Gait Comparisons for Below-Knee
Amputees Using a Flex-Foot(TM) Versus a
Conventional Prosthetic Foot
Pamela A. Macfarlane, Ph.D.
David H. Nielsen, L.P.T., Ph.D.
Donald G. Shurr, L.P.T., c.o.
Kenneth Meier, C.P.
Introduction
Whether limb loss is due to trauma or disease, a common goal of below-knee (BK)
amputees is to walk without endangering
their stability and with a gait that appears as
normal as possible (1,2). The design of BK
prosthetic feet needs to accommodate these
goals so that each limb can provide the control and support forces necessary for walking
over a range of grades and speeds. In addition, an asymmetrical gait has been linked to
an increase in the prevalence of degenerative
changes in the lumbar spine and knees (3).
Hurley et al. suggested that the asymmetrical
gait could tend to increase the joint forces on
one side and possibly predispose that side to
premature degenerative arthritis (2). A
prosthetic foot that can optimize the subject's gait in view of the above concerns
would be expected to decrease the perceived
and physiological difficulty of ambulation
that arises as a result of the loss of motor and
sensory nervous control with amputation.
The walking gait of BK amputees differs
significantly from that of normals. The selfselected walking velocity (S-SWV) is slower
due to shorter step lengths and a slower cadence (4,5,6,7,8). The BK walking gait is
also asymmetrical (2,6). Robinson et al.
found the distance stepped with the involved
(prosthetic) foot was longer than that with
the uninvolved (normal) foot (6). The type
of prosthetic foot was not found to substantially affect the walking gait when subjects
used either the Solid Ankle Cushion Heel
(SACH) foot or the uniaxial foot, which in
the past have been the most common designs
in the conventional prosthetic foot (CF) used
in the United States and Britain, respectively
(9,4,10).
Dynamic prosthetic feet, designed to store
and release energy during ambulation, have
recently become available due to the development of elastic lightweight materials like
those used in skis and tennis rackets. The
Flex-Foot?(FF) is one of these dynamic
prosthetic feet composed of a fiberglass and
carbon (graphite) shaft, which compresses
during heel contact and extends during heel
off (11). The design is intended to increase
the forward push-off of the user and assist in
the involved foot's recovery. Preliminary research has shown that FF walking is associated with faster S-SWVs and better gait efficiency than CF walking at speeds of 67 m/
mm and above and that subjects prefer the
FF for faster level walking and for incline
walking (12). No research was found that
compared the kinematics of the walking gait
with an FF to that with a CF during walking
over different grades or speeds.
The purpose of this study was to identify
differences in physical gait performance of
BK amputees during FF and CF walking
over a functional range of speeds and grades.
The focus was on linear, temporal and gaitsymmetry variables.
MethodDesign
Experimental design was based on subjects' measured performance in tests taken
on each subject for three selected walking
speeds (slow, medium and fast) over three
different grades (level, decline and incline)
on a motor-driven treadmill while subjects
wore the FF and CF. The S-SWV with each
prosthetic foot was also determined for each
individual during overground walking.
Subjects
The seven male subjects in the study were
paid volunteers who were unilateral traumatic BK amputees and considered by a certified prosthetist to be good ambulators with
both the CF and the FF. All subjects were
proficient treadmill walkers, having been
subjects in previous research involving
treadmill walking. The same prosthetist
fitted all subjects with the FF. Table 1
shows
descriptive data of the subjects. An admissions criterion was that the subjects have an
FF as well as a CF. Since we are located in a
relatively rural area and the subjects were
required to commute daily to and from the
test site, a maximum driving distance of 150
miles was adopted. Both of these criteria
limited the availability of subjects.
Prior to testing, the subjects were
screened to assess their general health and
ability to perform the walking tests, and each
subject was fully informed of the test
procedures. In accordance with the Human
Subjects Review Committee of the College
of Medicine at The University of Iowa, written consent for participation in the study was
obtained from each subject.
During testing the subjects wore shorts,
socks and their normal walking footwear. Six
of the subjects wore matching shoes. The
seventh did not normally wear a shoe with
his FF, so he walked without a shoe or sock
on that foot.
Equipment
A 15-meter-long segmental walkway that
has an electronic timer with portable lights
and photo conductive switches was used to
determine the S-SWVs. All other gait measurements took place while the subjects
walked on a motor-driven treadmill. A 16mm DC Locam camera (Photo Instrument
Division, Redlake Corp., Santa Clara,
Calif.) set to record at 100 frames/second
was positioned 8.85 m from the treadmill
with its optical axis perpendicular to the center of the treadmill belt. An internal timing
light, pulsing at 100 Hz, marked the film as it
ran through the camera. A vertical linear
scale was positioned in camera view 9.62 m
from the camera for use in the data reduction
and conversion of the image data to metric
units. A mirror was placed behind the subject and oriented so that a rear view of the
subject was simultaneously recorded on the
film by the camera. The subjects wore a
lightweight styrofoam dorsal fin attached by
means of a belt worn around the waist, which
was used to facilitate vertical trunk displacement measures.
Protocol
A preliminary practice session was provided to each subject to orient him to the required testing procedures and to allow him
to become familiar with walking on the
treadmill under the conditions of the study.
During this preliminary session, the S-SWV
was measured for both prosthetic feet based
on our standard protocol (12). Eive time
measurements were taken over the mid fivemeter section of the walkway after the subjects had walked at their own pace for five
minutes. The average of the five time measurements was used to calculate the S-SWV.
After the preliminary session, each subject completed two testing sessions on the
treadmill, three subjects using the CF first
and four using the FF first. This arrangement
was determined by random selection without
replacement. An attempt was made to have
subjects complete only one session per day;
however, due to scheduling conflicts, three
subjects completed all the testing on the
same day. The certified prosthetist considered each of these subjects well able to handle the double testing sessions on the same
day. These subjects were given at least a
three-hour recovery break between sessions
and did not experience difficulty completing
the second session. Each session consisted of
a level (0.0 percent grade), a decline (-8.5
percent grade), and then an incline (+8.5
percent grade) walking test with at least 30
minutes of recovery between each of the
three tests. The 8.5 percent grade magnitude
is the maximum grade suggested for use in
public building construction (13). Each test
consisted of a warm-up of three minutes followed by approximately three-minute walking bouts at slow (53.6 m/min = 2.0 mph),
medium (67.0 m/min = 2.5 mph), and fast
(80.5 m/min = 3.0 mph) speeds. The three
walking speeds were selected to represent a
functional range of walking velocities. The
S-SWVs were used as a guideline to assess
the maximum speed that appeared to be
comfortably attainable by all subjects at each
of the three grades.
Under each grade and speed condition,
subjects were filmed for three gait cycles after having walked for two minutes to adjust
to the grade and speed of the treadmill. The
subjects were asked to walk normally, and if
any atypical gait was noticed during recording, the filming was repeated.
Data Reduction
All data were obtained from the film,
which was projected onto an automated digi
tizing board so that an image of the subject
approximately 20 cm high was available for
analysis (Vanguard Projector, Model
MI6CP). The step lengths used in the data
analysis represented a mean of three consecutive steps taken by each leg. Each step length
was the scaled distance between the heel of the
backward foot to the heel of the forward foot
while the subject was in the double support
phase (Figure 1)
. The scaled measurements
were converted to meters using a scaling factor
calculated for each film segment.
The gait cycle was divided into the 10 temporal phases shown in Figure 2
, which were
determined by the critical instances of heel
strike, toe off, midstance and midswing for
the involved and uninvolved legs. The timing
light marks on the film were used to assure
that only film recorded at 100 frames/second
was used, which meant each frame was .01
seconds later than the previous one. The
number of the film frame that identified each
critical instance of the gait was recorded.
The duration of each phase was determined
as the difference in the frame numbers of the
two critical instances that limited each
phase. Two consecutive gait cycles were re
corded in this way; the mean of the two measurements was used in the analysis.
Symmetry ratios for step lengths and intracycle gait phases were calculated by dividing
the value of the variable relative to the involved foot by that of the uninvolved foot. In
this way, bilateral symmetrical walking
would be associated with a symmetry ratio of
1.0, and asymmetrical walking would have a
ratio of greater or less than 1.0.
To determine the vertical trunk displacement, a mark on the base of the dorsal fin,
viewed from the side, was traced while the
film ran slowly through the projector for the
duration of at least three gait cycles. The
vertical trunk displacement measure represented the scaled difference between the
traced marker at its highest and at its lowest
vertical position from the treadmill surface.
Scaled measurements were converted to
centimeters using the scaling factor previously presented. To identify the part of the
walking cycle where the vertical limits occurred, the path taken by the base of the fin
was traced from the mirror image (a rear
view) for the duration of at least one walking
cycle for each condition.
Statistical Analysis
Descriptive statistics were calculated on
all variables. A repeated measures analysis
of variance (ANOVA) with three levels
(foot type, grade and speed) was used to
analyze the gait data. An alpha-level of 0.05
was adopted for determining statistical significance. The Bonferroni adjustment procedure was used to control for possible type I
errors associated with the testing of multiple
variables (14). The alpha level (0.05) was
divided by the number of variables under
study (20), which resulted in using an
adjusted p-value of < 0.0025 to determine
the statistical significance of the foot type
effect.
Six subjects completed all testing satisfactorily, and the seventh completed all conditions but was unable to walk steadily on the
incline at the fast speed with his CF. The
data for this subject under this condition
were considered too variable and were not
used in the analysis. This resulted in nine
foot type comparisons (three speeds at three
grades) for six subjects and eight for the subject who did not complete the incline test.
Results
Figure 3
, Figure 4
, Figure 5
, Figure 6
, Figure 7
, Figure 8
, and Figure 9
represent graphs of the
descriptive statistics for the step length, temporal phase and vertical displacement variables. From these graphs the influences of
foot type, grade and speed can be seen on
each variable. In general the grade and
speed conditions affected the mean values of
the variables in a similar manner regardless
of the prosthetic foot being used.
The ANOVA revealed no significant interactions between foot type and speed or
grade. For this reason, the main effect of
foot type was assessed for all variables across
the grade and speed conditions. The summary of means, standard errors and p-values
according to foot type are presented in Table
2
. It can be seen that the foot type condition
had little effect on the involved step length;
however, the uninvolved step length was significantly longer under the FF condition than
the CF condition. The longer stride (involved step length + uninvolved step
length) with the FF allowed the subjects to
take fewer steps per minute than with the CF
yet still walk at similar speeds. This is reflected in the increase in the cycle time with the
FF compared to the CF, which approached
significance.
The analysis of the intracycle gait phases
identified the specific parts of the gait that
were significantly affected by the choice of
prosthetic foot. The involved late stance
phase, the involved early swing phase and
the uninvolved late swing phase were all of
significantly longer duration with the FF
than the CF. The foot type influence on the
uninvolved early stance phase approached
significance. The symmetry ratios of the late
stance phase and the late swing phase were
significantly better with the FF than the CF;
however, the early swing ratio was significantly more symmetrical with the CF. The
step length ratio approached significance in
favor of the CF. There was significantly less
vertical trunk displacement during FF walking than CF walking. The cause of this difference is illustrated in Figure 10
, taken from a
typical tracing of the movement of the base
of the fin viewed from the rear mirror image.
It can be seen that the source of the increase
in vertical trunk motion due to the choice of
foot type was the additional lowering of the
trunk that occurred immediately after uninvolved heelstrike during CF walking.
DiscussionOverall Gait Characteristics
The subjects in this study spent less time
on the involved foot in late stance than they
did on the uninvolved foot, which is in accordance with previous research on BK amputee walking (15,6). In particular, the period
of the late stance phase between uninvolved
heelstrike and involved toe off (the period of
uninvolved double support) was noticeably
shorter than the corresponding contralateral
period, resulting in poor double support
symmetry ratios for both foot types (1.43 for
the FF, 1.39 for the CF). Breakley suggested
the cause for the shortened late stance phase
was an early heel contact by the uninvolved
limb to decrease the time the body weight is
supported only on the involved limb (15).
While the late stance phase was shorter for
the involved foot than the uninvolved foot,
the early swing phase was longer for the involved foot. This suggests that after uninvolved heel strike, the subjects shifted their
weight quickly to the uninvolved foot and
recovered the prosthesis quickly off the
ground, but they took longer to get it to the
midswing position than they did with the uninvolved foot.
Foot Type
The significant foot type differences are
summarized in Figure 11
. As illustrated, several of the gait variables were significantly
affected by foot type and appear to be related to the subjects' willingness to spend more
time in the single support late stance phase
on the FF than on the CF. This period of the
gait is limited by the instances of involved
midstance and uninvolved heelstrike as indicated by the shaded area in Figure 11
. While
both phases between involved midstance
and involved midswing were longer with the
FF than the CF, the phase between uninvolved heelstrike and involved toe off (the
uninvolved double support phase) was not
significantly affected by foot type. This suggests the cause of the foot type difference
occurred early during the involved late
stance phase, which is also when the uninvolved foot is in the late swing phase. The
length of the uninvolved step is determined
during this phase and was longer when the
subjects wore the FF than when they wore
the CF. This was particularly evident during
level and incline walking, where the subjects
could stride out (Figure 3)
. Several subjects
expressed hesitancy during fast decline walking, which may have limited the length of
step they were willing to take under this condition.
Based on the main effect analysis for foot
type, the step length ratio (involved step
length/uninvolved step length) for the CF
was significantly higher than that for the FF,
but both were less than 1.0. This is in contrast to Robinson, Smidt and Aurora, who
reported a step length ratio of 1.07 (SD
0.08) for CF overground walking at S-SWV
(6). In an earlier study, we observed mean
step length ratios of 1.08 (SE 0.02) (for the
CF) and .99 (SE = 0.01) for the FF (16).
These statistics reflected ANOVA adjustments with S-SWV as the covariate.
The inconsistency between the results of
the present study and earlier reports may be
related to the difference in mode of walking
between the studies, i.e. overground versus
treadmill walking. Preliminary analysis of
the S-SWV step length data for treadmill
versus overground walking in the present
study has revealed that the subjects tended
to decrease their step lengths while walking
on the treadmill. There appears to have been
a disproportionately greater decrease in the
involved compared to the uninvolved step
lengths, which produced the net effect of
lower step length ratios (below 1.0) for both
prosthetic feet during the treadmill condition.
While there was no significant interaction
between foot type and grade, the step length
ratios for the CF and the FF were significantly lower for decline walking than for level or
incline walking, which may also explain why
the ratios were less than unity. Subsequently, care should be taken when comparing our
step length ratios to those in other studies
involving overground walking.
Although reasonably symmetrical, the
contralateral step lengths in the present
study were shorter than those reported for
normal subjects (7). The step lengths of BK
amputees are limited due to their reluctance
to bear weight solely on the prosthetic foot
and their need to protect the residual limb
during involved heelstrike. As the subjects
gain competence in single support on the
prosthetic foot, the uninvolved step length
would be expected to become longer and
more like the step length taken by someone
without an amputation. The results of our
study suggest that the dynamic action of the
FF facilitated a longer involved late stance
phase, resulting in longer uninvolved step
lengths than during CF walking. There was
no significant difference in the involved step
length due to foot type, which may have
been due to the amount of force the residual
limb could absorb at heelstrike regardless of
which prosthetic foot was being used. This
would account for the lack of significant foot
type differences in the involved step length.
The interaction of these foot type influences
explains the lower step length ratios for the
FF compared to the CF.
The involved early swing phase might
have been of longer duration during FF
walking than CF walking only as a result of
the longer involved late stance phase. The
FF would have a greater distance to cover
between involved toe off and mid swing. If
the speed of recovery of the prosthetic foot
remained the same, the greater distance
would necessitate a longer duration of early
swing phase. The uninvolved early stance
phase occurs simultaneously, so it is similarly
affected.
The increase in vertical trunk displacement during CF walking compared to FF
walking, caused by the lowering of the trunk
after uninvolved heelstrike, could be related
to the subjects' attempt to decrease the magnitude of the ground reaction force during
weight acceptance onto the uninvolved limb.
Hurley et al. found amputees experience significantly lower ankle and knee horizontal
components of joint reaction force on the
uninvolved side at weight acceptance than
non-amputees (2). They suggested that amputees walk more slowly to decrease the
forces acting across the joints of the uninvolved limb. In the present study, the subjects were forced to walk at matched speeds
on the treadmill with both types of prosthetic
feet and could not choose to walk slower
with the CF as they had during the S-SWV
test. They appear to have protected their
uninvolved leg during CF walking by increasing the time of weight acceptance by the
uninvolved leg. This resulted in a greater
lowering of the trunk and a significant increase in the mean vertical trunk displacement values. The foot type influence in the
vertical trunk displacement suggests difference in overall biomechanical efficiency and
gait symmetry in favor of the FF.
Summary
Several of the gait characteristics of the
unilateral BK amputees in this study differed
significantly when they used an FF compared
to a CF. The source of the difference appears
to have been the increased length of time the
subjects were willing to spend in single support on the FF as compared to the CF. The
increase in the duration of this phase with
the FF allowed them to take larger, more
normal steps with their uninvolved leg and,
hence, they needed fewer steps per minute
to match CF walking speeds. Walking with
the FF was associated with a smoother, more
uniform vertical trunk motion, which suggests an increase in biomechanical efficiency
when compared to CF walking.
Acknowledgments
This research was supported by a grant from
Flex-Foot, Inc., Irvine, Calif. Appreciation is extended to Dr. J.G. Hay for his advice and assistance during the planning and data collection of
this study, to Dr. J.C. Golden for her help during
data collection, and to the subjects who cooperated unbelievably.
Pamela A. Macfarlane, Ph.D., is assistant professor of physical education at Northern Illinois
University, DeKalb, Ill. 60115, and was a Ph.D.
student at The University of Iowa when this study
was conducted.
David H. Nielsen, L.P.T., Ph.D., is associate
professor of physical therapy education at the Division of Associated Medical Sciences/College of
Medicine for the University of Iowa, 2600
Steindler Building, Iowa City, Iowa 52242; (319)
335-9791.
Donald G. Shurr, L.P.T., C.O., is with American Prosthetics Inc. in Iowa City, Iowa, and an
adjunct lecturer for physical therapy education in
the Division of Associated Medical Sciences/College of Medicine for The University of Iowa, in
Iowa City, Iowa.
Kenneth Meier, C.P., is with American Prosthetics Inc. in Davenport, Iowa.
References:
- Fisher SV, Gullickson G. Energy cost of ambulation in health and disability: a literature review. Archives of Physical Medicine and Rehabilitation 1978; 59:124-33.
- Hurley GRB, MeKenney R, Robinson M,
Zadravec M, Pierrynowski MR. The role of the
contralateral limb in below-knee amputee gait.
Prosthetics and Orthotics International 1990;
14:33-42.
- Brouwer BJ, Allard P, Labelle H. Running
patterns of juveniles wearing SACH and singleaxis foot components. Archives of Physical Medicine and Rehabilitation 1989; 70:128-34.
- Doane NE, Holt LE. A comparison of the
SACH and single-axis foot in the gait of unilateral
below-knee amputees. Prosthetics and Orthotics
International 1983; 7:33-6.
- Huang C, Jackson JR, Moore NB, et al.
Amputation: energy cost of ambulation. Archives
of Physical Medicine and Rehabilitation 1979;
60:18-24.
- Robinson JL, Smidt GL, Aurora iS. Accelerographic, temporal and distance gait factors in below-knee amputees. Physical Therapy 1977;
57:8:898-904.
- Skinner HB, Effeney DJ. Gait analysis in
amputees. American Journal of Physical Medicine 1985; 64:2:82-9.
- Waters RL, Perry J, Antonelli D, et al. Energy cost of walking amputees: the influence of
level of amputation. The Journal of Bone and
Joint Surgery 1976; 58:1:42-6.
- Cuiham EC, Peat M, Newell E. Analysis of
gait following below-knee amputation: a comparison of the SACH and single-axis foot. Physiotherapy Canada 1984; 36:5:237-42.
- Goh JCH, Solomonidis SE, Spence WD,
Paul JP. Biomechanical Evaluation of SACH and
unilateral feet. Prosthetics and Orthotics International 1984; 8:147-54.
- Hittenberger D. Technical Manual, FlexFoot, Inc., 14 Hughes, Suite B210, Irvine, Calif.
92718, 1986.
- Nielsen DH, Shurr DG, Golden JC, Meier
K. Comparison of energy cost and gait efficiency
during ambulation in below-knee amputees using
different prosthetic feet: a preliminary report.
JPO Journal of Prosthetics and Orthotics 1989;
1:24-3 1.
- International Conference of Building Officials. Uniform building code, Whittier, Calif.
1985.
- Huberty CJ, Morris JD. Multivariate analysis versus multiple univariate analyses. Psychological Bulletin 1988; 105:2:302-8.
- Breakley J. Gait of unilateral below-knee
amputees. Orthotics and Prosthetics 1976;
30:3:17-24.
- Shurr DG, Nielsen DH, Golden IC, Meier
K. Ambulation with the flex-foot in bk amputees.
Academy of Orthotics and Prosthetics Annual
Meeting, Orlando, Fla., February 1989.
- Day HJB. The assessment and description of
amputee activity. Prosthetics and Orthotics International 1981; 5:23-8.
|
|
