Physical Therapy in Sport 11 (2010) 81e85

Contents lists available at ScienceDirect

Physical Therapy in Sport

journal homepage: www.elsevier.com/ptsp

Original research

Correlation of three different knee joint position sense measures

Dayanand Kiran a, *, Mary Carlson a, Daniel Medrano b, Darla R. Smith b
a
Physical Therapy Program, University of Texas at El Paso, 1101, N Campbell Street, El Paso 79902, USA
b
Department of Kinesiology, University of Texas at El Paso, El Paso 79902, USA

a r t i c l e i n f o a b s t r a c t

Article history: Objective: The purpose of this study was to investigate correlation during concurrent measurement
Received 17 March 2010 among three knee joint position sense (JPS) measures in sitting position and between two measures in
Received in revised form standing position.
27 May 2010
Methods: Isokinetic dynamometer, electrogoniometer, and two dimensional (2D) video analysis were
Accepted 7 June 2010
used for measuring knee JPS. The JPS was measured both in sitting and standing positions. All three
measures were employed concurrently to measure knee JPS in sitting position; however, only the
Keywords:
electrogoniometer and 2D video analysis were concurrently used in the standing position. The knee JPS
Joint position sense
Electrogoniometer
was recorded in sitting position at 15
, 30
, and 45
and in standing at high, mid and low knee exion
Isokinetic dynamometer positions.
Two dimensional video analysis Results: The results of the study suggest excellent correlation (0.94e0.98) between the electrogoniometer
Correlation and 2D video analysis measures in standing position. In sitting position, good to excellent correlation
(0.63e0.92) was found between the isokinetic dynamometer and electrogoniometer; however, fair to
good correlation was found between 2D video analysis and either of the two measures (electro-
goniometer [0.52e0.57] and isokinetic dynamometer [0.41e0.63].
Conclusion: Either 2D video or an electrogoniometer may be used to measure JPS in standing position;
however, in sitting position 2D video should not be used if the camera is required to be placed at 10

from the plane of motion.
2010 Elsevier Ltd. All rights reserved.

1. Introduction
Proprioception is the umbrella term for kinesthesia and joint
position sense (JPS). JPS refers to the awareness of joint position in
space and is mediated through various receptors called mechano-
receptors (Grob, Kuster, Higgins, Lloyd, & Yata, 2002). These
receptors are located in the joint capsule, ligaments, menisci,
musculotendinous unit, and skin (Kavounoudias, Roll, & Roll, 2001;
Lephart, Pincivero, & Rozzi, 1998).
Poor proprioception is suggested as a risk factor for the devel-
opment of functional inability in patients with knee osteoarthritis
(Sharma, Cahue, Song, Hayes, Pai, & Dunlop, 2003). The deteriora-
tion of proprioception results in increased postural sway, decreased
balance, increased risk of falls and changes in gait patterns (Bergin,
Bronstein, Murray, Sancovic, & Zeppenfeld, 1995; Manchester,
Woollacott, Zederbauer-Hylton, & Marin, 1989; Tinetti &
Speechley, 1989). The changes in proprioception not only affect
* Corresponding author. Tel.: 1 915 747 7218; fax: 1 915 747 8211.
E-mail addresses: dkiran@miners.utep.edu (D. Kiran), mcarlson@utep.edu (M.
Carlson), danielm@miners.utep.edu (D. Medrano), darsmith@utep.edu (D.R. Smith).
the function of older adults but also have impact on the younger
population, especially with anterior cruciate ligament (ACL)
injuries. Proprioception is important in the prevention of injuries as
reduced proprioception is one of the factors contributing to injury
in the knee, particularly the ACL. Although the causes of ACL injury
are multi-factorial, poor proprioception is one of the key causative
factors (Beynnon & Johnson, 1996; Grifn et al., 2000; Taimela,
Kujala, & Osterman, 1990). Additionally, the restoration of a fully
functional knee joint after ACL injury depends on regaining
proprioception as an important component (Aune, Holm, Risberg,
Jensen, & Steen, 2001; Friden, Roberts, Ageberg, Walden, &
Zatterstrom, 2001). Therefore, proprioception appears not only
important for the prevention of ACL injuries, but also for regaining
full function after ACL reconstruction.
ACL injuries can be treated with reconstruction surgery where
the patellar tendon or the hamstring tendon is used as a graft to
replace the torn ACL ligament. Although reconstruction is
successful in regaining joint stability, the recovery of propriocep-
tive function remains debatable (Henriksson, Ledin, & Good,
2001). MacDonald, Hedden, Pacin, and Sutherland (1996) repor-
ted no signicant improvement in proprioceptive decits in
patients 31 months after ACL reconstruction by measuring

1466-853X/$ e see front matter 2010 Elsevier Ltd. All rights reserved.
doi:10.1016/j.ptsp.2010.06.002
82 D. Kiran et al. / Physical Therapy in Sport 11 (2010) 81e85
kinesthesia. However, Reider et al. (2003) reported a signicantly
improved level of proprioception by measuring JPS in an ACL
reconstructed knee after six months of rehabilitation when
compared with the contralateral limb. Furthermore, Hopper,
Creagh, Formby, Goh, Boyle, and Strauss (2003) reported no
signicant difference in knee proprioception after 12e16 months
of ACL reconstruction by measuring JPS. Reider et al. (2003) used
an electrogoniometer to measure the JPS, however, Hopper et al.
(2003) used Peak Motus motion measurement to measure the
knee JPS. The results regarding proprioceptive function may be
contradictory not only due to differences in measurement
methods but also due to use of different equipment to measure
the proprioception.
Early measurement of JPS used a modied Thomas splint with
a Pearson knee piece or a copper frame with a motor and a cali-
brated scale (Barrett, 1991; Corrigan, Cashman, & Brady, 1992).
These measurements were subjected to a high degree of intertester
variability as the joint angle was visually estimated. Many studies
have used the isokinetic dynamometer, electrogoniometer, or 2D
video analysis to measure JPS (Birmingham et al., 1998; Hopper
et al., 2003; Reider et al., 2003).
Although all three measures of JPS have been used separately,
no studies have established if concurrent measurement will result
in similar values of JPS. Additionally, most of the studies tested JPS
in non-weight bearing conditions while functional activities are
performed in weight bearing conditions. It logically follows that
JPS should be tested in weight bearing conditions in order to
simulate functional activities. Therefore, this study aimed to
determine if three JPS measures (isokinetic dynamometer, elec-
trogoniometer, and 2D video analysis) produced equivalent
results.
2. Methods
2.1. Participants
A group (convenience sampling) of 30 participants (male and
female), ages 18e25 years with no history of injury or surgery to
the knee, were recruited from the university student population.
The number of participants was decided based on previous
studies (Grob et al., 2002; Reider et al., 2003) on proprioception
in ACL injuries. The inclusion criteria for the study were (a)
current enrollment at the university and (b) participation in
sports-related activities at least three times a week. The partici-
pation in sports activities was required so that everybody could
be at a similar level of proprioception. The exclusion criteria were
(a) any history of injury, pain, or swelling of the knee in the
previous year and (b) any history of medical problems which
could limit proprioception. All the inclusion and exclusion
criteria were determined by the participants answers on a health
questionnaire form. Approval of the proposal from the uni-
versitys ethical review board was obtained prior to data collec-
tion. All the participants were asked to sign an informed consent
form.
2.2. Procedures
Three measures were taken concurrently in sitting position:
isokinetic dynamometer, 2D video analysis, and electrogoniometer.
However, only two measures were taken concurrently in standing
position: 2D video analysis and electrogoniometer. A practice test
was given to familiarize the participant with the procedure before
testing. The order of the testing in sitting or standing positions was
determined randomly.
2.3. Sitting measurement
2.3.1. Calibration of electrogoniometer
The Penny and Giles M series twin axis electrogoniometer
(Penny Giles, UK) was used to measure JPS. It measures the
potential difference between the two endblocks, and converts the
difference in potential to the respective joint angle. Before data
collection, calibration of the electrogoniometer was done using
VICON Motus Motion System (VMMS) version 8.5 (Peak Perfor-
mance Technologies, Inc. CO, USA) and a universal goniometer. One
endblock of the electrogoniometer was placed on the xed and the
other on the movable arm of the universal goniometer for the
calibration (see Fig. 1). The movable arm was rotated in 10
incre-
ments, and the corresponding voltages were recorded from the
VMMS. Further, the following equation was obtained by plotting
a graph (R2 0.99) between the recorded voltage (x) and the angle
(y) to which the universal goniometer arm was moved:
Angley 88:118x 261:09 (1)
This equation was used to calculate the knee joint angle during JPS
testing.
2.3.2. Calibration of the video camera
A JVC camera (Victor Company of Japan, Japan) was used for
recording the video for the measurement of JPS by VMMS. Before
testing, the calibration of the cameras spatial eld was done by
VMMS using a 0.50 m  0.325 m frame with reective markers at
each corner.
2.3.3. Testing procedure
The participants were asked to sit on the isokinetic dynamom-
eter system 3 (Biodex Medical Systems, Shirley, New York) chair
with their trunks secured to the chair after calibrating the system as
per manufacturers guidelines. The endblocks of the goniometer
were placed on the lateral aspect of the dominant limb knee joint.
To measure the knee angle by 2D video analysis, the reective
markers were placed on the thigh, lateral knee joint line, and lateral
malleolus. The thigh marker was placed slightly forward and
downward to the greater trochanter after placing the other two
markers. All the markers were ensured to be in one line and in the
same plane in knee extended position. Additionally, the reective
marker at the knee joint was stemmed to bring all three markers in
one plane. The placement of the endblocks and the reective
markers were held constant throughout the experiment in sitting
and standing position.
Fig. 1. Placement of elctrogonometer endblocks.
 D. Kiran et al. / Physical Therapy in Sport 11 (2010) 81e85 83

Participants were seated with the knee exed to 90
on the

isokinetic dynamometer chair with eyes covered (Fig. 2). Testing
positions of 15
, 30
, and 45
were selected on the isokinetic
dynamometer protocol (Birmingham et al., 1998). The three test
positions were tested in the order of 15
, 30
, and 45
as the
dynamometer protocol does not permit random selection. The
researcher asked the participant to move the knee from the starting
position to the angle being demonstrated by the dynamometer and
then to return the knee to the starting position. The participants
were asked to remember each demonstrated position and then
return the knee joint to the demonstrated position. The participants
indicated the reproduction of the knee angle to the demonstrated
position by pressing a switch at the appropriate angle.
The measurements from camera and electrogoniometer were
also recorded by the VMMS at each corresponding demonstrated
and reproduced position. An error angle between the demonstrated
and the reproduced position was considered for analysis of JPS.

2.4. Standing measurement
The calibration and placement of the endblocks of the electro-
goniometer was the same as it was in the sitting position. However,
the cameras spatial eld was recalibrated by using VMMS and
1.83 m  0.92 m frame with reective markers at each corner.
2.4.1. Testing procedure
The participants were asked to stand with maximum weight on
the test limb and with eyes covered (Fig. 3). The foot of the non-
tested leg was touching the oor for stability only. They were asked
to ex the knee to a randomly determined position of around
0
e45
of knee exion. We considered the test position high, mid
and low when the knee was exed around 15
, 30
, and 45

respectively. The participants were asked to remember the
demonstrated position and return to a standing position. They were
then asked to place the knee joint in the demonstrated position.
Both the demonstrated and reproduced positions were recorded by
the electrogoniometer and the camera for the analysis (Hopper
et al., 2003; Reider et al., 2003).
Fig. 3. Placement of reective markers and endblocks for measurement of JPS in
standing.
tests. The independent variables were the testing procedures with
three levels (electrogoniometer, isokinetic dynamometer, and 2D
video analysis), and testing positions with two levels (sitting and
standing). The dependent variable, joint position sense, was
measured in degrees.
2.6. Data analysis
We conducted all planned comparisons using SPSS 15.0 statis-
tical software. First of all, we examined all the variables for distri-
butional assumptions and potential outliers using frequency
analysis and histograms. Following these analyses, descriptive
statistics were reported including means and standard deviations
for continuous variables (JPS score).
3. Results

2.5. Research design
The research design was a prospective within subjects design
that used concurrent measurement and random ordering of the JPS
Thirty healthy participants (13 females and 17 males) with no
history of injury in the lower limb were tested on their dominant
limb. Three trials of demonstrated and reproduced testing were
done at each position and the average absolute error values are
presented in Table 1. For the knee position at 15
only two absolute

Table 1
presents the descriptive statistics of absolute error of the knee joint position sense
with the mean, standard deviation, minimum and maximum values (in degrees) in
sitting and standing position.

Test Equipment Measured Minimum Maximum Mean

position position error error
Sitting Isokinetic 15 1 10 4.68  2.21
dynamometer 30 1 8 3.85  1.90
45 0.67 12 3.90  2.55
Electrogoniometer 15 0.27 7.30 3.24  1.63
30 0.53 7.10 2.90  1.61
45 0.26 8.53 3.42  2.04
2D video analysis 15 0.14 6.54 2.91  1.65
30 0.27 7.07 2.72  1.49
45 0.14 5.59 2.31  1.47

Standing Electrogoniometer High 0.31 9.01 3.34  2.33

Mid 1.2 9.15 3.85  2.07
Low 0.35 8.77 3.57  2.1
2D video analysis High 0.34 8.67 3.49  2.24
Mid 0.98 8.21 3.58  2.00
Fig. 2. Placement of reective markers and endblocks for measurement of JPS in sitting Low 0.31 7.26 3.24  1.99
position.
84 D. Kiran et al. / Physical Therapy in Sport 11 (2010) 81e85
errors were averaged due to loss of data. In the sitting position, the
mean absolute errors measured by the isokinetic dynamometer
were 4.68 (SD 2.21), 3.85 (SD 1.90), and 3.90 (SD 2.55) degrees in
the 15
, 30
and 45
test positions respectively. The mean absolute
errors measured by the electrogoniometer in the same positions
were 3.24 (SD 1.63), 2.90 (SD 1.61), and 3.42 (SD 2.04) degrees. The
2D video analysis mean absolute measures of the knee JPS were
2.91 (SD 1.65), 2.72 (SD 1.49), and 2.31 (SD 1.47) degrees in the
similar test positions. In the standing position, the mean absolute
errors measured by electrogoniometer were 3.34 (SD 2.33), 3.85
(SD 2.07), and 3.57 (SD 2.10) degrees in the high-, mid-, and low-
test positions respectively. In the same test positions, the mean
absolute errors measured by 2D video analysis were 3.49 (SD 2.24),
3.58 (SD 2.00), and 3.24 (SD 1.99) degrees. To assess the concurrent
reliability the correlations between average absolute error angles of
two measurement techniques were determined at each position in
both sitting (Table 2) and standing (Table 3). Excellent correlation
(0.94e0.98) in the standing knee measurements is noted at each
joint position between the electrogoniometer and the 2D video
analysis. However, good to excellent level of correlation
(0.63e0.92) is present between the electrogoniometer and iso-
kinetic dynamometer measurements at all three of the positions in
sitting. Furthermore, fair to good correlation (0.41e0.63) is present
between the two dimensional video and either of the other two
measures (electrogoniometer and isokinetic dynamometer
measurements).
4. Discussion
Although knee joint position sense (JPS) has been measured by
different equipment in the research literature, the correlation
among the measures has not been studied. The present study
aimed to evaluate if the concurrent measures of the knee JPS by the
three measures (isokinetic dynamometer, electrogoniometer, and
2D video analysis) were highly correlated. These measures have
been used by numerous researchers for several years (Birmingham
et al., 1998; Hopper et al., 2003; Reider et al., 2003). Further, we
tested the reliability of the measures in both sitting and standing
positions since both positions were used in the research literature.
In the standing position, we found excellent correlation
(0.94e0.98) between the electrogoniometer and 2D video analysis
measurement (Fleiss, 1981). The excellent correlation in standing
position appeared to be due to the minimal level of error in
measurement of joint angles by 2D video analysis when the camera
was placed at 90
from the plane of motion. Additionally, the
analysis of the measures of JPS in standing suggests that neither
measure is better than the other, as both measures had equivalent
Table 2
presents the inter-correlations between all the positions in sitting.
Equipment Measured 2D video analysis
position
15 30
Isokinetic 15 0.41
dynamometer (r2 0.17)
30 0.53
(r2 0.28)
45
Electrogoniometer 15 0.55
(r2 0.30)
30 0.52
(r2 0.27)
45
Table 3
presents the inter-correlations between all the test positions in standing.
Equipment Measured 2D video analysis
position
High Mid Low
Electrogoniometer High 0.98
(r2 0.96)
Mid 0.96
(r2 0.92)
Low 0.94
(r2 0.88)
values of JPS. Therefore, researchers may use either measure during
knee JPS measurement in standing position with condence.
In the standing measurements, we were able to place the
camera close to a 90
position; however, in sitting position
measurement the placement of the camera at 90
was not possible
due to the isokinetic dynamometer arm blocking the cameras
visual eld. Therefore, we had to place the camera at an angle of 10

where the dynamometer could not obstruct the visibility of the
reective markers.
In this study, we found that in the sitting position the correlation
between the measures of the isokinetic dynamometer and
the electrogoniometer was in the range of 0.63e0.92 within the
different tested positions. However, the correlation between the
measures of isokinetic dynamometer and the 2D video analysis or
between the 2D video analysis and the electrogoniometer was in
the range of 0.41e0.63 or 0.52e0.57 respectively. The fair to good
correlation between 2D video analysis and either of the other two
measures appeared to be due to the inability of the VICON Motus
motion system (VMMS) to measure joint angles accurately when
the camera position was close to 10
from the plane of motion. In
a previous study, (Kiran, Carlson, Medrano, & Smith, 2010), we
reported increased error when the camera was placed at 5
e10

from the plane of motion. Therefore, the increased error in JPS
measurement by 2D video analysis could be the reason for the fair
to good correlation between 2D video analysis measure and iso-
kinetic dynamometer or electrogoniometer measure. However,
when the camera was positioned perpendicular to the plane of
motion, the 2D video analysis measures were highly correlated
with the electrogoniometer measures as evidenced by the standing
JPS correlation values. Further studies may be done to discern the
correlation between concurrent JPS measurements by the electro-
goniometer and the 2D video analysis in the sitting position.
The correlation between the electrogoniometer and the iso-
kinetic dynamometer in the sitting position ranged from 0.63 to
0.92 with an increase in correlation from 15
knee exion to 45

knee exion position. The differential in correlation in these posi-
tions may be due to more skin movement in the thigh in a fully
Electrogoniometer
45 15 30 45
0.63
(r2 0.40)
0.80
(r2 0.64)
0.63 0.92
(r2 0.40) (r2 0.85)
0.57
(r2 0.32)
 D. Kiran et al. / Physical Therapy in Sport 11 (2010) 81e85
extended position than in a exed position. Kuo, Tully, and Galea
(2008) reported displacement of skin markers in the lateral thigh
during hipe knee exion in sitting position. The measurement of
the joint angle by electrogoniometer depends on the change in
voltage between the two endblocks. The differential skin move-
ment in the thigh from extension to exion of the knee may
interfere with the change in voltage between the two endblocks,
yielding altered JPS measurement values. Further research may be
done to nd out the accuracy of electrogoniometer measures at
different degrees of knee exion angle. Additionally, research may
be done to discover the amount of skin movement throughout the
knee range of motion. The future study may also include correlation
in sitting position of knee JPS measures by an electrogoniometer
and 2D video analysis with a camera perpendicular to the plane of
motion.
In conclusion, since excellent correlation between 2D video
analysis and electrogoniometer was found, either electro-
goniometer or 2D video analysis may be used in the standing
position to measure JPS; if the camera is placed perpendicular to
the plane of motion. Accordingly, either isokinetic dynamometer or
electrogoniometer may be used to measure knee JPS in sitting
position as the correlation between those two measures was good
to excellent. However, in sitting position, the correlation of the 2D
video analysis and either of the other two measures was only fair to
good. Therefore, the selection of 2D video analysis to measure knee
JPS in sitting position with the camera placed at 10
from the plane
of movement should be avoided.
Acknowledgements
The research was funded by College of Health Sciences, the
University of Texas at El Paso.
Conict of interest
None.
Ethical statement
The study was approved by the University of Texas at El Paso
review board (88060-4).
References
Aune, A. K., Holm, I., Risberg, M. A., Jensen, H. K., & Steen, H. (2001). Four-strand
hamstring tendon autograft compared with patellar tendon-bone autograft for
anterior cruciate ligament reconstruction. A randomized study with two-year
follow-up. The American Journal of Sports Medicine, 29(6), 722e728.
85
Barrett, D. S. (1991). Proprioception and function after anterior cruciate recon-
struction. The Journal of Bone and Joint Surgery. British Volume, 73(5), 833e837.
Bergin, P. S., Bronstein, A. M., Murray, N. M., Sancovic, S., & Zeppenfeld, D. K. (1995).
Body sway and vibration perception thresholds in normal aging and in patients
with polyneuropathy. Journal of Neurology, Neurosurgery, and Psychiatry, 58(3),
335e340.
Beynnon, B. D., & Johnson, R. J. (1996). Anterior cruciate ligament injury rehabilita-
tion in athletes. Biomechanical considerations. Sports Medicine (Auckland, N.Z.),
22(1), 54e64.
Birmingham, T. B., Kramer, J. F., Inglis, J. T., Mooney, C. A., Murray, L. J., Fowler, P. J.,
et al. (1998). Effect of a neoprene sleeve on knee joint position sense during
sitting open kinetic chain and supine closed kinetic chain tests. The American
Journal of Sports Medicine, 26(4), 562e566.
Corrigan, J. P., Cashman, W. F., & Brady, M. P. (1992). Proprioception in the cruciate
decient knee. The Journal of Bone and Joint Surgery. British Volume, 74(2), 247e250.
Fleiss, J. L. (1981). Statistical methods for rates and proportions (2nd ed.). New York:
John Wiley & Sons.
Friden, T., Roberts, D., Ageberg, E., Walden, M., & Zatterstrom, R. (2001). Review of
knee proprioception and the relation to extremity function after an anterior
cruciate ligament rupture. The Journal of Orthopaedic and Sports Physical
Therapy, 31(10), 567e576.
Grifn, L. Y., Agel, J., Albohm, M. J., Arendt, E. A., Dick, R. W., Garrett, W. E., et al.
(2000). Noncontact anterior cruciate ligament injuries: risk factors and
prevention strategies. The Journal of the American Academy of Orthopaedic
Surgeons, 8(3), 141e150.
Grob, K. R., Kuster, M. S., Higgins, S. A., Lloyd, D. G., & Yata, H. (2002). Lack of
correlation between different measurements of proprioception in the knee. The
Journal of Bone and Joint Surgery. British Volume, 84(4), 614e618.
Henriksson, M., Ledin, T., & Good, L. (2001). Postural control after anterior cruciate
ligament reconstruction and functional rehabilitation. The American Journal of
Sports Medicine, 29(3), 359e366.
Hopper, D. M., Creagh, M. J., Formby, P. A., Goh, S. C., Boyle, J. J., & Strauss, G. R.
(2003). Functional measurement of knee joint position sense after anterior
cruciate ligament reconstruction. Archives of Physical Medicine and Rehabilita-
tion, 84(6), 868e872.
Kavounoudias, A., Roll, R., & Roll, J. P. (2001). Foot sole and ankle muscle inputs
contribute jointly to human erect posture regulation. The Journal of Physiology,
532(Pt 3), 869e878.
Kiran, D., Carlson, M. E., Medrano, & Smith, D. R. (2010). Validity of two dimensional
VICON Motus motion software at varying camera angles. International Journal of
Exercise Science: Conference Abstract Submissions, 4(1).
Kuo, Y. L., Tully, E. A., & Galea, M. P. (2008). Skin movement errors in measurement
of sagittal lumbar and hip angles in young and elderly subjects. Gait & Posture,
27(2), 264e270. doi:10.1016/j.gaitpost.2007.03.016.
Lephart, S. M., Pincivero, D. M., & Rozzi, S. L. (1998). Proprioception of the ankle and
knee. Sports Medicine (Auckland, N.Z.), 25(3), 149e155.
MacDonald, P. B., Hedden, D., Pacin, O., & Sutherland, K. (1996). Proprioception in
anterior cruciate ligament-decient and reconstructed knees. The American
Journal of Sports Medicine, 24(6), 774e778.
Manchester, D., Woollacott, M., Zederbauer-Hylton, N., & Marin, O. (1989). Visual,
vestibular and somatosensory contributions to balance control in the older
adult. Journal of Gerontology, 44(4), M118eM127.
Reider, B., Arcand, M., Diehl, L., Mroczek, K., Abulencia, A., Stroud, C. C., et al. (2003).
Proprioception of the knee before and after anterior cruciate ligament recon-
struction. Arthroscopy: The Journal of Arthroscopic & Related Surgery, 19(1), 2e12.
Sharma, L., Cahue, S., Song, J., Hayes, K., Pai, Y. C., & Dunlop, D. (2003). Physical
functioning over three years in knee osteoarthritis: role of psychosocial, local
mechanical, and neuromuscular factors. Arthritis and Rheumatism, 48(12),
3359e3370.
Taimela, S., Kujala, U. M., & Osterman, K. (1990). Intrinsic risk factors and athletic
injuries. Sports Medicine (Auckland, N.Z.), 9(4), 205e215.
Tinetti, M. E., & Speechley, M. (1989). Prevention of falls among the elderly. The New
England Journal of Medicine, 320(16), 1055e1059.