This article appeared in a journal published by Elsevier.

The attached
copy is furnished to the author for internal non-commercial research
and education use, including for instruction at the authors institution
and sharing with colleagues.
Other uses, including reproduction and distribution, or selling or
licensing copies, or posting to personal, institutional or third party
websites are prohibited.
In most cases authors are permitted to post their version of the
article (e.g. in Word or Tex form) to their personal website or
institutional repository. Authors requiring further information
regarding Elsevier’s archiving and manuscript policies are
encouraged to visit:
http://www.elsevier.com/copyright
 Author's personal copy

Available online at www.sciencedirect.com

Clinical Biomechanics 23 (2008) 769–778

www.elsevier.com/locate/clinbiomech

Effects of cadence on energy generation and absorption at

lower extremity joints during gait
Luci Fuscaldi Teixeira-Salmela a,*, Sylvie Nadeau b, Marie-Hélène Milot b,
Denis Gravel b, Luis Fernando Requião b
a
Department of Physical Therapy, Universidade Federal de Minas Gerais, Avenida Antônio Carlos, 6627, Campus Pampulha,
31270-010 Belo Horizonte, Minas Gerais, Brazil
b
École de réadaptation, Université de Montréal and Centre de recherche Interdisciplinaire en Réadaptation (CRIR),
Institut de réadaptation de Montréal, Canada

Received 31 July 2007; accepted 15 February 2008

Abstract

Background. Information regarding kinetic changes associated with walking speed is important for identifying alterations in locomo-
tor disorders caused by pathological processes, as opposed to those arising solely from altered speeds.
Methods. Fourteen healthy subjects were assessed walking at both natural and imposed cadences of 60, 80, and 120 steps/min. A 3D
motion analysis system, force platforms, and related software were used to obtain kinematic and kinetic data. Net joint powers were
calculated across cycles and the area under the positive and negative phases of the power curves provided the mechanical work generated
and absorbed at the hip, knee, and ankle. The relative contributions to the total positive and negative work across the four cadences were
calculated for each joint. ANOVAs followed by planned contrasts were used to assess the effects of laterality, joint, and cadence.
Findings. Power and mechanical work, as well as the contributions of individual joints to the total energy generated and absorbed,
were shown to be influenced by walking cadence, independent of laterality. The ankle, knee, and hip contributions to the total limb gen-
eration and absorption at the lowest cadence were 53%, 21%, and 26%, and at the highest cadence, the corresponding values were 34%,
33%, and 33%, respectively.
Interpretation. Power and mechanical work, as well as the contributions of individual joints to the total energy generated and
absorbed, were shown to be influenced by the walking cadence, independent of laterality. These findings will be helpful for identifying
walking strategies and adaptations in populations with gait disorders.
Ó 2008 Elsevier Ltd. All rights reserved.

Keywords: Gait analysis; Biomechanics; Mechanical work; Energy; Walking speed

1. Introduction tends to increase at faster walking speeds, resulting in a lar-

It is well known in the study of normal human gait, that
muscles are the main sources of energy generation and
absorption. When walking, the ability to adjust the speed
is an important mechanism that requires different levels
of muscular activities for appropriate adaptations to
changes in the task demands. In general, muscle activity
*
Corresponding author.
E-mail address: lfts@ufmg.br (L.F. Teixeira-Salmela).
ger muscular force output (Den Otter et al., 2004). One of
the main roles of the muscles is the control of the magni-
tude and duration of acceleration and deceleration of indi-
vidual body segments to permit safe forward progression
(Bishop et al., 2004; Neptune et al., 2004).
Self-selected walking speed is a well-known indicator of
overall gait performance and is commonly used to assess
locomotor ability. However, gait speed alone does not con-
tribute substantially to the understanding of the nature of
gait deficiencies nor to guide intervention protocols. By
the same token, kinematic analyses yield little information

0268-0033/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.clinbiomech.2008.02.007
 Author's personal copy
770 L.F. Teixeira-Salmela et al. / Clinical Biomechanics 23 (2008) 769–778
about the mechanisms underlying normal and abnormal
movement patterns. On the other hand, kinetic analyses
provide a better understanding of normal motor patterns
and result in new directions for diagnosing the causes of
abnormal motor patterns observed in pathological gait
profiles (Olney et al., 1991, 1994).
Several kinetic parameters have been proposed to evalu-
ate gait. The moment of force parameter provides insights
about the predominant muscle groups at a given joint and
about the level of the mechanical effort that is produced
when relatively expressed to the maximal force capabilities
(Milot et al., 2006; Requião et al., 2005). The assessment of
the joint power parameter is more sensitive to instanta-
neous muscle effects to generate or absorb energy, but does
not take into account the duration of force application.
Work parameters, on the other hand, integrate the power
curve over time and provide, in some cases, additional
meaningful measures (Winter, 1991) to understand normal
(Chen et al., 1997; Siegel et al., 2004; Winter, 1983a) and
abnormal walking patterns (Mansour et al., 1982; McGib-
bon et al., 2001; Olney et al., 1991, 1994; Teixeira-Salmela
et al., 2001). Positive work performed during concentric
contractions, results in energy flow from muscles to the seg-
ments to accelerate the body, whereas negative work, orig-
inating from eccentric contractions, produces energy flow
from the segments to the muscles to decelerate the body.
The interplay between these energy sources facilitates suc-
cessful progression during walking (Chen et al., 1997; Win-
ter, 1983).
When walking on level surfaces, the ankle plantarflex-
ors, hip flexors, and hip extensors are the main muscle
groups that contribute to energy generation in the sagittal
plane (Chen et al., 1997; Olney et al., 1994; Winter, 1983,
1991). Previous studies have shown high correlations
between positive power bursts of these muscle groups and
gait velocity or cadence both with normal (Sadeghi et al.,
2001), low-performing elderly (Graf et al., 2005), and path-
ological subjects (Mueller et al., 1994a, 1994b; Olney and
Richards, 1996; Teixeira-Salmela et al., 2001). The knee
joint muscles act mainly eccentrically and the rate of
absorption of energy is greater with increases in walking
speed (Olney et al., 1994; Teixeira-Salmela et al., 2001;
Winter, 1983), while negative work at the hip and ankle
also results in increases at higher walking speeds.
Because decreasing gait speed or cadence is commonly
observed in locomotor disorders, information on kinetic
changes with gait speed or cadence in normal subjects is
important to pinpoint specific modifications related to the
pathology. The study of variations in positive and negative
energy contributions would provide baseline data and yield
information that would be useful in understanding and
treating gait deviations. Moreover, the interaction between
different muscle groups in generating and absorbing power
is relevant for the understanding the compensatory mecha-
nisms in pathological conditions. For example, patients
with diabetic peripheral neuropathy were able to attenuate
the decreased push-off by emphasizing hip flexor muscles
(Mueller et al., 1994a, 1994b), whereas higher power gener-
ation by the hip extensors in early stance was a typical, spe-
cific adaptation of above (Seroussi et al., 1996) and below-
knee amputees (Gitter et al., 1991; Hermodsson et al.,
1994).
In this context, the relative contributions of each muscle
group to the total energy generation or absorption should
be studied. In normal children, Chen et al. (1997) investi-
gated the influences of walking speed on mechanical joint
power during gait and demonstrated that the relative con-
tributions to mechanical work of muscles about the knee
and hip were positively correlated with speed, while nega-
tively correlated with ankle muscles. Nadeau et al. (2001)
investigated the relative contributions to energy generation
for stroke and healthy subjects walking at similar speeds
and observed that for the former, they were greater when
originated from the hip flexors, whereas for the later, the
plantarflexors predominated. Thus, the estimation of the
relative contribution of each joint to the total energy gen-
erated and absorbed during gait at different speeds seems
to be a useful approach to understand the nature, extent,
and degree of compensation across joints and to suggest
more efficient methods of correction. Moreover, informa-
tion which is related to kinetic changes to walking speed
is important for identifying alterations in locomotor disor-
ders caused by pathological processes, as opposed to those
arising solely from altered speeds.
Therefore, the purpose of the present study was to
employ an analytical model to estimate the effects of walk-
ing cadence and laterality on the positive and negative
mechanical work performed by the hip, knee, and ankle
muscles in the sagittal plane. According to Winter
(1983b), cadence does not affect the energy generation
and absorption pattern shapes over a stride and it is largely
controlled by the mechanical power generation and
absorption. Therefore, it was decided to use cadence as
the controlling parameter in the present study.
2. Methods
2.1. Participants
Fourteen healthy volunteers participated in the study,
recruited from the community, who had no previous his-
tory of lower limb injuries and showed no obvious gait
abnormalities. All participants provided consent prior to
their evaluation, based on ethical approval from the local
research review board. Data related to the mechanical level
of effort during gait at different cadences have been pub-
lished in a previous study for this group of participants
(Requião et al., 2005) and used as comparative data in
another study (Milot et al., 2006).
2.2. Gait assessment
Participants walked with their own low-heeled shoes
along a nine-m walkway over three embedded force plat-
 Author's personal copy

L.F. Teixeira-Salmela et al. / Clinical Biomechanics 23 (2008) 769–778 771

forms of standardized dimensions. First, subjects were approach (Winter, 1991) performed with Kingait3 software
required to walk at their natural cadence, followed by three (Mishac Kinetics, Waterloo, Canada) was used to estimate
other imposed cadences of 60, 80 and 120 steps/min, con- the net moments at the ankle, knee and hip joints.
trolled by a metronome. Five walking trials for each side Net joint powers (P) were calculated for each instant in
were collected at the four cadences, for a total of 40 trials. time as the product of net moments across the joint (M)
A successful trial was defined as one in which the subject and the relative angular velocity between the adjacent limb
appropriately contacted the force platform, i.e., the subject segments (x). When the product (M * x) was positive, the
hit one force platform with one foot, keeping the metro- muscles generated energy by contracting concentrically and
nomic rhythm. when the product was negative, the muscles absorbed
A three-dimensional (3D) Optotrak motion analysis sys- energy by contracting eccentrically. The power curves at
tem, three AMTI (Advanced Medical Technologies, New- the hip, knee, and ankle were analyzed on the basis of all
ton, MA, USA) force platforms, and related software cadences and all phases of power generation and absorp-
were used to obtain kinematic and kinetic data. The accu- tion were identified. The area under the positive phase of
racy for the Optotrak system and for the location of the the power curve provided the mechanical work generated
center of pressure of force platforms was 1 mm and 3 mm (Wpositive; Eq. (1)) at the hip, knee, and ankle, whereas
(RMS mean values), respectively. the integral of the negative phase (Wnegative; Eq. (2)), pro-
Prior to data collection, infrared markers were placed vided the absorbed work (Winter, 1983)
bilaterally on the following locations: foot (lateral heel, Z tf
dorsum, fifth metatarsal head), leg (lateral malleolus, mid W positive ¼ P dt for P > 0 ð1Þ
J
shank, and head of fibula), thigh (greater trochanter, ti
Z tf
mid-thigh, and lateral femoral condyle), pelvis (left and
right posterior superior iliac spine, left iliac crest), trunk W negative
J ¼ P dt for P < 0 ð2Þ
ti
(L3, T12 spinous process, right and left at the T8 level, C7
spinous process, and right and left acromium), and head
where J refers to the joint (ankle, knee or hip), P to the
(right and left occipital). In addition, 13 specific anatomical
joint power, and ti and tf to the initial and final gait cycle
points on the feet (heel posterior point, tip of foot), shanks
times, respectively. For all cadences, the energy generated
(medial malleoli), thighs (medial femoral condyles), pelvis
and absorbed at the ankle, knee, and hip was computed
(right and left ASIS, and iliac crests) and trunk (left gleno-
without consideration to the muscle groups involved at a
humeral joint) were digitized with a probe to define the
given joint.
local axis of each segment. Anthropometric measurements,
In addition, for each cadence, the work generated by the
consisting of length and diameter of segments, using the
flexors and extensors for all joints were summed to produce
landmark locations as references, were manually collected,
the total positive work (Eq. (3)) of each lower limb
along with body mass and height. Stride characteristics
ðW positive
Limb Þ. The same was carried out for the total absorbed
were recorded with three foot-switches located on the soles
work (W negative
Limb ; Eq. (4))
of their shoes (heel, mid-foot, and tip of the foot).
W positive
Limb ¼ W positive positive positive
Ankle þ W Knee þ W Hip ð3Þ
2.3. Data processing negative negative negative negative
W Limb ¼ W Ankle þ W Knee þ W Hip ð4Þ
The coordinate data, sampled at 60 Hz, were digitally
filtered using a 4th-order Butterworth zero-lag filter, with The relative contributions (Cont) of each joint to the
a cut-off frequency of 6 Hz. The data analysis (Mishac total positive or total negative work was calculated by
Kinetics, Waterloo, Canada) was used to calculate the rel- dividing the work values generated or absorbed by each
ative angles from a rotation matrix using a cardanic joint by the total energy value, and multiplied by 100.
sequence, so that the local x, y and z axes corresponded, For example, the contribution to the positive work by the
respectively, to abduction–adduction, longitudinal rotation ankle (Eq. (5)) in a gait cycle was calculated as followed:
and flexion–extension for the hip and knee joints, and ever- .
sion–inversion, transverse rotation, and dorsiflexion–plan- Contpositive
Ankle ¼ W positive
Ankle W positive
Limb
tarflexion for the ankle joint. .
Ground reaction forces were collected at 600 Hz with ContAnkle ¼ W Ankle W negative
negative negative
Limb ð5Þ
three force platforms embedded in the nine-m walkway.
These data were also filtered with a 4th-order Butterworth Finally, the joint contributions to the total work
zero-lag filter, with a cut-off frequency of 10 Hz and resam- observed in a gait cycle were also computed, using Eq.
pled at 60 Hz to match the kinematic data. The foot con- (6) as an example for the ankle
tact and ground reaction forces served to determine the h i.h i
gait cycles, which were normalized to 100%. For each ContAnkle ¼ W positive þ W negative
W positive
þ W negative
Ankle Ankle Limb Limb
imposed cadence, the three trials showing the closest
cadences were averaged for each side. An inverse dynamic ð6Þ
 Author's personal copy

772 L.F. Teixeira-Salmela et al. / Clinical Biomechanics 23 (2008) 769–778

2.4. Data analyses 3.2. Power profiles

All statistical analyses were carried out using SPSS for The overall average patterns for the ankle, knee, and hip
Windows. Descriptive statistics and tests for normality for the dominant side across all cadences are presented in
(Shapiro-Wilk) were performed for all outcome variables. super-imposed graphs. The curve patterns had several
Differences in spatio-temporal parameters between the four characteristics that were common to all subjects. At the
cadences were identified by repeated measures ANOVA hip, three major power phases were observed (Fig. 1A):
followed by planned contrasts with a significance level of
a < 0.05.
Since no main effects for laterality were found (see Sec- A 200
60
tion 3), the data for the dominant and non-dominant sides
150 80
were collapsed and a two-way within-factor repeated mea-
Natural
sures ANOVA was used to assess joint and cadence effects H3
100 120
and interactions. Whenever a main effect or interaction
effect was significant, focused contrasts were performed

Watts
50
H1
to locate the differences. Bonferroni corrections for multi-
ple comparisons in terms of cadence conditions were
0
applied to contrast results, changing the significance level
from 0.05 to 0.008 (0.05/6). Gait power profiles were
-50
described, but not statistically analyzed, to demonstrate
H2
changes associated with the cadence.
-100
0 10 20 30 40 50 60 70 80 90 100
Percent of Gait Cycle
3. Results
B 200
60
Fourteen healthy subjects (seven women and seven men)
80
completed the evaluation. The mean age, stature and body 150
Natural
mass were 46.0 (±13.3) years, 1.70 (±0.10) m, and 72.0
120
(±14.9) kg, respectively. Only one subject was left-side 100
dominant.
Watts

50
K2

3.1. Spatio-temporal measures

0

Table 1 presents the description of the spatio-temporal -50

variables for the natural and imposed cadences and the
actual cadences were close to the prescribed ones. Gait K1 K3 K4
-100
speed and stride length increased with cadences and greater 0 10 20 30 40 50 60 70 80 90 100
variability was observed for the natural and 120 prescribed Percent of Gait Cycle
cadences. ANOVA revealed that the four cadences pro-
duced highly significant gait speed differences C A2
200
(P = 0.0001). However, no differences were found between
the natural and the imposed 120 cadences (P = 0.16). 60
150 80
Natural
100 120
Watts

Table 1
50
Means (±SD) and range [min–max] of spatio-temporal measures for the
dominant side across cadences (n = 14)
0
Measure Cadence
60 80 Natural 120
-50
Cadence 65.8 ± 4.8 83.7 ± 2.1 105.3 ± 13.7 115.1 ± 12.0
reached [61.0–75.7] [79.3–85.7] [77.2–129.8] [95.9–141.5] A1
-100
(steps/min)
0 10 20 30 40 50 60 70 80 90 100
Stride length 1.2 ± 0.1 1.3 ± 0.1 1.4 ± 0.2 1.5 ± 0.2
(m) [1.0–1.4] [1.1–1.5] [1.2–1.7] [1.1–1.7] Percent of Gait Cycle
Speed (m/s) 0.67 ± 0.09 0.92 ± 0.10 1.25 ± 0.24 1.42 ± 0.22
Fig. 1. Power profiles (Watts) across cadences: (A) at the hip, (B) at the
[0.51–0.82] [0.75–1.07] [0.79–1.53] [1.08–1.85]
knee, and (C) at the ankle joint.
 Author's personal copy
L.F. Teixeira-Salmela et al. / Clinical Biomechanics 23 (2008) 769–778
the power was positive during the first half of the stance
phase, negative during the second half, and became posi-
tive during late stance (pull-off phase). For the knee, four
major power phases were identified, with power being pri-
marily negative (Fig. 1B). As shown in Fig. 1C, two major
phases were observed at the ankle: negative power was
recorded at early and mid-stance with a large positive peak
at late stance. As shown in Fig. 1A–C, a common pattern
for all joints was reflected by a consistent trend towards
higher values of power bursts for all joints as the cadence
increased.
3.3. Energy variables
3.3.1. Joint and cadence effects
Table 2 provides the descriptive data related to energy
generated and absorbed, as well as the energy contributions
at each joint for the natural and imposed cadences for all
subjects. Figs. 2 and 3 illustrate the mechanical energy gen-
erated and absorbed, as well as the relative contributions of
the hip, knee, and ankle joints across cadences.
For respective energy generation and absorption, signif-
icant effects for joint (F = 143.54; df = 2; P < 0.001 and
F = 3.98; df = 2; P = 0.03), cadence (F = 36.64; df=2;
P < 0.001 and F = 22.98; df = 2; P < 0.001), and joint-by-
cadence interactions (F = 10.79; df = 6; P < 0.001 and
F = 12.95; df = 6; P < 0.001) were demonstrated. Since
ANOVA revealed significant interactions between joints
and cadences, subsequent analyses used planned contrasts
for comparisons for joint or cadence effects.
For the energy generating values (Table 2 and Fig. 2A),
planned contrast analyses showed significant differences for
all joints across all cadences (P < 0.001), except between
the natural and 120 cadence. Moreover, there was a joint
effect for all cadences (P < 0.001), except between the ankle
and hip at the imposed cadence of 120.
For the energy absorbing values (Table 2 and Fig. 3A),
significant differences were found only for the knee and hip
joints (P < 0.001) across all cadences, except between the
natural and 120 cadence. Statistical analyses revealed that
ankle energy was significantly higher than at the knee
and hip energy at lower cadences (60 and 80), while the
knee energy level was significantly higher than the hip
and ankle energy for higher cadences (natural and 120
cadence).
3.3.2. Joint contributions
ANOVA revealed that the contributions to energy gen-
eration and absorption at the individual joints also chan-
ged with cadence (Table 2). Analyses for relative
contributions to energy generation demonstrated signifi-
cant effects for joint (F = 143.54; df = 2; P < 0.001 and
F = 3.98; df = 2; P = 0.03) and joint-by-cadence interac-
tion (F = 10.79; df = 6; P < 0.001 and F = 12.95; df = 6;
P < 0.001), respectively. No main effects were observed
for the relative contributions to energy absorption; how-
ever, analyses showed joint-by-cadence interactions. Since
773
the ANOVA revealed significant interactions between
joints and cadences, subsequent analyses used planned con-
trasts for comparisons of joint or cadence effects.
For the contributions to energy generation (Table 2 and
Fig. 2B), analyses showed significant decreases at the ankle
joint and significant increases for the hip joint across all
cadences (P < 0.001), except for those at the hip between
natural and the 120 cadence. For the knee, no differences
were found between 60 and 80 (P = 0.41), and between nat-
ural and the 120 cadence (P = 0.21). Moreover, there was a
joint effect for all cadences (P = 0.001).
For the contributions to energy absorption (Table 2 and
Fig. 3B), planned contrast analyses showed significant dif-
ferences for the knee across all cadences (P = 0.001), except
for the natural and the 120 cadence, which were not signif-
icantly different (P = 0.09). No differences were found at
the hip joint and for the ankle between 60 and 80
(P = 0.08), and between natural and the 120 cadence
(P = 0.51). Moreover, there was a joint effect for all
cadences, except for the knee and hip at the lower cadences
of 60 and 80 and for the ankle and hip for higher cadences
(natural and 120). The results indicated that at lower
cadences (60 and 80), similar contributions were observed
for the knee and hip, whereas at higher cadences (natural
and 120), the ankle and hip showed similar contributions.
4. Discussion
The present study assessed the work generation and
absorption, as well as the relative contributions of each
lower extremity joint to the total positive and negative
work during gait, in relation to cadence. Mechanical work
about the lower extremity joints was shown to be influ-
enced by the walking cadence, independent of the side of
the body. In addition, the contributions of the individual
joints to the total generated and absorbed energy during
gait was also influenced by cadence, suggesting that speed
differences should be minimized when comparing power
values between walking trials in other studies. Four main
issues will be discussed in the following sections.
4.1. Spatio-temporal measures
The four investigated cadences resulted in mean speeds
ranging from 0.67 to 1.42 m/s, which were very similar than
those of 0.65–1.35 m/s assessed with healthy older adult
subjects (Teixeira, 1998). In the present study, these
cadences were chosen to match those usually demonstrated
by subjects with locomotor disabilities walking at natural
speeds, whose speed values rarely surpass 1.42 m/s (Olney
et al., 1991, 1994; Olney and Richards, 1996; Teixeira-Sal-
mela et al., 2001). No differences in gait speed were found
between the higher natural cadences and the imposed one
of 120 steps/min. The reasons behind these non-significant
differences may be related to the large variations across
subjects. It appears that some subjects had a higher natural
cadence than the 120 steps/min, while others found it diffi-
 Author's personal copy

774
Table 2
Means (±SD), range [min–max] of positive and negative work (Joules) and contributions (%) to positive and negative work at ankle, knee, and hip joints across cadences (n = 28)

L.F. Teixeira-Salmela et al. / Clinical Biomechanics 23 (2008) 769–778

Ankle Knee Hip Limb(ankle+knee+hip)
P þ P P þ P þ
W+ W ðW þ W  Þ W+ W ðW þ þ W  Þ W+ W ðW þ W  Þ W+ W ðW þ W  Þ
60 steps
Mean 13.4 14.6 28 3.6 7.6 11.2 5.5 8.5 14 22.5 30.7 53.2
SD 3.7 4.5 9.7 2 2.6 5.3 2.5 5.4 6.8 8.5 11.14 18.5
Range [6.5–19.2] [10–28.3] [18–42] [0.8–8.0] [1.9–12.1] [3–20] [2.3–11.6] [0.7–23.4] [5–30] [10–33] [13–45] [29–75]
Cont (%) to WLimb 59.8 47.6 52.7 15.8 24.9 21 24.4 27.6 26.3 100 100 100
80 steps
Mean 15 14.3 29.3 4.3 11.3 15.6 8.3 10 18.3 27.6 35.6 63.2
SD 4.5 4.3 10.6 2.4 3.8 7.2 4.5 4.8 7.7 11 12.5 22.2
Range [8.3–23.6] [9.4–23.4] [18–46] [0.3–9.2] [4.8–20.9] [5–27] [3.7–24.5] [4.7–24.6] [9–33] [12–55] [19–55] [36–93]
Cont (%) to WLimb 54.4 40.3 46.5 15.6 31.7 24.6 30 28 28.9 100 100 100
Natural
Mean 19.2 12. 5 31.7 7 18.6 25.6 13.7 13 26.7 39.9 44.1 84
SD 5 4.5 7.7 3.1 6.7 10.4 7 5.8 10.3 13.2 13.9 25
Range [9.3–29.5] [6.9–22.9] [20–47] [2.9–13.7] [10–38.1] [15–52] [4–28] [5.5–26.0] [13–52] [23–67] [24–81] [57–139]
Cont (%) to WLimb 48.1 28.3 37.7 17.6 42.2 30.5 34.3 29.5 31.8 100 100 100
120 steps
Mean 19.7 12.3 32 8.5 22.3 30.8 17.1 14.5 31.6 45.3 49.1 94.4
SD 5.5 4.7 10.22 3.5 7.2 12.5 8.8 6.8 14.2 17.8 16.8 33.1
Range [11–33.7] [5–23.4] [21–48] [2.7–16.2] [7.5–35.9] [10–52] [5.5–47.8] [4.1–28.9] [12–77] [19–90] [17–77] [45–175]
Cont (%) to WLimb 43.5 25.1 33.9 18.7 45.4 32.6 37.8 29.7 33.5 100 100 100
 Author's personal copy

L.F. Teixeira-Salmela et al. / Clinical Biomechanics 23 (2008) 769–778 775

30
A A 30
Ankle

Mechanical Energy (J)

25 Ankle
Mechanical Energy (J)

25 Knee
Knee
Hip
Hip
20 20

15
15

10
10
5
5
0
0 60 steps 80 steps Natural 120 steps
60 steps 80 steps Natural 120 steps
B 70 Ankle
80
B 60 Knee
70 Hip
Ankle 50
60 Knee
Hip 40

%
50
30
(%)

40
20
30
10
20
0
10 60 steps 80 steps Natural 120 steps

0 Fig. 3. (A) Mechanical energy absorption at the hip, knee, and ankle
60 steps 80 steps Natural
Fig. 2. (A) Mechanical energy generation at the hip, knee, and ankle
joints (Joules; J) across cadences. (B) Relative contributions of the hip,
knee, and ankle joints (percentages; %) to energy generation across
cadences.
cult to reach the prescribed 120 steps/min cadence. The net
result was an overlap of the distributions of gait speed mea-
sured at the natural and 120 cadence resulting in non-sig-
nificant differences. Considering that energy changes were
closely associated with differences in gait speed, it was
not surprising that differences between the natural and
imposed 120 cadence for energy parameters did not reach
statistical significance.
4.2. Cadence effects
The influences of speed on kinetic parameters has been
extensively investigated (Bishop et al., 2004; Chen et al.,
1997; Den Otter et al., 2004; Graf et al., 2005; Miyoshi
et al., 2004; Olney et al., 1994) and independent of the
selected population, methodology, and parameters used,
the effects of cadence were consistent. Kim and Eng
(2004) observed that the magnitude of kinematic and
kinetic gait profiles were significantly related to gait speed
in all three planes. Significant correlations between the
same work phases across cadences indicate the same pat-
tern with increases in cadence. However, the absence of sig-
nificant correlations between different work phases at a
given cadence indicates that the subjects did not adopt con-
sistent patterns to generate or absorb work, i.e., they used
120 steps joints (Joules; J) across cadences. (B) Relative contributions of the hip,
knee, and ankle joints (percentages; %) to energy absorption across
cadences.
different strategies to accomplish the task. It is possible that
some subjects may have used their plantarflexors more
than hip flexors or extensors, while others may have done
the reverse. In fact, Vardaxis et al. (1998) suggested that
the variability observed in able-bodied gait may be the
result of multiple normal dynamic strategies employed by
different subjects and should be considered in studies inves-
tigating gait.
Cadence influenced the mechanical work performed by
the muscles acting at the hip, knee, and ankle. This sup-
ported previous findings (Chen et al., 1997) that the total
energy generated and absorbed by the knee showed the
largest increases (175%) in response to cadence changes.
The hip also showed substantial increases of 126%, whereas
the ankle demonstrated the smallest changes. The present
findings supported the evidence of a trade-off between the
hip and ankle joints, which were also shown by Graf
et al. (2005) and could reflect adjustments of the lower limb
to more efficiently meet different task demands or to atten-
uate fatigue and share the effort across muscles when the
cadence increases. Previous analyses of the level of effort
during the period of energy generation for this group of
subjects has revealed that the level of effort increased with
gait cadence but the increases were more pronounced for
the hip muscles (Requião et al., 2005). Energy generation
around the knee was minimal compared to that generated
at the ankle and hip joints, reinforcing the findings that
 Author's personal copy
776 L.F. Teixeira-Salmela et al. / Clinical Biomechanics 23 (2008) 769–778
the knee muscles mainly absorb, rather than generate
energy (Olney et al., 1991; Winter, 1983).
Negative work supplied by the knee has been reported
to be highly correlated with speed and one of the best pre-
dictors in gains in walking performance (Olney et al., 1994;
Winter, 1983). It appears that at increased cadences, mus-
cles generate and absorb energy at faster rates. Important
increases observed at the knee joint might be the result of
the work of numerous biarticular muscles in promoting
energy flow between segments and transferring energy from
the leg to the thigh to increase overall gait speed. The tran-
sition patterns between the energy generation/absorption
of the hip, knee, and ankle muscles suggest a shift of work
to larger muscle groups at faster cadences. This shift of
energy to larger and more proximal muscle groups may
permit muscles to work at a lower percentage of their max-
imum capacity and, therefore, optimize energy consump-
tion during gait.
For the different behaviors of the muscles at the ankle,
which showed the smallest increases in work values in
response to changes in cadence, there are parallel findings
in the literature. Andriacchi et al. (1977) demonstrated that
the muscle moment about the hip and knee showed greater
dependencies on speed, whereas, the moment at the ankle
did not substantially change. In addition, studies with chil-
dren (Chen et al., 1997), adults (Winter, 1983), and with
low-performing elderly subjects (Graf et al., 2005) have
shown that the work at the ankle demonstrated the least
sensitivity to speed variations.
4.3. Contributions to energy generation and absorption
When expressed as a percentage of work contributions,
the variations of behavior of the ankle, knee, and hip in
relation to cadence, become even more obvious. The con-
tributions to work from the muscles about the knee and
hip showed increases at higher cadences, while those from
the muscles about the ankle substantially decreased. For
instance, at the lowest cadence, about 60% of the total
energy generated resulted from the muscles about the
ankle. However, when the cadence was up to 120 steps/
min, only 44% of the total energy generated originated
from these muscles. The contributions of the muscles about
the hip to the total energy generated increased from 24% at
the lower to 38% at higher cadences, whereas knee contri-
butions increased from 16% to 19%, indicating large contri-
butions of hip muscles for speed modulation.
Few studies have evaluated the relative contributions of
the muscles about the ankle to the total energy generated
during gait at different speeds, but differences in the assess-
ment protocols make comparisons difficult. The findings of
the present study are somewhat in agreement with those of
Chen et al. (1997) on contributions of the lower extremity
joints to the total work generated within three walking
groups of healthy children. In the present study, cadences
were tightly controlled, while in the study of Chen et al.
(1997), they were not strictly imposed. Participants were
instructed to walk at various speeds, which were selected
post-hoc and subsequently classified into three categories
based on their actual walking speeds.
Regarding the respective contributions to energy
absorption, increases were seen at the hip and knee joints,
but the most striking changes were observed at the knee,
which increased from 25% at the lowest cadence to 45%
at the highest. Energy absorption is not only associated
with an active contraction but passive contributions to
energy absorption should also be considered. Neptune
et al. (2004) observed that forces developed from stretched
passive hip structures assisted plantarflexors during the
swing initiation. By the same token, the passive energy
absorbed by the rectus femoris at the end of stance may
have assisted hip flexors in limb propulsion.
It is interesting to note how the total contributions at
each joint to the sum of the total limb generation and
absorption changed when cadences were increased. At the
lowest cadence, the ankle, knee, and hip contributed
53%, 21%, and 26%, respectively, however, at the fastest
cadence, these discrepancies were less evident and the con-
tributions by each joint to the total energy were more bal-
anced. For instance, at the highest cadence, 34% of the
total energy was provided by the structures about the
ankle, 33% from the knee and 33% from the hip, indicating
that all three joints played similar roles in contributing to
the sum of the total energy used. These findings suggest
that, at faster cadences, the power-energy capability of
the knee muscles was functionally relevant, since they were
responsible for about one third of the total work performed
at the highest cadence. Therefore, the contributions of the
knee as a energy modulator should not be underestimated.
Although the positive and negative work increased at
about the same amount for the knee and hip joints across
cadences, they were actually quite different. There were rel-
atively small increases in the positive work for the knee
(136%) and much greater changes at the hip (221%). On
the other hand, there were much larger increases in the neg-
ative work for the knee (192%) and less for the hip (70.6%).
It is also important to note that the ankle had a larger
absolute increase in positive work (6.3 J) across cadences
than did the knee (4.9 J), a value that was about half of
the amount of the positive work at the hip (11.6). However,
the negative work at the ankle actually decreased (2.3 J,
16%) across cadences, whereas increases of 6.0 and 14.6 J
were found for the hip and knee, respectively.
In the present study, only power and work performed
in the sagittal plane during concentric and eccentric
contractions were evaluated and it was observed that the
differences between the positive and negative values for
the total limb work, were always higher for the negative
work, which supported the limitations of the sagittal plane
model. It has been reported that with normal subjects,
contributions to kinetic energy existed from sagittal and
non-sagittal plane motions (Vardaxis et al., 1998). In fact,
Graf et al. (2005) reported that the larger transverse pelvic
rotations employed by low-performing subjects may be a
 Author's personal copy
L.F. Teixeira-Salmela et al. / Clinical Biomechanics 23 (2008) 769–778
compensatory mechanism to increase step length and gait
speed. In addition, body support and forward progression
were found to occur because power redistribution between
the legs and trunk occurred through muscular force gener-
ation, regardless of whether produced from concentric,
isometric, or eccentric activity. It could be argued that
underestimation of power and work might have occurred
because of the absence of motion evaluation in the frontal
and transverse planes and that produced by isometric
muscle activity, especially for the hip joint. For instance,
if the lower limb was laterally rotated, so that the adduc-
tors substantially contributed to hip flexion, then work
would inappropriately be attributed to hip flexors.
However, the major portion of this movement is
performed in the sagittal plane, because the goal of
locomotion is to support body weight against gravity
while generating forward motion (Eng and Winter,
1995). However, future studies should also include
measures of power and work in the frontal plane of
movement, particularly at the hip joint, where concentric
and eccentric works are performed (Nadeau et al., 2003).
5. Conclusions
The findings demonstrated that the performance of the
dominant and non-dominant sides was very similar, inde-
pendent of the joint or cadence. Power and mechanical
work, as well as the contributions of individual joints to
the total energy generated and absorbed, were shown to
be influenced by the walking cadence, independent of later-
ality. The ankle, knee, and hip contributions to the total
limb generation and absorption at the lowest cadence were
53%, 21%, and 26%. At the highest cadence, the corre-
sponding values were 34%, 33%, and 33%, respectively.
The present findings provide baseline data for identifying
changes in individuals with locomotor disorders. The esti-
mation of the relative contributions of each joint to the
total energy generated and absorbed during gait at different
speeds could provide a useful approach to understand the
nature, extent, and degree of compensation between joints
and suggest more efficient methods of correction.
Acknowledgments
Canadian Institute of Health Research, CAPES (Brazil-
ian Government Agency), and Fonds de la Recherche en
Santé du Québec (FRSQ).
References
Andriacchi, T.P., Ogle, J.A., Galante, J.O., 1977. Walking speed as a basis
for normal and abnormal gait measurements. J. Biomech. 10, 261–268.
Bishop, M., Brunt, D., Pathare, N., Patel, B., 2004. The effect of velocity
on the strategies used during gait termination. Gait Posture 20, 134–
139.
Chen, I.H., Kuo, K.N., Andriacchi, T.P., 1997. The influence of walking
speed on mechanical joint power during gait. Gait Posture 6, 171–
176.
777
Den Otter, A.R., Geurts, A.C., Mulder, T., Duysens, J., 2004. Speed
related changes in muscle activity from normal to very slow walking
speeds. Gait Posture 19, 270–278.
Eng, J.J., Winter, D.A., 1995. Kinetic analysis of the lower limbs during
walking: what information can be gained from a three-dimensional
model? J. Biomech. 28, 753–758.
Gitter, A., Czerniecki, J.M., DeGroot, D.M., 1991. Biomechanical
analysis of the influence of prosthetic feet on below-knee amputee
walking. Am. J. Phys. Med. Rehabil. 70, 142–148.
Graf, A., Judge, J.O., Õunpuu, S., Thelen, D.G., 2005. The effect of
walking speed on lower-extremity joint powers among elderly adults
who exhibit low physical performance. Arch. Phys. Med. Rehabil. 86,
2177–2183.
Hermodsson, Y., Ekdahl, C., Persson, B.M., Roxendal, G., 1994. Gait in
male trans-tibial amputees: a comparative study with healthy subjects
in relation to walking speed. Prosthet. Orthot. Int. 18, 68–77.
Kim, C.M., Eng, J.J., 2004. Magnitude and pattern of 3D kinematic and
kinetic gait profiles in persons with stroke: relationship to walking
speed. Gait Posture 20, 140–146.
Mansour, J.M., Lesh, M.D., Nowak, M.D., Simon, S.R., 1982. A three
dimensional multi-segmental analysis of the energetics of normal and
pathological human gait. J. Biomech. 15, 51–59.
McGibbon, C.A., Krebs, D.E., Puniello, M.S., 2001. Mechanical energy
analysis identifies compensatory strategies in disabled elders’ gait. J.
Biomech. 34, 481–490.
Milot, M.H., Nadeau, S., Gravel, D., Requiao, L.F., 2006. Bilateral level
of effort of the plantar flexors, hip flexors, and extensors during gait in
hemiparetic and healthy individuals. Stroke 37, 2070–2075.
Miyoshi, T., Shirota, T., Yamamoto, S., Nakazawa, K., Akai, M., 2004.
Effect of the walking speed to the lower limb joint angular displace-
ments, joint moments and ground reaction forces during walking in
water. Disabil. Rehabil. 26, 724–732.
Mueller, M.J., Minor, S.D., Sahrmann, S.A., Schaaf, J.A., Strube, M.J.,
1994a. Differences in the gait characteristics of patients with diabetes
and peripheral neuropathy compared with age-matched controls. Phys.
Ther. 74, 299–308.
Mueller, M.J., Sinacore, D.R., Hoogstrate, S., Daly, L., 1994b. Hip and
ankle walking strategies: effect on peak plantar pressures and impli-
cations for neuropathic ulceration. Arch. Phys. Med. Rehabil. 75,
1196–1200.
Nadeau, S., Gravel, D., Olney, S.J., 2001. Determinants, limiting factors
and compensatory strategies in gait. Crit. Rev. Phys. Rehabil Med. 13,
1–24.
Nadeau, S., McFadyen, B., Malouin, F., 2003. Frontal and sagittal
analysis of the stair climbing task in healthy adults over 40 years: what
are the challenges compared to level walking? Clin. Biomech. 18, 950–
959.
Neptune, R.R., Zajac, F.E., Kautz, S.A., 2004. Muscle force redistributes
segmental power for body progression during walking. Gait Posture
19, 194–205.
Olney, S.J., Richards, C., 1996. Hemiparetic gait following stroke: part 1:
characteristics. Gait Posture 4, 136–148.
Olney, S.J., Griffin, M.P., Monga, T.N., McBride, I.D., 1991. Work and
power in gait of stroke patients. Arch. Phys. Med. Rehabil. 72, 309–
314.
Olney, S.J., Griffin, M.P., McBride, I.D., 1994. Temporal, kinematic, and
kinetic variables related to gait speed in subjects with hemiplegia: a
regression approach. Phys. Ther. 74, 872–885.
Requião, L.F., Nadeau, S., Milot, M.H., Gravel, D., Bourbonnais, D.,
Gagnon, D., 2005. Quantification of level of effort at the plantarflexors
and hip extensors and flexor muscles in healthy subjects walking at
different cadences. J. Electromyogr. Kinesiol. 15, 393–405.
Sadeghi, H., Allard, P., Duhaime, P.M., 2001. Muscle power compensa-
tory mechanisms in below-knee amputee gait. Am. J. Phys. Med.
Rehabil. 80, 25–32.
Seroussi, R.E., Gitter, A., Czerniecki, J.M., Weaver, K., 1996. Mechanical
work adaptations of above-knee amputee ambulation. Arch. Phys.
Med. Rehabil. 77, 1209–1214.
 Author's personal copy
778 L.F. Teixeira-Salmela et al. / Clinical Biomechanics 23 (2008) 769–778
Siegel, K.L., Kepple, T.M., Stanhope, S.J., 2004. Joint moment control of
mechanical energy flow during normal gait. Gait Posture 19, 69–
75.
Teixeira, L.F., 1998. The impact of a program of muscle strengthening
and physical conditioning on impairment and disability in chronic
stroke survivors. Ph.D. Dissertation, Queen’s University, Kingston,
Ontario.
Teixeira-Salmela, L.F., Nadeau, S., Mcbride, I., Olney, S.J., 2001. Effects
of muscle strengthening and physical conditioning training on tempo-
ral, kinematic and kinetic variables during gait in chronic stroke
survivors. J. Rehabil. Med. 33, 53–60.
Vardaxis, V.G., Allard, P., Lachance, L., Duhaime, M., 1998. Classifica-
tion of able-bodied gait using 3-D muscle powers. Hum. Mov. Sci. 17,
121–136.
Winter, D.A., 1983a. Energy generation and absorption at the ankle and
knee during fast, natural, and slow cadences. Clin. Orthop. Relat. Res.
175, 147–154.
Winter, D.A., 1983b. Biomechanical motor patterns in normal walking. J.
Motor Behav. 15, 302–330.
Winter, D.A., 1991. The Biomechanics and Motor Control of Human
Gait: Normal, Elderly and Pathological. University of Waterloo Press,
Waterloo.