Gait & Posture 98 (2022) 187–194

Contents lists available at ScienceDirect

Gait & Posture

journal homepage: www.elsevier.com/locate/gaitpost

Differences in causes of stiff knee gait in knee extensor activity or ankle

kinematics: A cross-sectional study
Kazuki Fujita a, *, Yuichi Tsushima b, Koji Hayashi c, Kaori Kawabata a, Mamiko Sato c,
Yasutaka Kobayashi a
a
Graduate School of Health Science, Fukui Health Science University, Fukui-city, Fukui, Japan
b
Department of Physical Therapy Rehabilitation, Fukui General Hospital, Fukui-city, Fukui, Japan
c
Department of Rehabilitation Medicine, Fukui General Hospital, Fukui-city, Fukui, Japan

A R T I C L E I N F O A B S T R A C T

Keywords: Background: Stiff knee gait (SKG), a common occurrence after the onset of stroke, is caused by hyperactivity of
Stiff knee the rectus femoris during the swing phase. Another cause of SKG is the weakness of push-off in hemiparetic gait.
Gait Prior research did not consider the effect of the magnitude of knee extensors in their subjects.
Stroke
Research question: Does the cause of SKG differ between patients with high and low knee extensor activities
Kinematics
during the swing phase?
Electromyography
Botulinum toxin Methods: We examined 38 patients with chronic stroke hemiplegia who presented with SKG. After placing an
inertia sensor and an electromyogram, patients walked 10 m at a comfortable speed. All patients were catego­
rized per the sign of the principal component 2 (PC2) as a component with large factor loadings of knee extensors
attained from the electromyographic amplitude during the early swing phase of the paretic limb. Then, the ki­
nematic parameters of knee flexion and other gait parameters in each group were compared, and a correlation
analysis was performed.
Results: In the high PC2 group, the timing of peak knee flexion during the swing phase was early, and vastus
lateralis activity during the preswing phase negatively correlated with the knee-flexion angle during the swing
phase. In the low PC2 group, the angular velocity of ankle plantar flexion at the toe-off was slow, which posi­
tively correlated with the knee-flexion angle during the swing phase.
Significance: The cause of SKG could be an inappropriate activity of the vastus lateralis rather than the rectus
femoris in patients with high knee extensor activity and slow plantar-flexion velocity at toe-off in patients with
low knee extensor activity. Not all causes of SKG in patients with hemiplegia are common, and different treat­
ment strategies are needed per the individuality of spastic knee extensor activity.

1. Introduction
Stiff knee gait (SKG), characterized by reduced flexion of the knee
joints during the swing phase, is a common occurrence after stroke [1,
2], increasing the risk of falls by decreasing foot clearance [3] and
exacerbating energy inefficiencies through compensatory motions [4].
Hyperactivity of the rectus femoris related to disorders of muscle ac­
tivity control, such as stroke or Parkinson’s disease, causes excessive
knee extension moments during the swing phase, primarily observed
during the initial preswing phase and could continue during the early
and mid-swing phases [5]. Reportedly, the specific cause of SKG is the
inappropriate activity of the rectus femoris [5–7].
After the onset of stroke, upper motor-neuron lesions initiate disor­
dered sensorimotor control causing intermittent or sustained involun­
tary muscle activation [8]. Particularly, the anomalous activity of the
quadriceps during the swing phase decreases knee flexion velocity [9]
and an excessive knee extension moment [6]. The rectus femoris-reflex
excitability at toe-off is increased in stroke patients, negatively corre­
lating with the peak knee flexion angle during the swing phase [10].
Moreover, the amplified activity of the vastus lateralis owing to multi­
joint heteronymous excitatory reflexes induced by hip extension and
anomalous activity of the vastus intermedius contributes to SKG [11,
12]. Hence, SKG treatment focuses on decreasing quadriceps activity
during the swing phase. Indeed, significant literature is available on SKG

* Correspondence to: Graduate School of Health Science, Fukui Health Science University, 55-13-1 Egami, Fukui-city 910-3190, Fukui, Japan.
E-mail address: k.fujita@fukui-hsu.ac.jp (K. Fujita).

https://doi.org/10.1016/j.gaitpost.2022.09.078
Received 20 June 2022; Received in revised form 20 August 2022; Accepted 19 September 2022
Available online 21 September 2022
0966-6362/© 2022 Elsevier B.V. All rights reserved.
K. Fujita et al.
treatment by administering botulinum toxin (BoNT) to the rectus fem­
oris [1,3,13].
Reportedly, an increase in the knee flexion angle during the swing
phase due to BoNT administration into the rectus femoris is < 10◦ on
average [1,3,14]. Moreover, 45–80% of BoNT treatments to relieve SKG
could be inappropriate [15]; this limited recovery could depend on the
inadequacy in the treatable patients [15]. For example, the weakness of
push-off by the plantar flexor decreases knee flexion velocity, often
causing SKG rather than rectus femoris activity [14]. However, in this
report, statistical analysis of electromyographic (EMG) activity of the
quadriceps was not performed. Furthermore, in previous studies of the
causes and therapeutic effects of SKG, subject selection criteria were too
simple as they included only a decrease in knee flexion angle during the
swing phase and did not consider the effects of muscle activity behind
the subject [9–15].
When exploring the cause of SKG, the results could vary per the knee
extensor muscle activities. To date, no study has examined patients with
SKG classified by electromyographic (EMG) activity and compared gait
parameters between groups. Hence, this study elucidates the difference
in the causes of SKG between high and low knee extensor activities
during the swing phase in a hemiparetic gait, and possibly provides
crucial insights into selecting SKG treatments customized to patients’
gait characteristics.
2. Methods
2.1. Study design and participants
This cross-sectional study used baseline data from clinically regis­
tered randomized controlled trials (clinical trial no.: UMIN000034270).
We enrolled patients who had a stroke, aged 40–75 years, and hospi­
talized to treat the sequelae of stroke in the chronic phase and for
rehabilitation during September 2018–December 2021. The inclusion
criteria were: (i) unilateral cerebral lesions; (ii) at least 6 months since
the onset of stroke; (iii) walking independently or under supervision and
use/nonuse of a cane but not a highly fixed orthotic device; (iv) SKG
diagnosed under: peak knee flexion during the swing phase, knee range
of motion (ROM) during the early swing phase, total knee ROM, and
timing of peak knee flexion during the swing phase (based on the
following average [Standard Deviations; SD] values: 68.1◦ [4.2], 19.1◦
[3.1], 65.8◦ [4.3], and 72.7%GC [1.3], respectively) [6,16]. A limb was
classified as “stiff” if ≥ 3 of these measures were > 2 SD below the
average control value. Control data were obtained from 14 healthy
participants of approximately the same average age, height, and weight
as the patients (age, 60.5 [4.7] years; sex, 9 men and 5 women; height,
164.9 [6.6] cm; weight, 62.9 [8.7] kg; gait velocity, 1.25 [0.12] m/s).
The exclusion criteria were: (i) decreased lower-limb joint ROM at rest
(hip, inside 0–90◦ ; knee, inside 10–100◦ ; ankle, inside 0–30◦ ); (ii) his­
tory of lower-limb joint surgery; and (iii) Mini-Mental State Examination
score of < 24 points [16].
Of 45 participants, 7 were excluded for not fulfilling the SKG re­
quirements, finally enrolling 38 patients. The Ethical Review Committee
of Nittazuka Medical Welfare Center approved this study (approval no.
Nittazuka Ethics 2022–11). We obtained written informed consent from
all patients.
2.2. Experimental setup and protocol
A straight walkway (16 m) was built to measure walking distance
(including 3 m extra at each end). We set a video camera (HD Pro
Webcam; Logicool, Inc., Tokyo, Japan), with a 30 Hz sampling fre­
quency at 5 m lateral to the midpoint of the walkway[17]. Participants
walked at a comfortable speed, and regular cane users were allowed to
use it during the experiment. Patients with a high risk of falling because
of ankle inversion during the stance phase could use ankle–foot orthosis
(AFO); however, we used ORTOP® (Pacific Supply, Inc., Osaka, Japan),
188
Gait & Posture 98 (2022) 187–194
with a short, light, and flexible plastic body, to minimize their impact on
the results [18]. The same three examiners conducted all experiments.
Kinematics data during gait were recorded using MyoMotion (Nor­
axon Inc., Scottsdale, AZ). MyoMotion inertial sensors were placed per
the lower-limb rigid-body model with seven joint segments used in the
MR3 software (Noraxon Inc.) on shoes (top of the upper foot), frontal on
the shanks, frontal on the thighs, and bony area of the sacrum [16].
Using the upright position, calibration was done to determine the
0◦ angle value in the joints considered. Then, the examiners manually
fixed the participants’ knee joint to obtain the maximum extension
angle. At that time, the hip and ankle joint angles were adjusted to 0◦ ±
5◦ . We set the sampling frequency for the inertial sensors at 100 Hz.
EMG recording during gait was done using TELEmyo DTS (Noraxon
Inc.). The sampling frequency was 1500 Hz, and the bandpass filter was
set at 10–500 Hz. Then, EMG was performed for the following muscles
on the paretic side of the body: rectus femoris, vastus lateralis, biceps
femoris, tibialis anterior, medial head of the gastrocnemius, and soleus.
Next, skin impedance was decreased to ≤ 10 kΩ using alcohol-soaked
cotton swabs and an abrasive cream. Furthermore, Ag–AgCl electrodes
(EM-272; Noraxon Inc.) were placed on positions recommended by the
Surface ElectroMyoGraphy for the Non-Invasive Assessment of Muscle
project.
To identify the gait phase, we placed foot switches (Noraxon Inc.) on
the soles of both the feet (each foot = 4 points) [17]. MyoSync and Sync
Light devices (Noraxon Inc.) were used to synchronize all instruments
and ensure the alignment of time frames.
2.3. Data analysis
The kinematic and EMG waveforms were analyzed using MR3
(version 3.14). The analysis interval was three continuous gait cycles
around the middle portion of the walkway. The duration of each gait
cycle was normalized after considering one gait cycle to be 100%. We
obtained the arithmetic mean of the three gait cycles and calculated
1000-point amplitudes at intervals of 0.1%. Each gait phase (loading
response, single support, preswing, and swing phases) was identified
from the foot switch data on both sides. We defined the early swing
phase as the interval from the toe-off to the peak knee flexion angle
during the swing phase [19]; if the knee flexion waveform was bimodal,
the early swing phase was defined up to the first peak.
We calculated kinematic parameters, such as the angle and angular
velocity, of the knee, hip, and foot related to the swing phase [19]. All
EMG waveforms were full-wave rectified, and the mean amplitude of
each gait cycle (preswing and early swing) was evaluated; they were
normalized by the mean amplitude of the entire gait cycle [17,20]. The
step length was measured using the still images obtained from the video
data when patients passed through the intermediate points on the
walkway [17]. Moreover, we calculated cadence and gait cycle dura­
tions using the footswitch data, and gait velocity using the measured
stride length and cadence. We calculated the mean values obtained from
three walking trials for all data evaluated.
2.4. Statistical analysis
To extract those with high activity of knee extensors, we performed
the principal component analysis (PCA) using four electromyogram
average amplitudes during the early swing phase (rectus femoris, vastus
lateralis, biceps femoris, and gastrocnemius). For grouping, the prin­
cipal components (PC) with eigenvalues of ≥ 1 and high-factor loadings
of the rectus femoris and vastus lateralis were targeted. As the PC scores
are centered on having zero mean, the groups were extracted per PC
scores sign and categorized into high and low knee extensor activity
groups of knee extensors [21].
No missing data was reported. Next, between-group statistical dif­
ferences were analyzed using the independent-sample t-, Welch’s t-,
Mann–Whitney U, or χ2 tests, depending on the normality, variance, and
K. Fujita et al. Gait & Posture 98 (2022) 187–194

scale of data. Besides, Cohen’s d was used to assess the effect size (ES) Table 1
[22]. Moreover, the correlation coefficient between all four SKG [6,16] Patients’ characteristics.
and each gait parameter was calculated for both groups. All statistical High PC 2 (n Low PC 2 (n P
analyses were performed using BellCurve for Excel version 3.20 (Social = 17) = 21)
Survey Research Information Co., Ltd., Tokyo, Japan), with P < 0.05 Age 61.0 (2.4) 62.7 (2.3) NS
statistically significant. The sample size was determined using G*Power Height (cm) 167.8 (2.3) 167.0 (2.3) NS
3.1 (Heinrich-Heine-University, Düsseldorf, Germany) [23]. Weight (kg) 66.6 (2.4) 68.0 (2.4) NS
Sex (female/male) 3/14 2/19 NS
Type of stroke (CI/ICH) 2/15 7/14 NS
3. Results Months since onset 42.7 (8.6) 55.5 (10.3) NS
Paretic side (L/R) 5/12 4/17 NS
3.1. PCA Assistive device (None/T-cane/ 6/11/3 8/13/5 NS
AFO)
Fugl–Meyer 20.1 (1.2) 21.8 (0.8) NS
The PCA comprised four types of PCs of which two valid PCs with
assessment LE
eigenvalues > 1 were obtained. While PC1 had an eigenvalue of 1.57 Motricity index
(variance: 39.22%), PC2 had an eigenvalue of 1.23 (variance: 30.72%; Hip 25 (6) 25 (0) NS
Fig. 1). Based on the PC2 signs, patients were categorized into the high Knee 25 (8) 25 (8) NS
Ankle 19 (16) 14 (10) NS
(17 cases) and low PC2 groups (21 cases), because PC2 had large factor
Modified Ashworth
loadings of the rectus femoris and vastus lateralis as knee extensors scale
(0.65 and 0.61, respectively) and small factor loadings of the biceps Hip flexors 1 (1) 1 (1) NS
femoris and gastrocnemius as knee flexors (− 0.43 and − 0.50, respec­ Knee extensors 2 (2) 1 (2) NS
tively). Table 1 presents the patients’ characteristics. Ankle plantar 2 (2) 3 (2) NS
flexors

3.2. Comparisons between PC2 groups
Regarding the EMG amplitude, the rectus femoris (P = 0.036; ES =
0.797) and vastus lateralis (P < 0.001; ES = 1.440) during the early
swing phase in the high PC2 group were significantly higher than the
low PC2 group (Table 2). The gastrocnemius during the early swing
Values are expressed as mean (standard error) or as median (quartile deviation).
P: Comparison between the high PC2 and low PC2 groups (t-test, χ2 test, or
Mann–Whitney U test).
A score of 1 + in the Modified Ashworth scale was assigned as 2, and scores of ≥
2 were revised upward by 1.
PC, principal component; CI, cerebral infarction; ICH, intracerebral hemorrhage;
LE, lower extremity;
AFO, ankle–foot orthosis; NS, no significant difference.

Table 2
Comparisons of the electromyogram data between the high PC2 and low PC2
groups.
High PC2 Low PC2 P ES
(n = 17) (n = 21)

RF Preswing 106.7 (12.8) 101.9 (10.0) NS 0.101

Early swing 193.7 (32.1) 114.5 (15.0) 0.036 0.797
VL Preswing 45.7 (5.7) 48.8 (11.8) NS 0.076
Early swing 43.6 (5.1) 19.5 (2.9) < 0.001 1.440
BF Preswing 53.6 (5.9) 60.3 (8.9) NS 0.199
Early swing 33.1 (3.9) 41.3 (5.2) NS 0.408
TA Preswing 125.6 (10.9) 129.0 (8.2) NS 0.086
Early swing 190.8 (18.5) 144.6 (15.9) NS 0.639
MG Preswing 70.2 (5.0) 86.7 (11.4) NS 0.410
Early swing 38.5 (5.4) 74.3 (11.3) 0.008 0.892
Sol Preswing 87.1 (6.7) 75.1 (8.5) NS 0.358
Early swing 47.5 (7.9) 53.4 (8.5) NS 0.169

Values are mean amplitudes (%) during each gait phase (mean ± standard
error).
P: Comparison between the high PC2 and low PC2 groups (t-test or Welch’s t-
test).
RF, rectus femoris; VL, vastus lateralis; BF, biceps femoris; TA, tibialis anterior;
MG, medial gastrocnemius; Sol, soleus; NS, no significant difference; ES, effect
size.

Fig. 1. The principal component analysis (PCA) plot. Black rhombus, the PC
Regarding the kinematics parameters of walking, the timing of peak
load of the four muscles based on electromyography (EMG). Gray squares, the
PC scores of the individual subject’s EMG.
knee flexion during the early swing phase in the high PC2 group was
significantly earlier than in the low PC2 group (P = 0.049; ES = 0.684;
Table 4). The hip extension angle at toe-off in the high PC2 group was
phase in the high PC2 group was significantly smaller than the low PC2
significantly higher than in the low PC2 group (P = 0.008; ES = 0.935).
group (P = 0.008; ES = 0.892). Regarding the spatiotemporal parame­
Furthermore, the angular velocity of ankle plantar flexion at toe-off in
ters of walking, the stride (P = 0.047; ES = 0.689) and step lengths on
the low PC2 group was significantly slower than in the high PC2 group
the nonparetic side (P = 0.026; ES = 0.779) in the high PC2 group were
(P = 0.041; ES = 0.712).
significantly higher than the low PC2 group (Table 3).

189
K. Fujita et al. Gait & Posture 98 (2022) 187–194

Table 3 In the high PC2 group, the hip extension angle on the paretic side at
Comparisons of the spatiotemporal parameters between the high PC2 and low the toe-off were large, indicating that the rectus femoris on the paretic
PC2 groups. side was more stretched. Stretching the rectus femoris with hip exten­
High PC2 Low PC2 P ES sion generates a sustained spastic knee extensor activity [11]. Indeed,
(n = 17) (n = 21) the rectus femoris activity was high in the high PC2 group, although the
Gait velocity (m/s) 0.60 (0.06) 0.50 (0.05) NS 0.480 amplitude of the rectus femoris did not correlate with the peak knee
Cadence (steps/ 84.9 (4.2) 85.7 (3.0) NS 0.056 flexion angle during the swing phase. Reportedly, the rectus femoris
min) reflex excitability limited knee flexion during the swing phase [10].
LR period (%GC) 13.9 (0.9) 16.6 (1.2) NS 0.574
Compared with healthy subjects, the rectus femoris reflex excitability
SS period (%GC) 28.1 (1.1) 27.2 (1.1) NS 0.188
PSw period (%GC) 16.3 (1.0) 17.4 (1.0) NS 0.285 increased by hip extension in stroke patients, with an angular velocity of
Sw period (%GC) 41.8 (0.9) 39.5 (1.0) NS 0.567 ≥ 120◦ /s [11]. However, the hip extension velocity at the same walking
Step length (cm) 43.3 (2.1) 39.6 (1.9) NS 0.436 speed as our patients was ≤ 40◦ /s [26]. Possibly, many of our subjects
(Ps) had slow the hip extension velocity and thus did not reach the threshold
Step length (cm) 44.2 (2.4) 34.8 (3.0) 0.026 0.779
(NPs)
for inducing reflex excitability. Additionally, despite the insufficient
Stride length (cm) 87.4 (4.1) 74.4 (4.6) 0.047 0.689 evidence for a relationship between hip extension angle and rectus
femoris extension during walking, it is possible that the rectus femoris
Values represent the mean ± standard error.
muscle was not stretched at approximately 5 degrees of the hip exten­
P: Comparison between the high PC2 and low PC2 groups (t-test or Welch’s t-
test).
sion. Therefore, an increase of the rectus femoris activity in the high PC2
LR, loading response; SS, single support; PSw, preswing; Sw, swing; DS, double group might play a role as an EMG module of propulsion of the swing leg
support; [24,25] rather than reflex excitability. Indeed, as the hip flexion velocity
Ps, paretic side; NPs, nonparetic side; GC, gait cycle; NS, no significant differ­ positively correlated with the knee flexion angle during the swing phase,
ence; ES, effect size. the rectus femoris, which has a hip flexion act, might not be the main
cause of knee flexion limitation.
3.3. Correlation analysis Nevertheless, the vastus lateralis activity in the high PC2 group
negatively correlated with the peak knee flexion angle and knee ROM at
Table 5 presents the correlation coefficient of the gait parameters the gait cycle. The vastus lateralis is paramount as a component of the
significantly correlating with any of the four SKG parameters [6]. The EMG module for weight acceptance at the early stance phase [24]. The
kinematic parameters that differed between the groups demonstrated a loss of descending drive from corticospinal pathways to the interneurons
correlation between the knee ROM during the early swing phase and the and motoneuron pools that generate the extensor phase modules might
hip extension angle at toe-off (high PC2 group, r = 0.730; low PC2 render the propulsion modules no longer independent from the weight
group, r = − 0.072), and the knee ROM during the entire gait cycle and acceptance module [25]. Moreover, among the quadriceps, the vastus
the angular velocity of ankle plantar flexion at toe-off (high PC2 group, lateralis and rectus femoris contribute the most to the knee extension
r = 0.080; low PC2 group, r = 0.462). moment [27], and the cross-sectional area of the vastus lateralis is the
Regarding the EMG amplitude, the vastus lateralis during the pre­ largest of the quadriceps in healthy and stroke individuals [28].
swing or early swing phases, the biceps femoris during the preswing Therefore, the discharge of the vastus lateralis at the preswing phase to
phase, and the tibialis anterior or soleus during the early swing phase the early swing phase might produce a high knee extension moment. In
negatively correlated with any of the SKG parameters in the high PC2 this study, inappropriate timing activity of the vastus lateralis of the
group (r = − 0.448 to − 0.589). In the low PC2 group, the rectus femoris, high PC2 group lead to changes in the early timing of peak knee flexion
vastus lateralis, gastrocnemius, and soleus during the early swing phase during the swing phase, which could have caused SKG rather than rectus
positively correlated with any SKG parameter (r = 0.475－0.564). femoris activity. Several previous studies have attributed rectus femoris
overactivity to the cause of SKG [1–3,5–7]. However, Merlo et al. sug­
4. Discussion gested that numerous BoNT treatments for the rectus femoris to reduce
SKG may be inappropriate [15], with the results of this current study
We performed the PCA on the EMG activity of knee extensors during supporting their report. In individuals with high knee extensor activity
the early swing phase to differentiate between high and low knee during the early swing, the vastus lateralis may be a better target for SKG
extensor activity. This analysis was adopted because two muscles, the treatment. However, abnormal vastus intermedius activity also con­
rectus femoris and vastus lateralis, were included as indicators of knee tributes to SKG development [12]; therefore, additional studies
extensor activity and defining an amplitude magnitude boundary value involving knee extensors other than the rectus femoris and vastus lat­
for grouping was difficult. As PC1 has large factor loadings on the biceps eralis are needed.
femoris and gastrocnemius, it could also act on knee flexion. PC2 In the low PC2 group, the activities of the rectus femoris and vastus
demonstrates superior knee extensor activity because only the rectus lateralis during the early swing phase were low; however, the peak knee
femoris and vastus lateralis have large factor loadings. Thus, we grouped flexion angle did not markedly differ from that in the high PC2 group.
patients based on the PC2 factor loadings. The activities of the rectus Moreover, the activities of the rectus femoris and vastus lateralis in the
femoris and vastus lateralis were higher in the high PC2 group than in low PC2 group did not significantly negatively correlate with the peak
the low PC2 group, establishing that PC2 exhibits overactivity of knee knee flexion and knee ROM during the swing phase. Hence, knee flexion
extensors. In the walking muscle activity pattern of healthy persons, the during the swing phase in the low PC2 group was limited, notwith­
rectus femoris activity increased during the early swing phase; however, standing the magnitude of quadriceps activity. Particularly, the planter-
the EMG component that increases the activity of both the rectus femoris flexion velocity at toe-off in the low PC2 group was slower than in the
and vastus lateralis during the early swing phase was not reported [24]. high PC2 group. Furthermore, a positive correlation was noted between
Besides, no EMG module increases both the rectus femoris and vastus the planter-flexion velocity at toe-off and the peak knee flexion angle
medialis during the early swing phase in stroke patients [25]. As all our during the swing phase in the low PC2 group. As the activity of the
patients presented with SKG, the PC2 observed could be specific to pa­ plantar flexor contributes to the increased knee flexion velocity during
tients with SKG. Perhaps, this component causes knee extension at an the preswing phase [29], and the peak of malleolus vertical acceleration
inappropriate timing because of the high activity of the vastus lateralis during the preswing phase positively correlates with knee flexion ve­
despite the early swing phase. locity [14], the lack of push-off could be involved in SKG development.
Owing to a strong positive correlation between plantar-flexion velocity

190
 K. Fujita et al.
Table 4
Comparisons of the kinematics parameters between the high PC2 and low PC2 groups.
High PC2 (n = 17) Low PC2 (n = 21) P ES

Knee Peak extension angle at stance (◦ ) 0.4 (1.8) 1.8 (1.7) NS 0.182
Flexion angle at toe-off (◦ ) 27.4 (2.1) 27.5 (2.1) NS 0.009
Peak flexion angle at ESw (◦ ) 34.1 (2.6) 31.9 (2.4) NS 0.207
Range of motion during GC (◦ ) 34.5 (3.0) 33.6 (2.6) NS 0.074
Range of motion during ESw (◦ ) 6.7 (1.8) 4.4 (0.8) NS 0.207
Timing of peak flexion at ESw (%GC) 63.9 (1.3) 68.2 (1.6) 0.049 0.684
Flexion velocity at toe-off (◦ /s) 143.0 (19.2) 116.7 (16.1) NS 0.354
Peak flexion velocity during PSw to ESw (◦ /s) 180.8 (20.9) 145.1 (18.5) NS 0.430
Hip Peak extension angle during stance (◦ ) 7.7 (1.5) 5.1 (0.8) NS 0.559
Extension angle at toe-off (◦ ) − 0.9 (1.3) − 5.3 (0.9) 0.008 0.935
Flexion velocity at toe-off (◦ /s) 65.4 (11.1) 66.8 (10.8) NS 0.031
Peak flexion velocity during PSw to ESw (◦ /s) 91.0 (11.0) 90.0 (11.2) NS 0.021
191

Ankle Plantar-flexion angle at toe-off (◦ ) 3.6 (1.8) 4.2 (1.6) NS 0.081

Plantar-flexion velocity at toe-off (◦ /s) 81.8 (11.0) 46.5 (12.0) 0.041 0.712
Peak plantar-flexion velocity during PSw to ESw (◦ /s) 113.2 (13.9) 104.4 (14.4) NS 0.145

Values represent the mean ± standard error.

P: Comparison between the high PC2 and low PC2 groups (t-test or Welch’s t-test).
PSw, preswing; ESw, early swing; GC, gait cycle; NS, no significant difference; ES, effect size.

Gait & Posture 98 (2022) 187–194

 K. Fujita et al.
Table 5
Correlation coefficient with the four parameters that determine SKG of each group.
Peak flexion angle at ESw Range of motion during GC Range of motion during ESw Timing of peak flexion at ESw

H L H L H L H L

Knee flexion velocity at toe-off 0.781 * 0.791 0.871 * 0.783 * 0.668 * 0.743 * 0.107 − 0.105
Hip flexion velocity at toe-off 0.793 * 0.772 * 0.765 * 0.738 * 0.367 0.625 * 0.102 − 0.134
Peak hip extension angle during stance 0.482 * 0.557 * 0.846 * 0.465 0.569 * 0.168 0.183 − 0.343
Hip extension angle at toe-off 0.168 − 0.151 0.443 − 0.424 0.730 * − 0.072 0.128 0.011
Ankle plantar-flexion velocity at toe-off 0.301 0.504 * 0.080 0.462 * − 0.185 0.301 − 0.242 − 0.144
EMG amplitude (%)
Rectus femoris during ESw 0.221 0.510 * 0.110 0.209 0.315 0.215 − 0.448 − 0.248
192

− − −
Vastus lateralis during PSw − 0.587 * − 0.183 − 0.529 * − 0.163 − 0.300 − 0.058 0.175 0.230
Vastus lateralis during ESw − 0.404 0.475 * − 0.371 0.564 * − 0.185 0.182 − 0.448 * − 0.177
Biceps femoris during PSw − 0.487 * − 0.079 − 0.310 − 0.085 − 0.431 0.032 0.038 0.233
Tibialis anterior during ESw − 0.109 0.291 − 0.147 − 0.078 − 0.203 0.137 − 0.498 * − 0.082
Gastrocnemius during ESw − 0.341 0.537 * − 0.246 0.369 − 0.228 0.420 0.010 − 0.079
Soleus during ESw − 0.497 * 0.446 * − 0.589 * 0.357 − 0.484 * 0.303 0.131 − 0.011

Only items that have a significant correlation (*P < 0.05) with any of the SKG parameters are displayed.
SKG, stiff knee gait; H, high PC2 group; L, low PC2 group; EMG, electromyogram; PSw, preswing; ESw, early swing; GC, gait cycle.

Gait & Posture 98 (2022) 187–194

K. Fujita et al.
and plantar-flexion power at push-off [30], the plantar-flexion velocity
at toe-off in this study might reflect the push-off power. Notably, the
push-off power of the low PC2 group cannot be estimated directly;
however, owing to the deep involvement of the kinematic chain of the
ankle and knee in the push-off [31], the decreased plantar-flexion ve­
locity could correlate with SKG development, especially in the low PC2
group.
This study has some limitations: First, as all patients were in the
chronic phase, the cause of SKG might differ in those functionally in the
recovery stage. Second, the insufficient sample size hindered a multi­
variate analysis. Hence, additional research is warranted to investigate
the causes that limit knee flexion from multiple factors for each group.
However, the stricter patient selection criteria in this study helped us to
extract data more reliably from the SKG population than in previous
studies. In contrast, the strict selection criteria of subjects and their
allocation to the two groups may have made the distribution range of the
correlation analysis narrower and less likely to show significance. Third,
63.2% of all subjects used a T-cane, whereas 21.1% used an AFO during
the experiment. The use of the T-cane increased walking speed, but did
not alter the angular kinematic profile of chronic stroke survivors [32].
Although no previous studies have investigated the effect of the AFO
used in this study on angular kinematic profiles, this AFO is lightweight
and highly flexible. Furthermore, no significant difference in the number
of assistive device users was observed between the groups. Therefore,
the effect of the assistive device used in this study on the statistical re­
sults is considered small.
5. Conclusions
This study suggests that the cause of SKG varies according to the knee
extensors activity during the early swing phase. In individuals with high
knee extensor activity, quadriceps activity causes SKG, although the
effect of the inappropriate vastus lateralis activity is higher than the
rectus femoris. In individuals with low knee extensor activity, quadri­
ceps activity exerts a reduced effect, while a decline in plantar-flexion
velocity at toe-off correlates with SKG development. Overall, the cau­
ses of SKG in hemiplegic patients are not all common, and different
treatment strategies are needed per the individuality of spastic knee
extensor activity.
CRediT authorship contribution statement
K.F.: Conceptualization, Methodology, Formal analysis, Investiga­
tion, Resources, Data curation, Writing – original draft, Visualization,
Supervision, Project administration, Funding acquisition, Supervision.
Y.T.: Conceptualization, Methodology, Investigation, Resources, Data
curation, Project administration. K.H.: Methodology, Formal analysis,
Writing – review & editing, Visualization. K.K.: Investigation, Writing –
review & editing, Project administration. M.S.: Methodology, Re­
sources, Data curation.Y.K.: Formal analysis, Resources, Writing – re­
view & editing, Supervision.
Acknowledgements
This work was supported by Grants-in-Aid for Scientific Research in
the Japan Society for the Promotion of Science (grant number:
18K17772).
Conflict of interest
The authors have no conflict of interest to declare.
References
[1] J. Boudarham, S. Hameau, D. Pradon, D. Bensmail, N. Roche, R. Zory, Changes in
electromyographic activity after botulinum toxin injection of the rectus femoris in
193
Gait & Posture 98 (2022) 187–194
patients with hemiparesis walking with a stiff-knee gait, J. Electromyogr. Kinesiol.
23 (2013) 1036–1043, https://doi.org/10.1016/j.jelekin.2013.07.002.
[2] D.C. Kerrigan, M.E. Karvosky, P.O. Riley, Spastic paretic stiff-legged gait: joint
kinetics, Am. J. Phys. Med. Rehabil. 80 (2001) 244–249, https://doi.org/10.1097/
00002060-200104000-00002.
[3] G.G. Stoquart, C. Detrembleur, S. Palumbo, T. Deltombe, T.M. Lejeune, Effect of　
botulinum toxin injection in the rectus femoris on stiff-knee　gait in people with
stroke: a prospective observational　study, Arch. Phys. Med. Rehabil. 89 (2008)
56–61, https://doi.org/10.1016/j.apmr.2007.08.131.
[4] J. Doke, J.M. Donelan, A.D. Kuo, Mechanics and energetics of swinging the human
leg, J. Exp. Biol. 208 (208) (2005) 439–445, https://doi.org/10.1242/jeb.01408.
[5] J. Boudarham, N. Roche, D. Pradon, E. Delouf, D. Bensmail, R. Zory, Effects of
quadriceps muscle fatigue on stiff-knee gait in patients with hemiparesis, PLoS One
9 (2014), e94138, https://doi.org/10.1371/journal.pone.0094138.
[6] S.R. Goldberg, S. Ounpuu, A.S. Arnold, J.R. Gage, S.L. Delp, Kinematic and kinetic
factors that correlate with improved knee flexion following treatment for stiff-knee
gait, J. Biomech. 39 (2006) 689–698, https://doi.org/10.1016/j.
jbiomech.2005.01.015.
[7] S.R. Goldberg, S. Õunpuu, S.L. Delp, The importance of swing-phase initial
conditions in stiff-knee gait, J. Biomech. 36 (2003) 1111–1116, https://doi.org/
10.1016/s0021-9290(03)00106-4.
[8] A.D. Pandyan, M. Gregoric, M.P. Barnes, D. Wood, F. Van Wijck, J. Burridge, et al.,
Spasticity: clinical perceptions, neurological realities and meaningful
measurement, Disabil. Rehabil. 27 (2005) 2–6, https://doi.org/10.1080/
09638280400014576.
[9] S.J. Piazza, S.L. Delp, The influence of muscles on knee flexion during the swing
phase of gait, J. Biomech. 29 (1996) 723–733, https://doi.org/10.1016/0021-9290
(95)00144-1.
[10] T. Akbas, K. Kim, K. Doyle, K. Manella, R. Lee, P. Spicer, et al., Rectus femoris
hyperreflexia contributes to Stiff-Knee gait after stroke, J. Neuroeng. Rehabil. 17
(2020) 117, https://doi.org/10.1186/s12984-020-00724-z.
[11] M.D. Lewek, T.G. Hornby, Y.Y. Dhaher, B.D. Schmit, Prolonged quadriceps activity
following imposed hip extension: neurophysiological mechanism for stiff-knee gait,
J. Neurophysiol. 98 (2007) 3153–3162, https://doi.org/10.1152/jn.00726.2007.
[12] D. Mazzoli, E. Giannotti, M. Manca, M. Longhi, P. Prati, M. Cosma, et al.,
Electromyographic activity of the vastus intermedius muscle in patients with stiff-
knee gait after stroke. A retrospective observational study, Gait Posture 60 (2018)
273–278, https://doi.org/10.1016/j.gaitpost.2017.07.002.
[13] N. Roche, R. Zory, A. Sauthier, C. Bonnyaud, D. Pradon, D. Bensmail, Effect of
rehabilitation and botulinum toxin injection on gait in chronic stroke patients: a
randomized controlled study, J. Rehabil. Med. 47 (2015) 31–37, https://doi.org/
10.2340/16501977-1887.
[14] I. Campanini, A. Merlo, B. Damiano, A method to differentiate the causes of stiff-
knee gait in stroke patients, Gait Posture 38 (2013) 165–169, https://doi.org/
10.1016/j.gaitpost.2013.05.003.
[15] A. Merlo, I. Campanini, Impact of instrumental analysis of stiff knee gait on
treatment appropriateness and associated costs in stroke patients, Gait Posture 72
(2019) 195–201, https://doi.org/10.1016/j.gaitpost.2019.06.009.
[16] K. Fujita, Y. Kobayashi, H. Miaki, H. Hori, Y. Tsushima, R. Sakai, et al., Pedaling
improves gait ability of hemiparetic patients with stiff-knee gait: fall prevention
during gait, J. Stroke Cerebrovasc. Dis. 29 (2020), 105035, https://doi.org/
10.1016/j.jstrokecerebrovasdis.2020.105035.
[17] K. Fujita, H. Miaki, H. Hori, Y. Kobayashi, T. Nakagawa, How effective is physical
therapy for gait muscle activity in hemiparetic patients who receive botulinum
toxin injections? Eur. J. Phys. Rehabil. Med. 55 (2019) 8–18, https://doi.org/
10.23736/S1973-9087.18.05168-7.
[18] Pacific Supply Co., Ltd [Web]. Available online: https://www.p-supply.co.jp/
products/index.php?act=detail&pid=9 (accessed on 6 Jan 2021).
[19] N.E. Akalan, S. Kuchimov, A. Apti, Y. Temelli, A. Nene, Contributors of stiff knee
gait pattern for able bodies: hip and knee velocity reduction and tiptoe gait, Gait
Posture 43 (2016) 176–181, https://doi.org/10.1016/j.gaitpost.2015.09.019.
[20] K. Fujita, H. Miaki, A. Fujimoto, H. Hori, H. Fujimoto, Y. Kobayashi, Factors
affecting premature plantar flexor muscle activity during hemiparetic gait,
J. Electromyogr. Kinesiol 39 (2018) 99–103, https://doi.org/10.1016/j.
jelekin.2018.02.006.
[21] N. Gaudreault, N. Mezghani, K. Turcot, N. Hagemeister, K. Boivin, et al., Effects of
physiotherapy treatment on knee osteoarthritis gait data using principal
component analysis, Clin. Biomech. 26 (2011) 284–291, https://doi.org/10.1016/
j.clinbiomech.2010.10.004.
[22] J. Cohen, A power primer, Psychol. Bull. 112 (1992) 155–159, https://doi.org/
10.1037//0033-2909.112.1.155.
[23] F. Faul, E. Erdfelder, A.G. Lang, A. Buchner, G* Power 3: a flexible statistical power
analysis program for the social, behavioral, and biomedical sciences, Behav. Res.
Methods 39 (2007) 175–191, https://doi.org/10.3758/bf03193146.
[24] Y.P. Ivanenko, R.E. Poppele, F. Lacquaniti, Five basic muscle activation patterns
account for muscle activity during human locomotion, J. Physiol. 556 (2004)
267–282, https://doi.org/10.1113/jphysiol.2003.057174.
[25] D.J. Clark, L.H. Ting, F.E. Zajac, R.R. Neptune, S.A. Kautz, Merging of healthy
motor modules predicts reduced locomotor performance and muscle coordination
complexity post-stroke, J. Neurophysiol. 103 (2010) 844–857, https://doi.org/
10.1152/jn.00825.2009.
[26] B.F. Mentiplay, M. Banky, R.A. Clark, M.B. Kahn, G. Williams, Lower limb angular
velocity during walking at various speeds, Gait Posture 65 (2018) 190–196,
https://doi.org/10.1016/j.gaitpost.2018.06.162.
K. Fujita et al.
[27] L.Q. Zhang, G. Wang, G.W. Nuber, J.M. Press, J.L. Koh, In vivo load sharing among
the quadriceps components, J. Orthop. Res. 21 (2003) 565–571, https://doi.org/
10.1016/S0736-0266(02)00196-1.
[28] A. Silva-Couto Mde, C.L. Prado-Medeiros, A.B. Oliveira, C.C. Alcântara, A.
T. Guimarães, F. Salvini Tde, et al., Muscle atrophy, voluntary activation
disturbances, and low serum concentrations of IGF-1 and IGFBP-3 are associated
with weakness in people with chronic stroke, Phys. Ther. 94 (2014) 957–967,
https://doi.org/10.2522/ptj.20130322.
[29] S.R. Goldberg, F.C. Anderson, M.G. Pandy, S.L. Delp, Muscles that influence knee
flexion velocity in double support: implications for stiff-knee gait, J. Biomech. 37
(2004) 1189–1196, https://doi.org/10.1016/j.jbiomech.2003.12.005.
194
Gait & Posture 98 (2022) 187–194
[30] M.G. Browne, J.R. Franz, Ankle power biofeedback attenuates the distal-to-
proximal redistribution in older adults, Gait Posture 71 (2019) 44–49, https://doi.
org/10.1016/j.gaitpost.2019.04.011.
[31] S.O. Schaarup, E. Wetke, L.A.G. Konradsen, J.D.F. Calder, Loss of the knee-ankle
coupling and unrecognized elongation in Achilles tendon rupture: effects of
differential elongation of the gastrocnemius tendon, Knee Surg. Sports Traumatol.
Arthrosc. 29 (2021) 2535–2544, https://doi.org/10.1007/s00167-021-06580-1.
[32] J.C. Polese, L.F. Teixeira-Salmela, L.R. Nascimento, C.D.M. Faria, R.N. Kirkwood,
G.C. Laurentino, et al., The effects of walking sticks on gait kinematics and kinetics
with chronic stroke survivors, Clin. Biomech. 27 (2012) 131–137, https://doi.org/
10.1016/j.clinbiomech.2011.08.003.