discussion
Our previous work concerned potential ankle injury
risk factors in a similar group of intercollegiate athletes
-who competed in the same sports [4] and showed that
individuals with an increased calcaneal eversion rotation
and muscle strength imbalances had a higher prevalence
of inversion ankle sprains. The present study confirms
our earlier findings, at least in part, since we found that,
1554527x, 2001, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1016/S0736-0266(00)90004-4 by Nat Prov Indonesia, Wiley Online Library on [24/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
2'
In'Gm Y
t I I ,
I
I
I i
(4
Fig. 3. Pre-season muscle reaction data from a female subject that
subsequently suffered an ankle sprain, presented for the purpose of
comparison with the data from the uninjured subject (Fig. l(a)-(c)).
Reaction of the anterior tibialis muscle in response to the 4" rotation of
the foot (Fig. ?(a)), the same perturbation used for the uninjured
subject (Fig. l(a)),is presented along with the EMG reactions for the
left leg (L. Tibia, Fig. ?(b)) and right leg (R. Tibia, Fig. 2(c)).The long-
loop response (LLI) appeared to be delayed for this injured subject
compared to uninjured subjects, occurring at 147 and 143 ms for the
left and right muscles correspondingly.
among women, ankle injuries were related to increased
calcaneal eversion range of motion. The present inves-
tigation also showed that women with increased tibia1
varum and men with increased talar tilt were more likely
to sustain ankle injuries. This finding was in contrast to
our earlier study, as was the fact that, in the present
study, we also found no relation between muscle
strength or muscle strength imbalances and subsequent
injury.
The test-retest reliability of the ankle injury risk
factor measurements we used has been shown to be
highly repeatable [5,33], and therefore differences be-
tween injured and uninjured subjects for the anatomic
alignment risk factors were not likely to have resulted
from the examiner's learning effect or measurement ar-
tifact. Differences between the present and our previous
study may derive from differences in methodology.
Unlike the previous study, this time we evaluated
women and men separately. If the effect of a risk factor
is gender dependent, it will be obscured in analyses using
combined data from men and women. Even risk factors

B.D. Beynnon et ul. I Journal of Orthopaedic Reseurch 19 (2001 I 213 - 2 0
having similar effects in men and women may not be
detected in analysis of combined data if high values for
women correspond to low values for men. Conversely,
variables whose values differ greatly between genders
may falsely appear to have an effect on risk if women are
inherently at higher risk. Also, in the current study we
evaluated exposure data and performed data analysis
using the Cox regression model to take into account
both time at risk and differences in risk associated with
different sports. Our earlier study did not document
exposure data and used the Student’s f test to analyze
the data without adjustment for sport.
Dahle et al. [12] and Barrett et al. 131 reported no
correlation between anatomic foot type (pronated,
neutral, or supinated) and the incidence of ankle sprains.
Our investigation supports these findings; however, 70%
of the ankle injuries in our study occurred in athletes
with either a neutral or supinated foot type.
Barrett et al. [2] reported that provocative testing of
ankle ligament stability with the anterior drawer and
talar tilt exams did not predict ankle injury. In our
earlier work, on this same subject [4], we found that
provocative testing with the anterior drawer test showed
a trend toward prediction, while the talar tilt exam did
not. In our present investigation, the same trend was
observed among women ( P = 0.1); however, we found
an association between talar tilt and injury incidence
among men.
Jackson et al. [I61 and our group [4]reported that
generalized joint laxity is not correlated with ankle lig-
ament injury. The present investigation confirms these
findings for men and women, considered separately.
There is also conflicting evidence regarding the effect
of limb dominance on the risk of ankle injury. Ekstrand
and Gillquist [ 131 reported an increased risk of injury for
the dominant ankle, but Surve et al. [34] found no dif-
ference in the incidence of ankle injury between non-
dominant and dominant ankles. The present study
agrees with the findings of Surve et al. [34] for both men
and women.
Our earlier work also investigated the relationship
between lower extremity strength and the incidence of
ankle injury [4]. We found that ankle sprains were as-
sociated with higher ratios between ankle eversion and
inversion peak torques, higher peak torque values dur-
ing plantar-flexion, and a lower ratio between dorsi-
flexion and plantarflexion peak torque values [4]. In the
current study, however, we found no difference in peak
torque values among athletes who subsequently sus-
tained injuries and those who did not for all test modes.
Likewise, neither the ratio between ankle eversion and
inversion peak torque values, nor the ratio between
dorsi-flexion and plantar flexion peak torque values was
related to subsequent injury.
The muscle reaction times were not different for in-
jured than for uninjured men. In contrast, we observed
1554527x, 2001, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1016/S0736-0266(00)90004-4 by Nat Prov Indonesia, Wiley Online Library on [24/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
219
an interesting trend for the women: athletes who sub-
sequently sustained an ankle injury had a gastrocnemius
muscle that required less time to react while the anterior
tibialis muscle required more time to react in response to
a dorsi-flexion perturbation of the foot compared to the
athletes that were not injured. This suggests that a
neuromuscular deficit may have existed in these women
athletes, and that the protective effect of the leg muscles
on joint stability may have been compromised. The
medium-loop gastrocnemius muscle reaction controls
local muscle properties while the long-loop anterior
tibialis reaction provides the first stabilization to the
ankle joint in response to a perturbation. The delay in
the anterior tibialis reaction may reflect some deficit of
the musculoskeletal system that compromises the pro-
tective effect of the leg muscles on ankle joint stability,
thereby predisposing these women athletes to ankle in-
jury.
Maintaining proper contact conditions between the
foot and floor during stance or gait may require in-
creased tibial varum rotation to be coupled with in-
creased eversion rotation of the calcaneous. Therefore,
our finding that amongst women, increased tibial varum
and calcaneal eversion range of motion were both as-
sociated with increased risk of suffering ankle trauma
introduced the prospect that these measurements were
related. Post-hoc correlation analysis did not reveal a
significant correlation between these two anatomic
measurements; however, there was a moderate correla-
tion between calcaneal eversion rotation (evaluated with
the subject unweighted) and calaneal varushalgus ori-
entation, evaluated with subject weightbearing
(Y = 0.79; P = 0.032). This finding suggests, at least in
part, that orientation of the hindfoot is an important
anatomic parameter to consider when evaluating risk
factors for inversion ankle trauma.
In summary, this study has demonstrated that the
risk factors that predispose an athlete to ankle ligament
injury were different between men and women. Women
with increased tibial varum rotation and a compensa-
tory increase of calcaneal eversion range-of-motion
sustained proportionately more ankle injuries. Men with
an increased talar tilt sustained more injuries. For
women, playing soccer (as opposed to field hockey or
lacrosse) significantly increased their risk of ankle inju-
ry, whereas for men, type of sport (soccer versus la-
crosse) had no effect on injury incidence. Strength and
postural sway (as characterized by A-P COG sway an-
gle) were not risk factors for both men and women.
Currently, we are expanding on the efforts described in
this investigation by including other sports. different
playing conditions, and younger athletes to broaden the
scope and applicability of our findings. Once we have
established the risk factors associated with male and
females athletes for different playing conditions, our
next step will be to perform an intervention study that is

220 B.D. Beynnon r t a/. I Journal of Orthopaedic Research 19 (2001) 213-220
designed to determine how we can reduce the high in-
cidence of ankle trauma.
Acknowledgements
This work was funded in part by a grant from the
National Collegiate Athletic Association (grant
#60120100) and the National Institutes of Health (R01-
AR45346).
References
Barker HB, Beynnon BD, Renstrom PA. Ankle injury risk factors
in sports. Sports Med 1997:23:69-74.
Barrett JR, Tanji JL, Drake C, et al. High versus lowtop shoes for
the prevention of ankle sprains in basketball players: a prospective
randomized study. Amer J Sports Med 1993;21:582-5.
Baumhauer JF. A comparison study of ankle inversion and
eversion strength in healthy and inversion ankle sprained individ-
uals as assessed by Cybex I1+ dynamometer. Thesis, Middlebury
College, Middlebury, Vermont, 1985.
Baumhauer JF, Alosa DM, Renstrom PA, Trevino S, Beynnon
BD. A prospective study of ankle injury risk Factors. Amer J
Sports Med 1995;33:564-70.
Baumhauer JF, Alosa DM, Renstrom PA, Trevino S, Beynnon
BD. Test-retest reliability of ankle injury risk factors. Amer J
Sports Med 1995;33:5714.
Beighton P, Solomon L, Soskolne CL. Articular mobility in an
African population. Ann Rheum Dis 1973:3(2):413-8.
Bergfeld J, Halpern B. Sports Medicine: Functional Management
of Ankle Injuries. Videotape Library of the American Academy of
Family Physicians, 1991, V.T. No. 894.
Black HM, Brand RL, Eichelberger MR. An improved technique
for the evaluation of ligamentous injury in severe ankle sprains.
Amer J Sports Med 1978;6:276-82.
Brand RL, Black HM, Cox JS. The natural history of inade-
quately treated ankle sprains. Amer J Sports Med 1977;5:248249.
Chrisman OD, Snook GA. Reconstruction of lateral ligament
tears of the ankle. J Bone Joint Surg 1969;51A90412.
Cox JS, Hewes TF. Normal talar tilt angle. Clin Orthop
1979;140:3741.
Dahle LK, Mueller M, Delitto A. et al. Visual assessment of foot
type and relationship of foot type to lower extremity injury.
J Orthop Sports Phys Ther 1991:14:70-4.
[13] Ekstrand J, Gillquist J. Soccer injuries and their mechanisms: a
prospective study. Med Sci Sports Exerc 19833 5:267-70.
[I41 Elveru RA, Rothstein JM, Lamb RL, et al. Methods for taking
subtalar joint measurements: a clinical report. Phys Ther
1988:68:678-82.
[I51 Jackson JP, Hutson MA. Castbrace treatment of ankle sprains.
Injury 1986;17:251 5.
[16] Jackson DW, Jarrett H, Bailey D, et al. Injury predictlon in the
young athlete: a preliminary report. Amer J Sports Med 1978;6:6-
11.
1554527x, 2001, 2, Downloaded from https://onlinelibrary.wiley.com/doi/10.1016/S0736-0266(00)90004-4 by Nat Prov Indonesia, Wiley Online Library on [24/03/2023]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
[I71 James SL, Bates B, Osternig LR. Injuries to runners. Amer J
Sports Med 1978;6:40-50.
[I81 Johnson EE, Markolf KL. The contribution of the anterior
talofibular ligament to ankle laxity. J Bone Joint Surg 1983:
65A:81-8.
[I91 Karlsson J, Bergsten T, Lansinger 0, Peterson L. Reconstruction
of the lateral ligaments of the ankle for chronic lateral instability.
J Bone Joint Surg 1988;70A:581-8.
Karnofel H, Wilkinson K, Lentell G. Reliability of isokinetic
muscle testing at the ankle. J Orthop Sports Phys Ther
198931:150-4.
Lawson GD, Shepard NT, Oviatt DL, Wang Y. Electromyo-
graphic responses of lower leg muscles to upward toe tilts as a
funciton of age. J Vistibular Res 1994;4:203-14.
Lynch SA, Eklund U, Gottlieb DJ, Renstrom PA, Beynnon
BD. Electromyographic latency changes in the ankle muscula-
ture during inversion moments. Amer J Sports Med
1996;24:362-9.
Lysens R, Steverlynck A, van den Auweele Y, Lefevre J, Renson
L. Claessens A, Ostyn M. The predictability of sports injuries.
Sports Med 1984;l:fLlO.
Magee DJ. Orthopaedic physical assessment. 2nd ed. Lower leg,
ankle and foot. London: Saunders, 1987 (Chapter 12).
Makhani JS. Lacerations of the lateral ligaments of the ankle. An
experimental appraisal. J Int Coll Surg 1962;38:45&66.
Milgrom C , Shlamkovitch N, Finestone A, Eldad A, Laor A,
Danon YL, Lavie 0, Wosk J, Simkin A. Risk factors for the
lateral ankle sprain: a prospective study among military recruits.
Foot Ankle 1991;1226-30.
Moretz JA. Flexibility as a predictor of knee injuries in college
football players. Phys Sportsmed 1982: 10:93-7.
Nicholas JA. Injuries to knee ligaments; relationship to looseness
and tightness in football players. J Amer Med Assoc 1970;
2122236-9.
Rasmussen 0. Stability of the ankle joint. Analysis of the function
and traumatology of the ankle ligaments. Acta Orthop Scand
(Suppl.) 1985;211:1 -75.
Renstrom P, Theis M. Biomechanics and function of ankle
ligaments: experimental results and clinical application. J German
Orthop Trauma 1993;7:29-35.
Sitler M, Ryan J, Wheeler B, et al. The efficacy of a semirigid
ankle stabilizer to reduce acute ankle injuries in basketball. Am J
Sports Med 1994;22:45461.
Smith-Oricchio K, Harris A. Interrater reliability of subtalar
neutral, calcaneal inversion and eversion. J Orthop Sports Phys
Ther 1990;12:10-5.
Soboroff SH, Pappius EM, Komaroff AL. Benefits, risks and costs
of alternative approaches to the evaluation and treatment of
severe ankle smains. Clin Orthor, 1984;183:160-8.
[34] Surve I, Schwellnus MP, Noakes T, et al. A fivefold reduction
in the incidence of recurrent ankle sprains in soccer players
using the sportstirrup orthosis. Amer J Sports Med
199422601-6.
[35] Tohyama H, Beynnon BD. Renstrom PA, Theia MJ, Fleming BC,
Pope MH. Bioniechanical analysis of the ankle anterior drawer
test for anterior talofibular ligament injuries. J Orthop Res
1995:13:609-14.