1441
Vastus Medialis Obliquus and Vastus Lateralis Activity in
Open and Closed Kinetic Chain Exercises in Patients With
Patellofemoral Pain Syndrome: An Electromyographic Study
Simon F.T. Tang, MD, Chih-Kuang Chen, MD, Robert Hsu, MD, Shih-Wei Chou, MD, PhD,
Wei-Hsien Hong, MS, Henry L. Lew, MD, PhD
ABSTRACT. Tang SFT, Chen C-K, Hsu R, Chou S-W,
Hong W-H, Lew HL. Vastus medialis obliquus and vastus
lateralis activity in open and closed kinetic chain exercises
in patients with patellofemoral pain syndrome: an electro-
myographic study. Arch Phys Med Rehabil 2001;82:1441-5.
Objective: To evaluate the electromyographic activities of
vastus medialis obliquus (VMO) and vastus lateralis (VL)
muscles in open and closed kinetic chain exercises in subjects
with patellofemoral pain syndrome (PFPS).
Design: Case-controlled study.
Setting: Rehabilitation science center in a tertiary medical
center.
Participants: Ten patients with bilateral knee pain diag-
nosed with PFPS and 10 healthy volunteers.
Interventions: Subjects performed open kinetic chain exer-
cise on an isokinetic dynamometer and closed kinetic chain
exercise by squat-to-stand and stand-to-squat tasks. Surface
electromyography was done for the VMO and VL muscles.
Main Outcome Measures: VMO/VL ratios were calculated
after normalization of muscle activities.
Results: The VMO/VL ratios of PFPS subjects were signif-
icantly lower than were those of unimpaired subjects during
knee isokinetic closed kinetic chain exercises (p ⫽ .047).
However, there was no statistical difference in VMO/VL ratio
between subjects with and without PFPS during closed kinetic
chain exercises (p ⫽ .623). Maximum VMO/VL ratio was
obtained at 60° knee flexion in closed kinetic chain exercise.
Conclusion: In closed kinetic chain exercises, more selec-
tive VMO activation can be obtained at 60° knee flexion.
Maximal VMO/VL ratio was observed at this knee flexion
angle, and muscle contraction intensity was also greatest.
Key Words: Electromyography; Exercise; Femur; Knee;
Pain; Patella; Rehabilitation.
© 2001 by the American Congress of Rehabilitation Medi-
cine and the American Academy of Physical Medicine and
Rehabilitation
From the Departments of Physical Medicine and Rehabilitation (Tang, Chen, Chou,
Hong) and Orthopedic Surgery (Hsu), Chang Gung Memorial Hospital, Tao-Yuan,
Taiwan; and Physical Medicine and Rehabilitation Service, VA Palo Alto Health Care
System, Stanford University School of Medicine, Palo Alto, CA (Lew).
Accepted in revised form January 3, 2001.
Supported by the National Science Council, Taiwan, Republic of China (grant no.
NSC85-2331-B-182-064).
No commercial party having a direct financial interest in the results of the research
supporting this article has or will confer a benefit upon the author(s) or upon any
organization with which the author(s) is/are associated.
Reprint requests to Simon F.T. Tang, MD, Dept of Physical Medicine and Reha-
bilitation, Chang Gung Memorial Hospital, 5 Fu-Shin St, Kwei-Shan, Tao-Yuan 333,
Taiwan, e-mail: fttang@adm.cgmh.org.tw.
0003-9993/01/8210-6419$35.00/0
doi:10.1053/apmr.2001.26252
ATELLOFEMORAL PAIN SYNDROME (PFPS) is a
P common problem that affects both the athletic and nonath-
letic populations. It is among the most common knee problems
1
in physically active young adults.2 PFPS is defined as retropa-
tellar or peripatellar pain, or both, that results from physical
and biochemical changes in the patellofemoral joint.3 The basic
origin and exact pathogenesis of PFPS are unknown, but many
predisposing factors have been proposed, including acute
trauma, knee ligament injury and surgery, instability, overuse,
immobilization, overweight, genetic predisposition, malalign-
ment or dysfunction of the knee extensor mechanism, congen-
ital anomalies of the patella, prolonged synovitis, recurrent
hemorrhage into a joint, joint infection, and repetitive intraar-
ticular injections of corticosteroids. In many cases, however,
there are no obvious reasons for the symptoms, and there is no
clear association between the severity of the symptoms and the
radiologic and arthroscopic findings.2
The biomechanics of the patellofemoral joint are controlled
by its static and dynamic components. The primary dynamic
components are the 4 parts of the quadriceps femoris complex,
with accessory input from the iliotibial band, the adductor
magnus and longus, the pes anserine group, and the biceps
femoris. Some theories for the origin of nontraumatic gradual
onset of PFPS are: (1) neuromuscular imbalances of the vastus
medialis obliquus (VMO) and the vastus lateralis (VL) mus-
cles; (2) tightness of the lateral knee retinaculum, hamstrings,
iliotibial band, and gastrocnemius; and (3) overpronation of the
subtalar joint. Excessive lateral tracking of the patella in the
trochlear groove of the femur is believed to be caused by the
previously mentioned factors. Lateral excessive tracking of the
patella is evident in many people with PFPS.1,3,4
The Q angle deserves special mention because a large Q
angle is often implicated as a predisposing factor in patients
with PFPS. It is defined as the angle between the quadriceps
force vector and the patella-tibial tubercle axis.
However, there is no consensus about its importance, and it
has not been shown to vary significantly between symptomatic
and asymptomatic subjects.3
As stated earlier, PFPS has multifactorial causes. Many
rehabilitation strategies have been implemented for patients
with PFPS. In general, the goals of patellofemoral rehabilita-
tion are to maximize quadriceps strength while minimizing the
patellofemoral joint reaction forces and stress.5-10 Strengthen-
ing of the quadriceps mechanism is typically performed in the
last 30° of knee extension, but the selectivity of this method for
strengthening only the VMO is debatable. Recent works by
Steinkamp et al9 indicates that the best method to strengthen
the quadriceps group while incurring the least patellofemoral
joint reaction force and stress is with a short arc (⬍45° flexion
to extension), closed kinetic chain exercise.9 Selective VMO
strengthening may possibly be enhanced by hip adduction
exercises, which may also give the VMO a stable origin from
which to contract. Hanten and Schulthies11 showed that during
Arch Phys Med Rehabil Vol 82, October 2001
1442 EXERCISE AND PATELLOFEMORAL PAIN SYNDROME, Tang

hip adduction, the electrical activity of the VMO is signifi-

cantly greater than that of the VL. Therefore, hip adductor
contraction, in conjunction with quadriceps sets and straight
leg raises, is recommended to facilitate VMO strengthening.
This study assessed the VMO and VL concentric and eccen-
tric contraction performances in isokinetic open kinetic chain
and closed kinetic chain exercises. The open kinetic chain
exercise was performed with a KIN-COM isokinetic dyna-
mometer.a The closed kinetic chain exercise was performed
with squat-to-stand and stand-to-squat exercises. The raw elec-
tromyographic data for VMO and VL muscles were rectified
and quantified. VMO/VL ratios were calculated after normal-
ization of muscle activities.

METHODS

Subjects
Ten subjects (5 women, 5 men; age range, 21–32yr; mean,
25.7yr) with asymptomatic knees and 10 subjects (6 women, 4
men; age range, 19 – 48yr; mean, 28.2yr) with bilateral PFPS
were included in the study. Asymptomatic subjects were de-
fined as individuals with pain-free knees who had never had
significant knee or lower extremity pathology or surgery.
The symptomatic subjects had no history of direct trauma to
the patellofemoral joint, but had bilateral anterior knee pain for
a minimum of 6 weeks. The pain was elicited by squatting, stair
climbing, kneeling, and prolonged sitting. Patients with con-
comitant ligamentous injury were excluded. Examinations of
the knees with merchant’s views confirmed patellofemoral
malalignment. The symptomatic group was diagnosed and ex-
amined by the same physiatrist (SFTT).

Procedures
Subjects performed all tasks under instructions from an
engineer at the Rehabilitation Science & Engineering Service
Center of Chang Gung Memorial Hospital, Taiwan. Informed
consent was obtained from all subjects before they entered the
study.
To prepare the area for attachment of surface electromyo-
graphic electrodes, the anterior thigh and medial tibial crest
were cleaned with isopropyl alcohol. Surface electrodes were
centered on the muscle bellies of the VMO and distal VL, as
described by Basmajian and Blumenstein.12 A MyoSystem
200b was used to acquire electromyographic signals from 9
channels, of which 8 signals were from the surface electrodes
placed on the lower limbs. The last channel was used as an
event marker. The raw electromyographic signals were band-
pass filtered (15–500Hz) and sampled at 1000Hz. The signals
were rectified and low-pass filtered (cutoff frequency, 6Hz).
This system used bipolar gold-plated surface electrodes to
collect electromyographic signals.
Both legs of subjects in the symptomatic group were tested.
Because there is no evidence of a relation between unilateral
patellofemoral dysfunction and leg dominance, we tested ran-
domly chosen legs of the asymptomatic subjects. Every subject
in both groups was tested under 2 exercise modes, the open
kinetic chain and the closed kinetic chain.
In the isokinetic open kinetic chain exercise, a KIN-COM
isokinetic dynamometer was used. The testing procedure began
with a 5-minute warm-up on a stationary bicycle, followed by
stretching of the quadriceps, hamstrings, iliotibial band, and
calf musculatures. Subjects’ positioning on the dynamometer
was according to the manufacturer’s protocol, with the the back
supported and the hip flexed to approximately 80°. Trunk and
Fig 1. Closed kinetic chain exercise by squatting and standing.
thigh straps were fastened for stabilization. Quadriceps con-
centric and eccentric contractions were recorded at between 0°
and 90° of knee flexion. All the tests were done at the angular
velocity of 120°/s. Both legs of symptomatic subjects were
tested, with the order of testing chosen randomly. After testing
there was a 5-minute cool-down period, after which ice was
applied for 10 minutes to the examined knees.
In the closed kinetic chain exercise, stand-to-squat, and
squat-to-stand tasks were repeated twice in our gait laboratory
with a Vicon 370 motion analysis systemc with 6 cameras
containing infrared light-emitting diodes. Reflective markers
were placed on the sacrum, bilateral anterosuperior iliac spines,
greater trochanters, anterior thighs, medial and lateral malleoli,
anterior tibia, dorsum of the foot, fifth metatarsal heads, and
posterior heels. Motion analysis of the lower extremity was
then determined for the sagittal plane. The trunk was kept
upright and, to obtain better balance control, the subjects were
asked to hold the 2 arms of a walker (fig 1). Quadriceps
eccentric contraction intensity was recorded during squatting,
whereas quadriceps concentric contraction intensity was re-
corded during squatting-to-standing tasks. The contraction in-
tensities were recorded by surface polyelectromyography. Sub-
jects performed all the tasks at a speed comfortable to them.

Arch Phys Med Rehabil Vol 82, October 2001

 EXERCISE AND PATELLOFEMORAL PAIN SYNDROME, Tang 1443

In the normalization process of muscle activity, subjects

were positioned at the dynamometer. Maximum voluntary iso-
metric contraction of the quadriceps was performed from 90°
knee flexion to complete full knee extension at 15° intervals,
and the electromyographic activities of the VMO and VL were
recorded. Each maximum contraction was done for 5 seconds.
In the PFPS group, we found that when knee flexion angle
increased, knee pain became more obvious, and there was less
quadriceps isometric contraction intensity. The quadriceps iso-
metric contraction activities were similar in both groups during
complete knee extension. Therefore, the maximum quadriceps
isometric contraction data obtained during complete knee ex-
tension was used as the reference value.
Normalized VMO and VL electromyographic data obtained
in both exercise modes were presented as the VMO/VL ratio.
A ratio greater than 1 meant that the VMO was exerting a Fig 2. VMO/VL ratio of quadriceps contraction in isokinetic exer-
cise. Abbreviations: con, concentric; ecc, eccentric.
larger muscular contractile intensity than was the VL. In both
exercise modes, electromyographic data were obtained at 15°
intervals (15°, 30°, 45°, 60°, 75°, 90°). During knee flexion
from 0° to 90°, 6 electromyographic data points could be VMO/VL ratios were smaller in the PFPS symptomatic
obtained for quadriceps eccentric contractions. The same data group (p ⫽ .047). When the concentric contractions in both
acquisition applied for quadriceps concentric contractions. Six groups were compared with eccentric contractions, the con-
electromyographic data points could be obtained during knee tractile intensity was smaller for concentric contraction
extension from ⫺90° to 0°. forces (p ⫽ .015).
In the closed kinetic chain exercise during stand-to-squat
Statistical Analysis and squat-to-stand tasks, VMO/VL ratios greater than 1
In the symptomatic group, the mean VMO/VL ratio for each were observed during concentric contractions at 60°
subject was calculated by averaging the VMO/VL ratio of both (1.052 ⫾ .430) and 90° (1.025 ⫾ .204) of knee flexion for
legs. SYSTAT softwared and repeat-measures analysis of vari- the PFPS symptomatic group. Eccentric contraction with
ance were used to analyze the VMO/VL ratios between asymp- VMO/VL ratio greater than 1 was noted during 60° (1.153 ⫾
tomatic and symptomatic subjects in both exercise modes. .299) and 75° (1.054 ⫾ .298) of knee flexion for the same
Statistical significance was set at p less than .05. group. In the asymptomatic group, VMO/VL ratio greater
than 1 was observed only for eccentric contraction during
RESULTS 60° of knee flexion (1.080 ⫾ .233) (table 2, fig 3). As in
the open kinetic chain exercise, we compared VMO/VL
In the open kinetic chain exercise with the dynamometer,
ratios in both groups; however, no statistical significance
we observed a VMO/VL ratio greater than 1 during eccen-
was found, nor was there statistical significance when con-
tric contractions at 75° (1.083 ⫾ .582) and 90° (1.105 ⫾
centric and eccentric contractions from both groups were
.695) of knee flexion in the PFPS symptomatic group. In the
compared.
asymptomatic group, VMO/VL ratio was greater than 1
during eccentric contractions at 60° (1.030 ⫾ .380), 75° DISCUSSION
(1.138 ⫾ .551), and 90° (1.259 ⫾ .916) of knee flexion
(table 1, fig 2). When the entire symptomatic group, which PFPS is frequently encountered in physically active young
includes concentric and eccentric contractions in all angles, adults and in sports injury clinics. The causes for this syndrome
was compared with the asymptomatic group, we found that remain enigmatic.13 Witvrouw et al14 proposed that a shortened
quadriceps muscle, an altered VMO muscle reflex response
time, a decreased explosive strength, and a hypermobile patella
had a significant correlation with the incidence of patellofemo-
Table 1: VMO/VL Ratio of Quadriceps Muscle Contraction in KIM- ral pain and are risk factors for PFPS. Predisposing factors such
COM Open Chain Isokinetic Exercises as repetitive loading of the patellofemoral joint, trauma, con-
Normal Asymptomatic genital anomalies, and malalignment of the patella are also
Knee
PFPS Symptomatic Subjects Subjects believed to cause PFPS.1,2
Flexion Concentric Eccentric Concentric Eccentric
(deg) Contraction* Contraction† Contraction† Contraction§
Table 2: VMO/VL Ratio of Quadriceps Muscle Contraction in
15° .803 ⫾ .149 .862 ⫾ .177 .945 ⫾ .227 .932 ⫾ .320 Closed Kinetic Chain Exercise by Squatting and Standing
30° .832 ⫾ .160 .875 ⫾ .173 .857 ⫾ .209 .942 ⫾ .387
45° .877 ⫾ .145 .897 ⫾ .188 .884 ⫾ .154 .994 ⫾ .304 Knee PFPS Subjects Asymptomatic Subjects
Flexion
60° .822 ⫾ .142 .916 ⫾ .234 .888 ⫾ .223 1.030 ⫾ .380
(deg) Concentric Eccentric Concentric Eccentric
75° .851 ⫾ .203 1.083 ⫾ .582 .923 ⫾ .164 1.138 ⫾ .551
90° .831 ⫾ .255 1.105 ⫾ .695 .959 ⫾ .295 1.259 ⫾ .916 15° .793 ⫾ .307 .881 ⫾ .405 .726 ⫾ .197 .807 ⫾ .289
30° .889 ⫾ .222 .854 ⫾ .444 .830 ⫾ .235 .833 ⫾ .354
* p ⫽ .047 when comparing the entire symptomatic group, which 45° .989 ⫾ .218 .965 ⫾ .359 .932 ⫾ .224 .938 ⫾ .388
includes concentric and eccentric contractions in all angles (* and †),
60° 1.052 ⫾ .430 1.153 ⫾ .299 .980 ⫾ .265 1.080 ⫾ .233
with the asymptomatic group († and §).
†
p ⫽ .015 when comparing the concentric contractions in both 75° .995 ⫾ .214 1.054 ⫾ .298 .978 ⫾ .200 .985 ⫾ .234
groups (* and †), with the eccentric contractions in both groups († 90° 1.025 ⫾ .204 .972 ⫾ .276 .984 ⫾ .171 .979 ⫾ .319
and §).

Arch Phys Med Rehabil Vol 82, October 2001

1444 EXERCISE AND PATELLOFEMORAL PAIN SYNDROME, Tang
Fig 3. VMO/VL ratio of quadriceps contraction in closed kinetic
chain exercise by squatting and standing.
Many rehabilitation strategies have been suggested for quad-
riceps strengthening, especially in training the VMO. Closed
kinetic chain exercises, such as squatting, working with mul-
tiple joint proprioceptive reactions, and muscular cocontrac-
tion, are generally assumed to be more functional.10,15 The
VMO/VL ratios in our study for squat-to-stand tasks showed
no significant differences between asymptomatic and PFPS
subjects. This may explain why similar firing patterns of the
VMO and VL muscles in both groups were generated during
closed kinetic chain exercises. Steinkamp et al9 found that the
best method to strengthen the quadriceps group is via short arc
(⬍45° flexion to extension), closed kinetic chain exercises.
However, in our squat-to-stand closed kinetic chain exercise
for the PFPS symptomatic group, knee flexion angles from 15°
to 45° revealed a VMO/VL ratio smaller than 1, with the
optimal ratio exceeding 1 noted at 60° knee flexion. Whether
the short-arc closed kinetic chain exercise should be increased
to 60° knee flexion is debatable, in view of our small sample
size.
We observed an interesting phenomenon in our study. As
knee flexion angle increased, there were some angles with
VMO/VL ratios greater than 1. However, subjects from the
PFPS group complained of knee pain when the knees were
more flexed. From the electromyography standpoint, the quad-
riceps contraction intensity was smaller, although the
VMO/VL ratio was greater than 1. Whether the knees should
be flexed to the angle with most optimal VMO/VL ratio for
VMO strengthening requires further investigation. We must
consider possible knee pain and muscle contraction intensities.
During closed kinetic chain exercise, the knee flexion angle
with a VMO/VL ratio greater than 1 and with optimal quadri-
ceps muscle contraction activity was 60°.
Souza and Gross16 have shown that the VMO/VL ratio was
significantly greater in reality subjects than in patients with
PFPS. The same result was noted in our study—the VMO/VL
ratio was greater in the asymptomatic group (p ⫽ .047). Thus,
the isokinetic dynamometer may be an ideal testing tool with
which to diagnose and assess PFPS. However, this remains
controversial, given that Callaghan et al17 have performed
multijoint isokinetic and isometric assessments in patients with
PFPS. The knee angular velocity with an isokinetic dynamom-
eter is important in muscle performance. A reasonable and
comfortable range for test velocity is between 60°/s and 180°/s.
We chose 120°/s because it was close to the average knee
angular velocity observed during squatting and standing tasks.
We found greater quadriceps eccentric muscle contraction
during closed kinetic chain exercise than with concentric con-
traction. Dvir et al18 showed that eccentric contractions result
Arch Phys Med Rehabil Vol 82, October 2001
in significantly higher pain ratings than do concentric con-
tractions. Our results showed that the maximal VMO eccentric
contractions occurred at 75° to 90° of knee flexion. Knee pain
is usually induced in PFPS subjects during this range of motion
(ROM). With this increase in knee flexion and in quadriceps
tension, patellofemoral joint reaction force is increased over
the patellofemoral joint. A larger patellofemoral joint reaction
force may yield high articular stresses and heighten the
chances of subchondral degenerative changes.5 Although there
is a higher VMO/VL ratio generated in eccentric isokinetic
exercise during knee flexion, the clinical application of eccen-
tric contraction in VMO training should be considered cau-
tiously.
CONCLUSION
Closed kinetic chain exercise from 0° to 60° of knee
flexion can induce maximal VMO firing, and the overall
VMO/VL ratio is lower for PFPS patients than for asymp-
tomatic subjects. However, our sample size was relatively
small. A larger sample size is needed to verify whether
closed kinetic chain exercise from 0° to 60° of knee flexion
is more beneficial than short arc, closed kinetic chain exer-
cises. Further study is also needed to observe whether knee
pain will be induced during this ROM.
References
1. Sheely P, Burdett R, Irrgang J. An electromyographic study of
vastus medialis oblique and vastus lateralis activity while ascend-
ing and descending steps. J Orthop Sports Phys Ther 1998;27:
423-9.
2. Natri A, Kannus P, Jarvinen M. Which factors predict the long-
term outcome in chronic patellofemoral pain syndrome? A 7-yr
prospective follow-up study. Med Sci Sports Exerc 1998;30:
1572-7.
3. Blond L, Hansen L. Patellofemoral pain syndrome in athletes: a
5.7 year retrospective follow-up study of 250 athletes. Acta Or-
thop Belg 1998;64:393-400.
4. Kramer PG. Patellar malalignment syndrome: rationale to reduce
excessive lateral pressure. J Orthop Sports Phys Ther 1996;8:
301-9.
5. Braddom R. Physical medicine and rehabilitation. In: Casazza B,
Young J, editors. Musculoskeletal disorders of the lower limbs.
2nd ed. Philadelphia: WB Saunders; 2000. p 834-7.
6. Bechman M, Craig R, Lehman RC. Rehabilitation of patellofemo-
ral dysfunction in the athlete. Clin Sports Med 1989;8:841-60.
7. Shelton GL, Thigpen LK. Rehabilitation of patellofemoral dys-
function: a review of literature. J Orthop Sports Phys Ther 1991;
14:243-9.
8. Henry JH, Crosland JW. Conservative treatment of patellofemoral
subluxation. Am J Sports Med 1979;7:12-4.
9. Steinkamp LA, Dillingham MF, Markel MD, Hill JA, Kaufman
KR. Biomechanical considerations in patellofemoral joint rehabil-
itation. Am J Sports Med 1993;21:438-44.
10. Woodall W, Welsh J. A biomechanical basis for rehabilitation
program involving the patellofemoral joint. J Orthop Sports Phys
Ther 1990;11:535-42.
11. Hanten WP, Schulthies SS. Exercise effect on electromyographic
activity of the vastus medialis oblique and vastus lateralis mus-
cles. Phys Ther 1990;70:561-5.
12. Basmajian JV, Blumenstein R. Electrode placement in EMG
biofeedback. Baltimore: Williams & Wilkins; 1980
13. Kannus P, Natri A, Paakkala T. An outcome study of chronic
patellofemoral pain syndrome. Seven year follow-up of patients in
a randomized, controlled trial. J Bone Joint Surg Am 1999;81:
355-63.
14. Witvrouw E, Lysens R, Bellemans J, Cambier D, Vanderstraeten
G. Intrinsic risk factors for the development of anterior knee pain
in an athletic population. A two-year prospective study. Am J
Sports Med 2000;28:408-9.
 EXERCISE AND PATELLOFEMORAL PAIN SYNDROME, Tang
15. Palmitier RA, An KN, Scott SG, Chao EY. Kinetic chain exercise
in knee rehabilitation. Sports Med 1991;11:404-13.
16. Souza DR, Gross MT. Comparison of MVO:VL integrated elec-
tromyographic ratios between healthy subjects and patients with
patellofemoral pain. Phys Ther 1991;71:310-20.
17. Callaghan MJ, McCarthy CJ, Al-Omar A, Oldham JA. The repro-
ducibility of multi-joint isokinetic and isometric assessments in a
healthy and patient population. Clin Biomech 2000;15:678-83.
18. Dvir Z, Halperin N, Shklar A, Robinson D. Quadriceps function
and patellofemoral pain syndrome. Part I: pain provocation during
1445
concentric and eccentric isokinetic contractions. Isokinet Exerc
Sci 1991;1:26-30.
Suppliers
a. Chattecx Corp, 4717 Hixson Rd, PO Box 489, Hixson, TN 34343.
b. Noraxon USA Inc, 13430 N Scottsdale Rd, Ste 104, Scottsdale, AZ
85254.
c. Oxford Metrics Ltd, Unit 14, 7 West Way, Oxford OX2 OJB,
England.
d. SPSS Inc, 233 S Wacker Dr, 11th Fl, Chicago, IL 60606.
Arch Phys Med Rehabil Vol 82, October 2001