536762

research-article2014
POI0010.1177/0309364614536762Prosthetics and Orthotics InternationalAgrawal et al.

INTERNATIONAL
SOCIETY FOR PROSTHETICS
AND ORTHOTICS

Original Research Report

Prosthetics and Orthotics International

Comparison of four different categories 1­–10

© The International Society for
Prosthetics and Orthotics 2014
of prosthetic feet during ramp ambulation Reprints and permissions:
sagepub.co.uk/journalsPermissions.nav
in unilateral transtibial amputees DOI: 10.1177/0309364614536762
poi.sagepub.com

Vibhor Agrawal1,2, Robert S Gailey1,2, Ignacio A Gaunaurd1,2,

Christopher O’Toole3, Adam Finnieston4, and Ronald Tolchin5

Abstract
Background: Comparative effectiveness of prosthetic feet during ramp ambulation in unilateral transtibial amputees, who
function at different Medicare Functional Classification Levels, has not been published.
Objective: To determine differences in symmetry in external work between four categories of prosthetic feet in K-Level-2
and K-Level-3 unilateral transtibial amputees during ramp ascent and descent.
Study design: Randomized repeated-measures trial.
Methods: Ten subjects completed six testing sessions during which symmetry in external work was calculated using
F-scan in-sole sensors. Between testing sessions 1 and 2, subjects received standardized functional prosthetic training.
In Sessions 3–6, subjects tested four feet—solid ankle cushion heel, stationary attachment flexible endoskeleton, Talux
(categories K1, K2, and K3, respectively), and Proprio-Foot (microprocessor ankle)—using a study socket and had a
10- to 14-day accommodation period with each foot.
Results: During ramp descent, K-Level-2 subjects demonstrated higher symmetry in external work values with Talux and
Proprio-Foot compared to the solid ankle cushion heel foot. K-Level-3 subjects also had higher symmetry in external
work values with the Talux foot than the solid ankle cushion heel foot. Ramp ascent symmetry in external work values
were not significantly different between feet.
Conclusions: Prosthetic foot category appears to influence symmetry in external work more during decline walking than
incline walking. K-Level-2 unilateral transtibial amputees achieve greater symmetry from K3 dynamic response prosthetic
feet with J-shaped ankle and microprocessor ankles while descending ramps.

Clinical relevance
The findings suggest that K-Level-2 unilateral transtibial amputees benefit from K3 dynamic response prosthetic feet with
J-shaped ankle. These results support the prescription of K3 feet for K-Level-2 amputees who frequently negotiate ramps.

Keywords
Biomechanics of prosthetic/orthotic devices, biomechanics, gait analysis, gait, prosthetic feet, prosthetics,
physiotherapy, rehabilitation

Date received: 13 February 2012; accepted: 23 April 2014

Background 1Department of Physical Therapy, Miller School of Medicine, University

of Miami, Coral Gables, USA
In the United States, the governmental definition of foot/ 2Research Department, Miami Veterans Affairs Healthcare System,
ankle assemblies (Table 1) defines five Medicare Miami, USA
Functional Classification Levels (MFCL) for amputees 3The Miami Project to Cure Paralysis, Miami, USA

(K-Level-0 through K-Level-4) with corresponding cate-
gories (K0–K4) for prosthetic feet.1,2 Prosthetic prescrip-
tion and subsequent reimbursement frequently mandates
that the amputee’s assigned MFCL correspond to a pros-
thetic foot that has the same class assignment. However, to
date, there has been little objective data demonstrating the
4Arthur Finnieston Prosthetics and Orthotics, Miami, USA
5Baptist Hospital Neuroscience Center, Spine Center, Miami, USA
Corresponding author:
Vibhor Agrawal, Department of Physical Therapy, University of Miami,
5915 Ponce De Leon Blvd, Coral Gables, FL 33146, USA.
Email: agrawal.vr@gmail.com

Downloaded from poi.sagepub.com at GEORGIAN COURT UNIV on March 3, 2015

2 Prosthetics and Orthotics International

Table 1. DMERC prosthetic policy definition of foot/ankle assemblies by Medicare Functional Classification Levels.

MFCL Description MFCL Foot/ankle

modifiers assemblies
Level 0 Does not have the ability or potential to ambulate or transfer K0 Not eligible for
safely with or without assistance, and a prosthesis does not prosthesis
enhance quality of life or mobility.
Level 1 Has the ability or potential to use prosthesis for transfers or K1 External keel, SACH
ambulation in level surfaces at a fixed cadence. Typical of the foot, or single-axis ankle/
limited and unlimited household ambulators. foot
Level 2 Has the ability or potential for ambulation with the ability to K2 Flexible keel foot or
transverse low-level environmental barriers such as curbs, stairs, multiaxial ankle/foot
or uneven surfaces. Typical of the limited community ambulator.
Level 3 Has the ability or potential for ambulation with variable cadence. K3 Flex foot system, energy
Typical of the community ambulator who has the ability to storing foot, multiaxial
transverse most environmental barriers and may have vocational, ankle/foot, dynamic
therapeutic, or exercise activity that demands prosthetic response, or flex-walk
utilization beyond simple locomotion. system or equivalent
Level 4 Has the ability or potential for prosthetic ambulation that exceeds K4 Any ankle/foot system
basic ambulation skills, exhibiting high impact, stress, or energy appropriate
levels. Typical of the prosthetic demands of the child, active adult,
or athlete.

DMERC: Durable Medical Equipment Regional Carrier; MFCL: Medicare Functional Classification Levels; SACH: solid ankle cushion heel.

functional correlation between MFCL and the assignment
of prosthetic components.
Unilateral transtibial amputees (TTAs) have a tendency
to load the intact limb more than the prosthetic limb during
level ground walking.3,4 The resulting kinetic asymmetry
is influenced by prosthetic gait training5 and the category
of prosthetic foot used.5–7 Comparisons between level
ground walking and incline/decline walking reveal higher
demands on the musculoskeletal system and greater
ground reaction forces (GRFs) during ramp ambulation in
non-amputees.8 If TTAs have higher intact limb loads dur-
ing level walking and the non-amputee population experi-
ences greater forces during incline/decline walking, it
would stand to reason that during incline/decline walking
TTAs would load their intact limb more than level walking
and the asymmetry in forces would be magnified.
The only related study found on ramp GRFs with TTAs
reported that elderly unilateral TTAs (K-Level-2) who
used a solid ankle cushion heel (SACH) foot (category K1
foot) to negotiate ramps experienced significantly higher
GRFs on the intact limb compared to the prosthetic limb.9
They also exhibited a shorter single limb support time on
the prosthetic limb than the intact limb. The authors attrib-
uted the kinetic and temporal asymmetry to the design fea-
tures of the SACH foot, such as its limited stance phase
ankle motion and lack of power generation.9 Dynamic
response (DR) prosthetic feet (category K3) have greater
energy return and provide greater ankle dorsiflexion dur-
ing stance period than the SACH foot because of their car-
bon fiber composition and “J-shaped” ankle design.10,11 A
K3 prosthetic foot with a J-shaped ankle has been shown
to promote greater symmetry in external work (SEW) dur-
ing level walking in K-Level-2 amputees due to its ankle
design and a heel-to-toe footplate.5 To date, influence of
foot design related to the prosthetic foot category on SEW
during incline and decline walking in unilateral TTAs has
not been investigated.
Traditional methods of assessing lower limb loading
involve analyzing GRFs at discrete points during a gait
cycle. Since the lower limbs experience variable GRFs
during the stance period, assessment of forces at discrete
points does not provide a complete picture of the conse-
quences of loading throughout the entire stance period.
Furthermore, traditional methods using direct compari-
sons of discrete events between prosthetic feet have not
consistently demonstrated the sensitivity to detect sig-
nificant differences between feet during level walk-
ing.12,13 The SEW measure has been employed as an
alternative to traditional biomechanical measures for
assessing kinetic and functional differences between
prosthetic feet.5,14
External work is the change in energy of the body center
of mass (CoM) that is brought about by GRFs.14 External
work can be calculated from the displacement of the CoM
and the GRFs applied to the CoM during the entire stance
period.14 The SEW measure calculates external work by
the intact and prosthetic limbs and generates an index of
symmetry between the limbs. This measure determines
collectively the similarities in GRFs, CoM displacement,
and step time between the intact and prosthetic limbs.
While the SEW measure has been successfully used to
detect functional differences between prosthetic feet

Downloaded from poi.sagepub.com at GEORGIAN COURT UNIV on March 3, 2015

Agrawal et al.
Table 2. Demographic characteristics of subjects for the test–retest reliability study.
Age Gender
(years)a
64 Male
55 Female
45 Female
59 Male
47 Male
Median 55 (45–64)
Range 45–64
Mean ± SD 54 ± 8
PVD: peripheral vascular disease; SD: standard deviation.
aNumbers have been rounded to the nearest whole number.
during level walking and stair negotiation, ramp ascent
and descent would be another appropriate application.5,15
The purpose of this study was to determine the func-
tional differences, as measured by SEW, between four cat-
egories of prosthetic feet during ramp ascent and descent
in K-Level-2 and K-Level-3 unilateral TTAs. It was
hypothesized that following standardized ramp training,
unilateral TTAs will demonstrate differences in SEW val-
ues between four categories of prosthetic feet during ramp
ascent and descent. Amputees classified with these func-
tional levels were selected because they represent potential
ramp ambulators who may benefit from the design propri-
eties of the four test prosthetic feet.
Methods
The study protocol was reviewed and approved by the
Institutional Review Board at the Miami VA Medical
Center and informed written consent was obtained from all
subjects prior to study enrollment. The inclusion criteria
for study participation were males and/or females between
the age 40 and 65 years with limb loss related to dysvascu-
lar reasons, trauma, and tumor and who wore a comforta-
bly fitting prosthesis for at least 6 months. All subjects
were able to negotiate ramps independently without an
assistive device. Potential subjects were screened by a
physician and prosthetist and were excluded if they had
any medical conditions or prosthetic limitations that would
cause pain or a potential health risk during the study. Data
collection was started upon physician and prosthetist clear-
ance. The study was conducted in two phases: Phase I—
Reliability study and Phase II—Intervention study.
Reliability study
Since the reliability of F-scan sensors (Tekscan Inc.,
Boston, MA, USA) has only been established for level
walking in TTAs,5 test–retest reliability of these sensors
was determined during incline and decline walking with
five unilateral TTA subjects (Table 2). The mean (standard
Downloaded from poi.sagepub.com at GEORGIAN COURT UNIV on March 3, 2015
3
Amputated Cause of Time since amputationa
side amputation (years)
Left PVD 3
Right Trauma 7
Left Trauma 3
Right PVD 2
Right Tumor 2
3 (2–7)
2–7
3.4 ± 2.1
deviation (SD)) age and time since amputation of the sub-
jects were 55.84 ± 6.5 years and 22.4 ± 18.8 months,
respectively. The median (range) age and time since ampu-
tation were 57.5 (43–64) years of age and 21 (6–72)
months, respectively. The same team of researchers per-
formed sensor calibration and data collection using previ-
ously validated procedures.16,17 During each testing
session, a new F-scan sensor was individually prepared for
each subject’s standardized shoe provided by the labora-
tory. Sensors were warmed up inside the shoes for 10
min.16 They were then calibrated using the manufacturer
recommended calibration and a force plate calibration
with two Kistler force plates embedded in a 10-m-long
level walkway. Vertical GRFs from force plates and F-scan
sensors were compared and a calibration factor for F-scan
sensors was calculated, as described by Mueller et al.17
Data were collected for 10 s at a sampling frequency of 50
Hz. Subjects then ascended and descended a 24-foot-long
custom fabricated wooden ramp, inclined at 5°—designed
per the Americans with Disabilities Act guidelines—
without using handrails. They were tested twice within a
48- to 72-h period using standardized shoes (Aetrex
Ambulator, type T1220) and socks. Intraclass correlation
coefficient (ICC) was used to determine the correlation
between test–retest SEW values.
Intervention study
Ten unilateral TTAs (five K-Level-2 and five K-Level-3
subjects) (Table 3) underwent six testing sessions to test
four categories of prosthetic feet. The test feet were: (1)
SACH foot (Kingsley Manufacturing Co., Costa Mesa,
CA, USA), (2) Stationary Attachment Flexible
Endoskeleton (SAFE) foot (Campbell-Childs Inc., White
City, OR, USA), (3) Talux foot (Ossur hf, Reykjavik,
Iceland) and (4) the Proprio-Foot (Ossur hf) (microproces-
sor ankle). These four prosthetic feet were selected to rep-
resent each category of foot as described in Table 1, that is,
K1—the SACH foot identified by name in the Durable
Medical Equipment Regional Carrier (DMERC) prosthetic
4 Prosthetics and Orthotics International

Table 3. Demographic characteristics of subjects for the intervention study.

MFCL class Age Gender Amputated Cause of Time since Height (cm) Mass Original Time in
(years)a side amputation amputation (kg) prosthetic current
(years)a foot prosthesis
(months)
K-Level-2 59 Male Right PVD 2 180.3 107.3 Otto Bock 18
Dynamic
Motion
K-Level-2 64 Male Left PVD 3 167.6 100 Endolite 24
Multiflex
K-Level-2 61 Male Right PVD 6 191.8 105.9 Seattle Catalyst 24
K-Level-2 58 Male Left PVD 2 185.4 102.7 SACH 12
K-Level-2 61 Male Left PVD 2 175.3 115.6 Endolite 24
Multiflex
K-Level-3 47 Male Right Tumor 2 172.7 77.6 Seattle Litefoot 12
K-Level-3 55 Female Right Trauma 7 167.6 88.8 Otto Bock 24
Dynamic
Motion
K-Level-3 43 Male Right Trauma 1 166.4 101.8 SACH 8
K-Level-3 53 Male Left Trauma 33 175.3 100 Otto Bock 6
Dynamic
Motion
K-Level-3 57 Male Left Trauma 37 167.6 93.2 Otto Bock 72
Springlite
Median 57.5 174.0 100.9 21
Range 43–64 166.4–191.8 77.6–115.6 6–72
Mean ± SD 55.8 ± 6.5 175 ± 8.6 99.3 ± 10.6 22.4 ± 18.8

MFCL: Medicare Functional Classification Levels; PVD: peripheral vascular disease; SACH: solid ankle cushion heel; SD: standard deviation.
aNumbers have been rounded to the nearest whole number.

policy definition of foot/ankle assemblies; K2—the SAFE
foot a truly flexible keel foot; K3—the Talux foot because
it is a Flex-Foot with a J-shaped ankle and DR design (also
named by DMERC) combined with multiaxial capabilities
because of the elastomer block. The Proprio-Foot, a K3
foot, was included as a microprocessor-controlled ankle
(MPA) that provides active dorsiflexion and plantarflexion
during ramp walking. Data collection procedures were
identical to the reliability study described above.
At Session 1 (Baseline), subjects used their existing
prosthetic socket and foot. There was a 2-week period
between Sessions 1 and 2, during which subjects received
up to 4 h of standardized functional prosthetic training, as
described below. During this time period, their residual
limb was also scanned to create a computer-aided design
(CAD) image (BioSculptor; Maramed Inc., Miramar, FL,
USA) which was used to fabricate a total surface bearing
socket having suction suspension with an Iceross Seal-in
liner (Ossur Inc.) or a cushion liner and external sleeve.
The purpose of standardizing the socket design and sus-
pension was to minimize the socket-related variables that
could influence gait performance during testing. During
Session 2 (Training), subjects were tested again in their
existing socket and foot to determine the effects of training
on their gait and were fit with the study socket and one of
the four randomized test feet (F1). The same study socket
was used for subsequent testing sessions and throughout
the duration of the study. During the 10- to 14-day
accommodation period with F1, they participated in 1–4 h
of training to learn to maximize the use of this foot design.
Any issues with socket fit or alignment were addressed
within 48 h of notification and all prosthetic or training-
related issues were resolved during one return visit. At
testing Session 3, effects of F1 on subjects’ gait were
measured. Following testing, subjects were fit with the
randomly selected second test foot (F2) and again were
given 1–4 h of foot-specific training during the 10- to
14-day accommodation period. The effects of F2 on sub-
jects’ gait were determined in Session 4. The same testing
and training procedure was repeated for the third and
fourth test feet during Sessions 5 and 6, respectively. All
socket fittings and prosthetic alignment procedures were
performed by the same board-certified prosthetist.
Standardized functional prosthetic training was admin-
istered to minimize gait deviations resulting from habit or
lack of training and also to maximize the use of each pros-
thetic foot’s functional properties. All subjects were able to
ascend and descend the ramp leg-over-leg without the use
of the handrail although they were instructed to hold onto
the handrail if required for safety. For ramp ascent, subjects
were instructed to place each foot flat on the surface in
front of the other foot, keeping a width of approximately 10
cm between feet. During training, cues were given to assist
with the maneuver when necessary, such as roll over the
toe, keep the trunk flexed forward, and keep the movement
continuous (i.e. do not hesitate). No cues were given during

Downloaded from poi.sagepub.com at GEORGIAN COURT UNIV on March 3, 2015

Agrawal et al.
testing. During descent, the physical therapist used a gait
belt to slow the transition of body weight over the pros-
thetic foot. Verbal cueing was provided to encourage more
time over the prosthetic heel (for the cushion feet design)
and over the forefoot for maximum deflection of the foot-
plate. With all four prosthetic feet, a slow and controlled
decent down the ramp with both limbs was promoted.
During each 1-h training session, subject’s gait profi-
ciency was assessed on predefined criteria. They were
trained only in areas in which training was needed. When
they satisfactorily met the predefined criteria, training
was concluded. All training sessions were standardized
and were administered by the same physical therapist.
Data analysis
After the subject’s third step with the prosthetic limb from
ascent or descent initiation, the next three consecutive
steps with both the intact and prosthetic limbs were uti-
lized for work calculations. External work was calculated
using the following equation
t2
 
∫
W = F ⋅ ds
t1

where W is the work in Joules, F is the GRF vector in

Newton, s is the displacement of CoM in meter, t1 is the
time at the beginning of a step, and t2 is the time at the end
of a step. GRFs obtained from F-scan sensors were used to
calculate vertical CoM acceleration using the following
equation
av =
( Fv − mg )
m except for the SACH foot (p = 0.02).
where av is the acceleration in the vertical direction (m/s2),
Fv is the vertical GRF (N), m is the body mass (kg), and g
is the acceleration due to gravity (m/s2).
The acceleration thus calculated was integrated to obtain
CoM velocity in the vertical direction. By integrating velocity
and using the geometry of the ramp, vertical CoM displace-
ment was calculated. Since F-scan sensors measure only the
normal force, vertical force was obtained from normal force
by using cosine of 5°, which was the angle of the ramp.
SEW between limbs was calculated for each stride
using the following equation,14 and a mean SEW value
was obtained by averaging the three strides
 WI − W p 
SEW = 100 − 100 ×  
 WI + W p 
 
where SEW is expressed in percentage; WI is the work
done by the intact limb due to vertical GRF, in Joules; WP
is the work done by the prosthetic limb due to vertical
GRF, in Joules.
Downloaded from poi.sagepub.com at GEORGIAN COURT UNIV on March 3, 2015
5
A SEW value of 100% indicates equal work by each
limb, value greater than 100% indicates more work by
the prosthetic limb, and a value less than 100% indicates
more work by the intact limb. Shapiro–Wilk test for nor-
mality of SEW values was performed for the sample
based on K-level, at Baseline, after initial prosthetic
training, and after prosthetic foot prescription. The intra-
class correlation coefficient (ICC) was calculated to
determine the test–retest reliability of the SEW values
using Model 1,3. A paired t-test was used to determine
differences in gait symmetry between the Baseline and
Training sessions to test the effect of standardized func-
tional prosthetic training. Repeated measures analysis of
variance (ANOVA) was used to determine differences
between feet after completion of the training protocol.
When ANOVA revealed significant differences, for post
hoc analysis, pairwise comparisons using related pairs
t-tests were used to identify the source of differences.
The critical alpha was set at 0.05.
Results
Subject characteristics for the reliability and intervention
studies are described in Tables 2 to 4, respectively.
Shapiro–Wilk test of normality
The results for the Shapiro–Wilk test for normality indi-
cate that the SEW values were normally distributed for
K-Level-2 subjects during decline walking and for
K-Level-3 subjects during incline and decline walking.
The SEW values for K-Level-2 subjects during incline
walking were normally distributed for all conditions
Reliability study
The ICC for test and retest SEW values was 0.87 for ramp
ascent and 0.89 for ramp descent. The high ICC indicated
that SEW can be measured consistently using F-scan sen-
sors with unilateral TTA’s walking on ramps and also cor-
roborated the stability of subjects’ gait during ascent and
descent.
Intervention study
Tables 4 and 5 show the mean (SD) SEW values during
ramp ascent and descent for both groups at the six testing
sessions.
Ramp ascent. After receiving standardized functional
prosthetic training, K-Level-3 subjects had a significant
(p < 0.05) increase in SEW values (16% increase over
Baseline). In the K-Level-2 subjects, there was a 5%
increase in SEW values following Training, which was
6 Prosthetics and Orthotics International

Table 4. Mean ± SD SEW values (in percentage) in K-Level-2 and K-Level-3 amputee subjects before and after standardized gait
training for ramps.

K-Level-2 subjects K-Level-3 subjects

Ascending Descending Ascending Descending

ramp SEW ramp SEW ramp SEW ramp SEW
(%) (%) (%) (%)
Baseline 67.91 ± 15.2 72.15 ± 21.3 72.87 ± 8.60 75.57 ± 12.0
Training 71.04 ± 18.3 80.9 ± 12.2 88.72 ± 16.0a 92.77 ± 5.4a

SD: standard deviation; SEW: symmetry in external work.

aSignificantly different from Baseline (p = 0.05).

Table 5. Mean ± SD SEW values (in percentage) for K-Level-2 and K-Level-3 amputee subjects while ascending and descending a
24-foot ramp with four test feet.

K-Level-2 subjects K-Level-3 subjects

Ascending Descending Ascending Descending

ramp SEW ramp SEW ramp SEW ramp SEW
(%) (%) (%) (%)
K1_Foot 74.03 ± 13.9 87.43 ± 5.0 78.08 ± 13.8 80.3 ± 11.9
K2_Foot 73.61 ± 21.7 80.02 ± 15.4 75.84 ± 8.6 79.50 ± 14.4
K3_Foot 83.51 ± 11.2 101.49 ± 7.7b,c 87.22 ± 23.1 96.13 ± 7.3b
MPA_Foot 74.76 ± 10.6 98.28 ± 6.1b 86.26 ± 19.8 87.47 ± 13.3

SD: standard deviation; SEW: symmetry in external work; MPA: microprocessor-controlled ankle.
bSignificantly different from the K1 foot (p = 0.05).
cSignificantly different from the K2 foot (p = 0.05).

not statistically significant (p < 0.05). There were no sig-
nificant differences between any pair of test feet.
Ramp descent. Following prosthetic training, there was a
significant increase in SEW values for the K-Level-3 group
(23% over Baseline), while in the K-Level-2 group (12%
over Baseline) increase in SEW was not statistically signifi-
cant (p < 0.05). In K-Level-2 subjects, a higher SEW value
with the Talux K3 foot (p < 0.05) was found when compared
to both the K1 and K2 feet. The MPA foot had significantly
higher SEW than the K1 foot (p < 0.05) and the difference in
SEW between MPA foot and the K2 foot did not reach statis-
tical significance (p = 0.06). In the K-Level-3 subjects, SEW
values were significantly higher with the K3 foot (p < 0.05)
when compared to the K1 foot. Again the K3 foot approached
statistical significance, but was not significantly different
from the K2 foot (p = 0.07). The Talux foot had the greatest
SEW values in four of the five K-Level-2 subjects and in
four of the five K-Level-3 subjects. There were no signifi-
cant differences between the K3 foot and the MPA foot.
Discussion
Ramp ascent
During ramp ascent, there were no significant differ-
ences in SEW values between prosthetic feet in either
K-Level-2 or K-Level-3 groups; however, the K3 and
MPA feet resulted in greater SEW values in the majority
of subjects in both groups. Vickers et al.9 reported sig-
nificant variations in GRFs during incline and decline
walking, not only with the prosthetic limb but also with
the intact limb. The results of this study demonstrate that
adaptations that impact external work due to vertical
force while ascending a 5° ramp are not influenced by
the type of prosthetic foot used. Although not an original
objective of the study, it was found that the prosthetic
training administered to the subjects resulted in numeri-
cally higher SEW values for the K-Level-2 subject group
and significantly higher SEW values for the K-Level-3
group. The methodology for this study ensured that all
subjects received training, specific to each prosthetic
foot’s design in order to maximize their performance
with the foot’s functional properties. As a result, the
improved symmetry following the ramp ascent was
maintained with all test feet.
During incline walking, lower limb muscles generate
power that is required to move the body’s CoM against the
force of gravity. Since the test prosthetic feet were not
designed to generate power, external work with the pros-
thetic limb may have been accomplished through the use of
residual limb musculature. During the standardized pros-
thetic training, subjects were taught to appropriately con-
tract the knee and hip extensor muscles as well as properly

Downloaded from poi.sagepub.com at GEORGIAN COURT UNIV on March 3, 2015

Agrawal et al.
contract their residual limb muscles to maintain balance
and use each prosthetic foot as designed for performing the
activity. Therefore, following the completion of the training
protocol, SEW values improved in both K-Level-2 and
K-Level-3 subjects and the test feet were not found to be
different during ramp ascent.
Contrary to the results of Fradet et al.18—who used a
7.5° slope—this study did not find a significant difference
between the Proprio-Foot and other test feet during incline
walking. The benefits of active dorsiflexion of the Proprio-
Foot may become more apparent on ramps with higher
incline angles (e.g. a 7.5° slope, as described by Fradet
et al.) or a larger sample size may be needed to detect sig-
nificant differences with the Proprio-Foot on a 5° incline.
Ramp descent
Both the K-Level-2 and K-Level-3 groups demonstrated
highest SEW values with the K3 foot, with statistically sig-
nificant differences between the K3/MPA feet and K1/K2
feet. Standardized training also resulted in statistically
higher SEW for the K-Level-3 group and numerically
higher symmetry in the K-Level-2 group. The differences
in SEW values could be attributed to the design and func-
tional properties of the K3 feet.
The K3 foot (Talux foot) used in this study has a dynamic
heel that promotes plantarflexion during early stance and a
J-shaped ankle design that simulates anatomical tibial
advancement (or dorsiflexion) as the body progresses over
the stationary foot in late stance.19 Macfarlane et al.20
reported that amputees capable of balancing over the fore-
foot during late stance with a Flex-Foot were able to take a
slower and longer step with the intact limb resulting in
greater symmetry during ambulation. The similarity in
designs between the Flex-Foot and the Talux foot indicates
that the temporal gait characteristics seen with these feet
would be comparable. In this study, at the time of training,
subjects were instructed to balance over the prosthetic fore-
foot and to take a controlled, slower step with the intact
limb. The combination of training and functional properties
of the prosthetic foot resulted in similar time needed to
reach the peak force (loading time) with the intact and pros-
thetic limbs, in both subject groups (Table 6). Also, the for-
ward deflection of the J-shaped ankle spring likely permits
a systematic lowering of the CoM in terminal stance, which
allows a smooth transfer of weight to the intact limb. This
functional property of the K3 foot resulted in reduced load-
ing of the intact limb as the first peak of the vertical GRF
curve with the K3 foot found to be lower than K1 and K2
feet in both subject groups (Table 7). The symmetry in ver-
tical GRF peak and loading times between the intact and
prosthetic limbs further corroborates the high SEW values
recorded with the K3 foot.
The MPA foot (Proprio-Foot) also has a heel–toe foot-
plate design, which permits late stance stability, and an
MCA joint that actively plantarflexes in late swing to
Downloaded from poi.sagepub.com at GEORGIAN COURT UNIV on March 3, 2015
7
match the grade of the slope. The Proprio-Foot’s ankle
joint has limited dorsiflexion during the stance period
which likely impacts the transition of body weight over the
stationary foot during the late stance period. The differ-
ences in functional ankle dorsiflexion between the K3 and
MPA feet may explain the observed differences in SEW
values, vertical GRF peak values, and loading times
between these DR feet (Tables 4 to 7).
In both the non-DR feet, the pylon is directly attached
to the foot without a mechanical axis at the ankle. This
rigid ankle design eliminates functional ankle dorsiflex-
ion during the stance period of gait. The absence of dorsi-
flexion during stance causes premature knee flexion
resulting in an early heel rise.19 Because of the shorter,
flexible keel, the body tends to progress more rapidly to
terminal stance as the CoM essentially runs out of base-
of-support. The maximum forward progression of the
center of pressure (CoP) (expressed as a percentage (%)
of foot length) was approximately 70% and 79% with the
SACH and SAFE, respectively, compared to 85% with
Talux/Proprio feet (Table 8). The resulting instability dur-
ing late stance likely caused the contralateral limb to
advance more rapidly to catch the body as it fell forward.
The sudden deceleration of body CoM as it transitioned
from the prosthetic to intact limb resulted in a greater ver-
tical GRF peak on the intact limb—particularly in the K2
group—and a faster loading time on the intact limb
(Tables 6 and 7). The functional properties of non-DR feet
thus resulted in greater work by the intact limb and lower
SEW values with K1/K2 feet in both groups. These results
suggest that K3 prosthetic feet can have significant func-
tional benefits for those K-Level-2 unilateral TTAs who
frequently negotiate ramps, as the K3 feet promote greater
SEW during decline walking.
The SEW measure determines the similarities in GRFs,
CoM displacement, and step time between the intact and
prosthetic limbs. Since these gait parameters are influenced
by the design related to prosthetic foot category, the foot
design that mimics the function of physiological foot/ankle
during ramp walking would result in greater inter-limb sym-
metry. In addition to the contributions from the prosthetic
foot, the use of residual limb musculature can significantly
influence gait dynamics while ascending and descending a
ramp. During ramp descent, the SEW measure is correlated
with traditional methods of gait assessment, such as the ver-
tical GRF peak (r = −0.5) and loading time of the intact limb
(r = 0.9), as well as the maximum forward progression of the
CoP on the prosthetic foot (r = 0.7). Subjects with low SEW
values demonstrated greater vertical GRF peak and faster
loading on the intact limb, as well as shorter CoP progres-
sion on the prosthetic foot. However, unlike these traditional
methods—which assessed discrete points of the stance
period and could only indicate trends about prosthetic feet—
the SEW measure was able to detect statistically significant
differences between feet. As the SEW measure combines
different gait parameters during the entire stance period into
8 Prosthetics and Orthotics International

Table 6. Mean ± SD time required to reach the first peak of vertical ground reaction force (expressed as percent of gait cycle) in
K-Level-2 and K-Level-3 amputee subjects while descending a 24-foot ramp.

K-Level-2 subjects K-Level-3 subjects

Amp limb Sound limb Prosthetic limb Intact limb

(% GC) (% GC) (% GC) (% GC)
K1_Foot 16.75 ± 3.3 14.66 ± 3.1 17.93 ± 3.8 13.88 ± 3.2
K2_Foot 17.67 ± 3.6 14.78 ± 3.1 19.56 ± 3.3 13.66 ± 4.4
K3_Foot 18.09 ± 5.3 18.19 ± 4.8 18.32 ± 4.4 17.65 ± 2.3
MPA_Foot 17.44 ± 4.1 17.64 ± 2.2 17.88 ± 2.1 16.41 ± 2.8

SD: standard deviation; GC: gait cycle; MPA: microprocessor-controlled ankle.

Table 7. First peak of vertical ground reaction force (mean ± SD), in percent body weight of K-Level-2 and K-Level-3 amputee
subjects while descending a 24-foot ramp.

K-Level-2 subjects K-Level-3 subjects

Amp limb force Sound limb force Amp limb force Sound limb force
(% BW) (% BW) (% BW) (% BW)
K1_Foot 115.07 ± 26.5 144.44 ± 24.5 103.44 ± 14.3 102.74 ± 18.6
K2_Foot 129.20 ± 27.3 120.38 ± 12.9 109.66 ± 21.1 101.63 ± 8.6
K3_Foot 110.92 ± 22.9 111.16 ± 28.1 108.9 ± 14.1 97.43 ± 10.3
MPA_Foot 109.18 ± 10.9 111.14 ± 24.7 110.26 ± 16.2 106.7 ± 6.5

SD: standard deviation; BW: body weight; MPA: microprocessor-controlled ankle.

Table 8. Forward progression of the center of pressure
(mean ± SD), expressed as a percentage of foot length in
K-Level-2 and K-Level-3 amputee subjects while descending a
24-foot ramp.
K-Level-2 subjects K-Level-3 subjects
(foot length %) (foot length %)
K1_Foot 70.51 ± 1.9 69.61 ± 1.67
K2_Foot 80.25 ± 4.1 76.96 ± 4.7
K3_Foot 86.47 ± 3.1 85.4 ± 1.6
MPA_Foot 85.22 ± 1.6 84.9 ± 2.0
SD: standard deviation; MPA: microprocessor-controlled ankle.
a single value, it appears to have the sensitivity for detecting
subtle gait differences between prosthetic feet on ramp
incline and decline.
Since increased loading of the intact limb and high
loading rates are associated with the higher prevalence of
osteoarthritis in the TTA population,21 the SEW measure
could potentially be used as a measure for determining the
efficacy of prosthetic feet in a clinical setting. Currently
available self-report outcome measures only assess per-
ceived difficulty associated with ramp ambulation and the
performance-based measures do not quantify movement
during incline and decline walking. Since the SEW meas-
ure can quantify gait asymmetry during ramp ambulation,
it could be used to supplement existing clinical outcome
measures. With the recent advancements in wireless in-
sole technology and mobile computing, the SEW measure
could potentially be implemented in a clinical setting rela-
tively easily. Future research to test this measure on a
larger population of subjects with different amputation
levels is needed to further determine the clinical applica-
bility of this measure with regard to different pathologies
such as osteoarthritis or neuropathic foot disease.
Limitations
The study was underpowered in terms of sample size.
While SEW values were significantly different between
prosthetic feet during ramp descent, differences during
ramp ascent may not have been evident due to a relatively
small size. In addition, the accommodation time with pros-
thetic feet and minimal exposure to incline/decline ambu-
lation during the accommodation period due to
geographical topography of Miami may have reduced the
opportunities for practice between testing sessions.
Standardized prosthetic ramp training had an impact on
performance because it was specific to the design of each
foot and promoted use of each component, such as a cush-
ion heel, rubber toes, heel-to-toe footplate, foot keel,
J-shaped design, and microprocessor ankle. Since ramp-
specific training is not always available in all clinical envi-
ronments, the clinically observed differences between
feet may vary as some feet could be more intuitive to use
than others. Future work could examine the differences
of SEW in TTAs who received training to maximize the
use of a prosthetic foot’s design properties for ramp
ambulation as opposed to those TTA who naturally

Downloaded from poi.sagepub.com at GEORGIAN COURT UNIV on March 3, 2015

Agrawal et al.
adapted to prosthetic foot without formal training; that is,
is there a prosthetic foot design that promotes SEW during
ramp ambulation and does not require training?
Similar to other commercially available in-sole sensor
systems, the F-scan sensors can only measure normal
forces during gait. In able-bodied gait on sloping surfaces,
the magnitude of normal and vertical forces is very simi-
lar—differing by less than 1%—while ascending and
descending a 5° ramp.20 The shear force also has a compo-
nent in the vertical direction, which is the trigonometric
sine of the angle of the slope. For a 5° ramp, this compo-
nent of shear force has a negligible contribution to the ver-
tical force, and hence was ignored in this analysis. Future
studies could examine SEW differences between feet at
higher grades of ramps, as the vertical and shear GRFs
change appreciably with the angle of inclination.8
The inability of F-scan sensors or any other currently
available in-sole sensor to measure shear forces limits our
ability to include shear values in the SEW calculation.
Since the GRFs during ramp negotiation are three-
dimensional in nature and the CoM moves in all three
planes, total work during ramp ambulation could be com-
puted using both the vertical and shear forces. Although
the magnitude of shear GRFs is much lower than the ver-
tical GRFs,22 the contribution of work due to shear forces
on the total work during ramp ascent and descent is
unknown. Moreover, subjects were asked to walk at a
self-selected speed and walking speed was not controlled
during incline or decline walking. While walking speeds
may have influenced SEW values, the effect of speed on
normal and shear GRFs in unilateral TTAs walking on
slopes is unknown. Future work to include shear forces in
the SEW calculation may be beneficial for determining
the aggregate effects of forces on the foot and lower limb
during ramp ambulation.
Conclusion
SEW was found to be different between four categories of
prosthetic feet during ramp descent of K-Level-2 unilat-
eral TTAs. The design features of K3 DR prosthetic feet
improved dynamic gait symmetry in K-Level-2 subjects,
while descending a 5° ramp compared to K1 and K2 feet.
The J-shaped ankle design of the K3 foot (Talux) resulted
in significantly greater SEW values than both the K1 and
K2 feet (SACH and SAFE, respectively), while the micro-
processor ankle (Proprio-Foot) had a significantly greater
symmetry than the K1 foot. The test feet were not signifi-
cantly different in either K-Level-2 or K-Level-3 subjects
while ascending the ramp. In K-Level-3 TTAs, SEW val-
ues increased significantly following standardized gait
training for ramp ascent and descent. Study results suggest
that both K-Level-2 and K-Level-3 amputees experience
greater SEW from a K3 prosthetic foot with a J-shaped
ankle and heel-to-toe footplate design as well as a micro-
processor foot while descending ramps.
Downloaded from poi.sagepub.com at GEORGIAN COURT UNIV on March 3, 2015
9
Author contribution
All authors contributed equally in the preparation of this
manuscript.
Declaration of conflicting interests
The authors declare that there is no conflict of interest.
Funding
This material is the result of the work supported with resources
and use of the Bruce W Carter Department of Veterans Affairs
Medical Center, Miami, FL, USA. The authors would like to
thank Ossur® Americas for their support of this research study
and the South Florida Veterans Affairs Foundation for Research
and Education, Miami, FL, USA, for administering the research
project.
References
1. HCFA Common Procedure Coding System HCPCS 2001.
Washington, DC: US Government Printing Office, 2001.
2. Medicare Region C. Durable Medical Equipment Prosthetics
Orthotic Supplier (DMEPOS) manual. Columbia, SC:
Palmetto GBA, 2005, pp. 53.5–53.6.
3. Gailey R, Allen K, Castles J, et al. Review of secondary
physical conditions associated with lower-limb amputa-
tion and long-term prosthesis use. J Rehabil Res Dev 2008;
45(1): 15–29.
4. Royer TD and Wasilewski CA. Hip and knee frontal plane
moments in persons with unilateral, trans-tibial amputation.
Gait Posture 2006; 23(3): 303–306.
5. Agrawal V, Gailey R, O’Toole C, et al. Influence of gait
training and prosthetic foot category on external work sym-
metry during unilateral transtibial amputee gait. Prosthet
Orthot Int 2013; 37(5): 396–403.
6. Morgenroth DC, Segal AD, Zelik KE, et al. The effect of
prosthetic foot push-off on mechanical loading associated
with knee osteoarthritis in lower extremity amputees. Gait
Posture 2011; 34(4): 502–507.
7. Powers CM, Torburn L, Perry J, et al. Influence of pros-
thetic foot design on sound limb loading in adults with uni-
lateral below-knee amputations. Arch Phys Med Rehabil
1994; 75(7): 825–829.
8. McIntosh AS, Beatty KT, Dwan LN, et al. Gait dynamics on
an inclined walkway. J Biomech 2006; 39(13): 2491–2502.
9. Vickers DR, Palk C, McIntosh AS, et al. Elderly unilateral
transtibial amputee gait on an inclined walkway: a biome-
chanical analysis. Gait Posture 2008; 27(3): 518–529.
10. Perry J and Shanfield S. Efficiency of dynamic elastic
response prosthetic feet. J Rehabil Res Dev 1993; 30(1):
137–143.
11. Gitter A, Czerniecki JM and DeGroot DM. Biomechanical
analysis of the influence of prosthetic feet on below-knee
amputee walking. Am J Phys Med Rehabil 1991; 70(3):
142–148.
12. Hafner BJ, Sanders JE, Czerniecki J, et al. Energy storage
and return prostheses: does patient perception correlate
with biomechanical analysis? Clin Biomech 2002; 17(5):
325–344.
13. Van der Linde H, Hofstad CJ, Geurts AC, et al. A system-
atic literature review of the effect of different prosthetic
10
components on human functioning with a lower-limb pros-
thesis. J Rehabil Res Dev 2004; 41(4): 555–570.
14. Agrawal V, Gailey R, O’Toole C, et al. symmetry in exter-
nal work (SEW): a novel method of quantifying gait dif-
ferences between prosthetic feet. Prosthet Orthot Int 2009;
33(2): 148–156.
15. Agrawal V, Gailey R, Gaunaurd I, et al. Comparison
between microprocessor-controlled ankle/foot and conven-
tional prosthetic feet during stair negotiation in people with
unilateral transtibial amputation. J Rehabil Res Dev 2013;
50(7): 1–10.
16. Koch M. Measuring plantar pressure in conventional shoes
with the TEKSCAN sensory system. Biomed Tech 1993;
38(10): 243–248 (Plantardruckmessung im Konfektionsschuh
mit der TEKSCAN Sensorfolie).
17. Mueller MJ and Strube MJ. Generalizability of in-shoe peak
pressure measures using the F-scan system. Clin Biomech
1996; 11(3): 159–164.
Downloaded from poi.sagepub.com at GEORGIAN COURT UNIV on March 3, 2015
Prosthetics and Orthotics International
18. Fradet L, Alimusaj M, Braatz F, et al. Biomechanical
analysis of ramp ambulation of transtibial amputees with
an adaptive ankle foot system. Gait Posture 2010; 32(2):
191–198.
19. Torburn L, Perry J, Ayyappa E, et al. Below-knee amputee
gait with dynamic elastic response prosthetic feet: a pilot
study. J Rehabil Res Dev 1990; 27(4): 369–384.
20. Macfarlane PA, Nielsen DH, Shurr DG, et al. Gait compari-
sons for below-knee amputees using a flex-foot(TM) versus
a conventional prosthetic foot. J Prosthet Orthot 1991; 3(4):
150–161.
21. Morgenroth DC, Gellhorn AC and Suri P. Osteoarthritis in
the disabled population: a mechanical perspective. PM R
2012; 4(5 Suppl.): S20–S27.
22. Zmitrewicz RJ, Neptune RR, Walden JG, et al. The effect of
foot and ankle prosthetic components on braking and pro-
pulsive impulses during transtibial amputee gait. Arch Phys
Med Rehabil 2006; 87(10): 1334–1339.