The influence of lower-extremity muscle force on gait characteristics in individuals with below-knee amputations secondary to vascular disease.

[Powers CM, Boyd LA, Fontaine CA, Perry J. The influence of lower-extremity muscle force on gait characteristics in individuals with below-knee amputations secondary to vascular disease. Phys Ther. 1996;76:369-377.]

Seventy percent of lower-extremity amputations result from complications associated with diabetes mellitus and peripheral vascular occlusive disease.[1-3] Peripheral vascular compromise, resulting from diabetes mellitus, leads to multiple health problems, including poor ability to heal wounds, infections, ischemia, and neuropathy.[4] Due to these factors, people who have diabetes are 15 times more likely to have an amputation.[5] Fifty-five percent of lower-limb amputations occur at the transtibial level.[3]

After amputation, gait speed usually declines and the energy cost of walking increases.[6-8] Although regaining strength on the amputated side and augmenting sound-limb musculature have been emphasized for increased function,[1,3,9-12] the direct relationship between specific muscle groups' strength and gait ability has not been clearly established. Renstrom et al[13] found a significant correlation (P<.01) between knee extensor and flexor strength and step length, as well as a relationship between strength and maximal walking speed, in a group of individuals with below-knee amputations. Only the knee flexors and extensors, however, were examined. Correlations between the strength of various muscles and gait variables have been found in patients with post-polio syndrome[14] and stroke,[15,16] indicating that strength plays a key role in ambulatory function.

Although many factors influence the gait of individuals who have had a dysvascular amputation, including decreased sensation, poor balance, and impaired vision, traditional rehabilitation programs have emphasized increasing muscle strength and coordination to facilitate successful prosthetic gait.[3,12,17] Most of these reports have not quantified the relative contributions of specific muscles to gait ability; however, many of the associations between strength and gait function have been derived from clinical observations. For example, a relationship between knee extensor strength and stride length in persons with below-knee amputations has been noted in the clinical setting.[18] The particular limitations that weakness in other major muscle groups impose on gait, however, remain to be discerned.

The purpose of our study was to establish the association between muscular torque capability and stride characteristics for the individual who has had a below-knee amputation secondary to vascular disease. In addition, comparisons of strength between the sound and residual limbs were made. Once the importance of torque in both residual and sound limbs is determined, rehabilitation teams will be able to focus their intervention to address gait dysfunction more efficiently and effectively.

Materials and Methods

Twenty-two individuals (15 men, 7 women) participated (Tab. 1). All of the subjects had a well-healed, unilateral below-knee amputation. Subjects were recruited from the Long Beach Veterans Administration Medical Center (Long Beach, Calif) and the Rancho Los Amigos Medical Center (Downey, Calif) prosthetic services. A chart review ensured that each subject had been diagnosed with vascular disease resulting from diabetes mellitus by a physician prior to inclusion in this study. Each subject was capable of independent ambulation, and customary assistive devices were used by 11 of the subjects. An informed consent statement was signed by each participant prior to the study.

Table 1.
Subject Characteristics
                       Males (n=15)   Female (n=7)
Age (y)
  X                     59.6          60.5
  SD                    11.1           5.1
  Range                 35-72         54-68

Height (cm)
  X                     178.9         157.1
  SD                      8.4           7.4
  Range                 165.1-190.5   144.8-167.6

Weight (kg)
  X                     86.5          68.9
  SD                     8.4           7.4
  Range                 72.117        49.5-91.8

Time since prosthetic
  fitting (mo)
  X                      24.6          13.0
  SD                     40.0           9.6
  Range                   1-156         1-24

Testing procedures included quantifying the force of the muscles that are primarily active during normal gait[19] and gait analysis to define the subjects' stride characteristics. Maximal isometric torque was measured for the hip extensor and abductor and knee extensor muscle groups bilaterally, and for the ankle plantar flexors on the sound side. The forces produced by the hip extensors and abductors were recorded with a cable tensiometer and then recalculated as torque by multiplication with the test lever-arm lengths. A Lido isokinetic dynamometer(*) was used to measure the torque values of the knee extensor group bilaterally and of the ankle plantar flexors on the sound side. The prosthesis was removed for all measurements of residual-limb strength. All equipment was calibrated prior to each test.

Hip extensor force was measured with each subject positioned supine and with the hip flexed to 20 degrees. The subject's leg was suspended by a chain from the ceiling with a tensiometer cuff attached just above the knee joint. Hip extensor data were corrected to account for the weight of the limb by recording a baseline measurement once the leg had been suspended. Subject positioning was designed to ensure that the cuff and the cable tensiometer were aligned perpendicular to the long axis of the femur. A knee cage was utilized on the sound limb to prevent knee flexion, and the pelvis was strapped down. The test lever arm was measured as the distance along the lateral aspect of the thigh between the perpendicular extensions of the center of the knee cuff and the superior margin of the greater trochanter.

Hip abductor force was measured with the subject positioned supine, with both knees extended and hips in the neutral position. The tensiometer was fixed at one end to a pole that was secured lateral to the opposite limb, aligned perpendicular to the long axis of the femur on the side being tested. A belt extending from the free end of the tensiometer was passed around both legs at the level of the knee. The contralateral pelvis was blocked by a padded wedge. Thus, the abducting limb pulled against the strap at the knee. The test lever arm was measured from the same landmarks used in the hip extension test.

Knee extensor force was determined with the subject seated with the knee flexed to 60 degrees and the hip flexed to 90 degrees. Positioning was reassessed prior to each trial. The dynamometer head was positioned so that its rotational axis was in line with that of the knee joint. The resistance pad was placed just below the tibial condyles to accommodate the shortened residual limb. A specially designed resistance pad allowed for adequate contact between the tibia and the lever arm once the prosthesis was removed. The position and placement of the resistance pad were the same as for the sound limb.

Ankle plantar-flexion force on the sound side was assessed with the subject in a sitting position, with the knee extended and the ankle in neutral (0 [degree] of dorsiflexion or plantar flexion). The dynamometer axis was positioned to approximate the axis of rotation of the ankle.

Each subject performed two maximal isometric efforts lasting 5 seconds for each torque test described. A 3-minute rest interval was given between efforts. Peak torque was averaged over the highest 1-second interval of each trial. All torque values were expressed in kilogrammeters and then as a percentage of normal values based on gender.

Unpublished normative data from the Pathokinesiology Laboratory, Rancho Los Amigos Medical Center, were used to calculate isometric strength as a percentage of normal (Tab. 2). These isometric torque data were collected previously from 35 men and 37 women without pathology. Calculation of isometric torque as a percentage of our data set permitted comparison between male and female subjects.

[TABULAR DATA 2 OMITTED]

The reliability of isometric torque measurements using dynamometers and cable tensiometers has been reported previously. Clarke[20] found that the reliability of measurements of maximal isometric hip extension and hip abduction torque obtained using a cable tensiometer was good (r=.94 and .82, respectively). Similar results were reported by Sleivert and Wenger[21] for maximal isometric knee extension torque (with a distal lever arm) (r=.94) and ankle plantar-flexion torque (r=.72) measured with an isokinetic dynamometer. Due to the need for placement of the lever arm proximal during knee extension torque testing, a reliability study was conducted to determine whether positioning would affect repeatability. Five adult subjects were tested using the proximal lever-arm apparatus on two consecutive days. These data demonstrated good reliability for this procedure (r=.77).

Stride characteristics were recorded with the Stride Analyzer System.([dagger]) This system utilizes footswitch insoles taped to the bottoms of the subjects' shoes. These insoles contain compression-closing switches located beneath the heel, the first and fifth metatarsal heads, and the great toe. The signals from the individual sensors were transmitted by FM-FM telemetry to the footswitch stride analyzer, which calculated the subjects' walking speed, stride length, and cadence.

Level walking was conducted on a 10-m walkway. The middle 6 m, designated by photoelectric cells, was used as the data-collection area. Measurements were collected from two trials of a self-selected (ie, free) walking speed. For those subjects who were able to walk at a faster speed, measurements were collected from two fast walking trials. These individuals were instructed to walk as fast as safely possible. Values for both free-speed and fast walking trials were then calculated and expressed as a percentage of the normal free speed during ambulation, by gender, for each subject (Tab. 3).

Table 3.
Stride Characteristics for Free-Speed (n=22) and Fast Walking
(n=14)
                    Normative        Subjects with
                    Values(a)        Amputations
                    Males  Female    Male  Female
Free speed (m/min)
  X                 77.0   77.0      60.6   32.2
  SD                                 15.4    2.5.

Fast speed (m/min)
  X                                  84.9   51.5
  SD                                 15.5    6.1

Fast-speed cadence
  (steps/min)
  X                 107.5  114.5     99.4   82.1
  SD                                 10.2   19.5

Free-speed stride
  length (m)
  X                  1.54    1.33     1.2    0.7
  SD                                  0.2    0.2

Fast-speed stride
  length (m)
  X                                   1.4     0.9
  SD                                  0.2     0.0
"Normative values based on unpublished data (male subjects' mean age=45.9
y, SD=17.7, range=22-65; female subjects' mean age=50.6 y, SD=19.4,
range=20-84).

Data of both trials of all torque tests and stride characteristics from the walking runs were averaged for analysis. Stepwise regression analysis using forward stepping was performed to determine whether any of the independent variables (muscle group torques expressed as a percentage of normal) were predictive of any of the stride characteristics during free-speed walking. Only those variables for which F-to-enter values were statistically significant at the P<.05 level were entered into the regression equation. Separate analyses were performed for each of the dependent variables (speed, stride length, and cadence). These analyses were repeated for fast walking. Isometric torque values were compared between the sound and residual limbs using paired t tests. The alpha level was adjusted to compensate for the inflated alpha of multiple correlated t tests. Statistical analysis was performed using BMDP statistical software.([double dagger])

Results

Isometric Torque

Isometric torque values showed a wide variability for both the residual and sound limbs (Tab. 2). The hip abductor and hip extensor torques were not different between the amputated and sound limbs. The knee extensors of the sound limbs, however, were stronger than the knee extensors of the amputated limbs (54% versus 46% of normal; P=.01).

In general, torques of the various muscle groups were found to be only moderately correlated with each other (Tab. 4). The strongest association was observed between the hip extensor torques of the sound limbs and the hip extensor torques of the residual limbs (r=.86), whereas the weakest correlation was evident between the torques of the hip abductors and the knee extensors of the residual limbs (r=.42).

Table 4.
Correlation Coefficient Matrix for Isometric Muscle Torques(a)
          HABD-S  HEX-S  KEX-S  APF-S  HABD-R  HEX-R  KEX-R
HABD-S    1.00
HEX-S      .63    1.00
KEX-S      .46     .57   1.00
APF-S      .56     .61    .55   1.00
HABD-R     .77     .64    .46    .62    1.00
HEX-R      .79     .86    .66    .71     .72   1.00
KEX-R      .53     .58    .79    .53     .42    .66    1.00

Walking Speed

The mean free walking speed was limited to 59% of normal. Of the 22 subjects tested, only 14 subjects were capable of walking at a fast speed that averaged 91% of normal self-selected speeds (Tab. 3). Hip extensor torque of the residual limb was the only predictor for both free-speed walking (r=.50; F= 6.60; df= 1,20; P<.01; Fig. 1) and fast walking (r=.72; F=13.25; df=1,12; P<.01; Fig. 2). This finding accounted for 25% of the variance for free-speed walking and for 52% of the variance for fast walking ([R.sup.2]) . Once hip extensor torque entered the regression equation, no other torque variable demonstrated a partial correlation that improved predictability (Tab. 5).

Table 5.
Simple and Partial Correlation Coefficients for Muscle Torque and
Speed(a)
          Free Speed            Fast Speed
          r        r (Partial)  r      r (Partial)
HABD-S    .42     .04           .67    .29
HEX-S     .48     .13           .64    .08
KEX-S     .43     .16           .47    .22
APF-S     .46     .16           .58    .05
HABD-R    .46     .16           .56    .07
HEX-R     .50     ...           .72    ...
KEX-R     .42     .16           .56    .22
(a) r=simple correlation; r (partial) = partial correlation; HABD-S = hip
abductors, sound limb; HEX-S= hip extensors, sound limb; KEX-S=knee
extensors, sound limb; APF-S=ankle plantar flexors, sound limb; HABD-R=hip
abductors, residual limb; HEX-R=hip extensors, residual limb; KEX-R=knee
extensors, residual limb.

Cadence

The average free-speed walking cadence for all subjects was 83% of normal, whereas the cadence for the subjects who were able to walk fast was 105% of normal freespeed gait (Tab. 3). Hip abductor torque of the sound limb was the only predictor of cadence for free-speed walking (r=.57; F=9.8; df= 1,20; P<.005; Fig. 3), as well as for fast walking (r=.57; F=5.98; df= 1,12; P<.025; Fig. 4). This finding accounted for 33% of the variability in cadence for both walking speeds ([R.sup.2]). As with walking speed, once hip abductor torque entered the regression equation, no other torque variable demonstrated a partial correlation that improved predictability (Tab. 6).

Table 6
Simple and Partial Correlation Coefficients for Muscle Torque and
Cadence(a)
          Free-Speed Cadence    Fast-Speed Cadence
          r        r (Partial)  r      r (Partial)
HABD-S    .57     ...           .58    ...
HEX-S     .44     .23           .27    .12
KEX-S     .31     .07           .03    .03
APF-S     .44     .18           .36    .36
HABD-R    .51     .13           .44    .44
HEX-R     .48     .06           .38    .38
KEX-R     .37     .09           .09    .09
(a) r=simple correlation; r (partial) = partial correlation; HABD-S = hip
abductors, sound limb; HEX-S= hip extensors, sound limb; KEX-S=knee
extensors, sound limb; APF-S=ankle plantar flexors, sound limb; HABD-R=hip
abductors, residual limb; HEX-R=hip extensors, residual limb; KEX-R=knee
extensors, residual limb.

Stride Length

Stride length was limited to 69% of normal for free-speed walking and to 86% of normal for fast walking (Tab. 3). Knee extensor torque of the sound limb was the only predictor of free-speed stride length (r=.44; F=4.89; df=1,20; P<.025), accounting for 20% ([R.sup.2]) of the variability (Fig. 5). Knee extensor torque of the residual limb was the only predictor of fast-walking stride length (r=.67; F=9.81; df=1,12; P<.005), accounting for 45% ([R.sup.2]) of the variability (Fig. 6). No other variable demonstrated a partial correlation coefficient that improved predictability once knee extensor torque was allowed into the regression equation (Tab. 7).

Table 7.
Simple and Partial Correlation Coefficients for Muscle Torque and
Stride Length
          Free-Speed            Fast-Speed
          Stride-Length         Stride-Length
          r        r (Partial)  r      r (Partial)
HABD-S    .27     .08           .52    .46
HEX-S     .37     .16           .61    .35
KEX-S     .44     ...           .66    .34
APF-S     .36     .16           .51    .28
HABD-R    .31     .15           .38    .42
HEX-R     .39     .16           .66    .44
KEX-R     .39     .07           .67    ...
(a) r=simple correlation; r (partial) = partial correlation; HABD-S = hip
abductors, sound limb; HEX-S= hip extensors, sound limb; KEX-S=knee
extensors, sound limb; APF-S=ankle plantar flexors, sound limb; HABD-R=hip
abductors, residual limb; HEX-R=hip extensors, residual limb; KEX-R=knee
extensors, residual limb.

Discussion and Conclusions

Lost mobility at the ankle and foot, lack of distal muscular control, and absent proprioception all contribute to altered gait characteristics in persons with a below-knee amputation.[19] Because of their retention of normal hip and knee control, these individuals have the potential for very high function; however, strength deficits in the remaining musculature often limit ambulation ability.[13,19,22] This is especially true for individuals who have had an amputation as a result of vascular disease and who potentially are at risk for generalized deconditioning, decreased vision, and impaired sensation, which may further compromise their functional abilities.

Torques of the various muscle groups in both lower extremities of these subjects ranged from 45.9% to 54.8% of normal (Tab. 2), This weakness contributed to a substantial reduction in free walking speed (59% of normal), which is consistent with other reports of slowed speed in this patient population.[7,8] Decreased speed was the result of both diminished stride length and cadence, which averaged 70% and 83% of normal, respectively.

Only 14 of the 22 subjects were able to demonstrate an increase in gait speed. Increased cadence to 105% of normal (from 83%) was the primary cause of this faster speed, as stride length improved only moderately (from 69% to 86% of normal). This finding is suggestive of a torque limitation, as faster speed was achieved by increasing the number of steps per minute rather than by lengthening strides, which imposes a greater muscular demand.[19]

Both free and fast gait speeds were found to be related to hip extensor torque of the residual limb. This finding appears to contradict the findings of Renstrom et al,[13] who found a correlation between residual-limb knee extensor strength and maximum walking speed. In our study, residual-limb knee extensor torque had a low partial correlation of .22 and was not statistically significant once residual-limb hip extensor torque had entered into the equation. Renstrom and colleagues did not assess other muscle groups, making comparison with our results difficult.

In individuals with intact limbs, the hip extensors are active from terminal swing through loading response and control the external hip flexion moment during initial double-limb support. This activity usually occurs until approximately 10% of the gait cycle has been completed and has been reported to require an electromyographic intensity of about 20% of maximum.[19] Prolonged hip extensor activity in people with below-knee amputations has been reported by Torburn et al.[23] Their subjects demonstrated more than twice the duration and intensity of normal, indicating that a greater demand is placed on this muscle group in individuals with below-knee amputations.

Normal ankle mobility provides shock absorption during the loading response and progression during stance, which are essential components of gait.[19] This normal ankle mobility, however, is compromised in most prosthetic foot designs. Kinematic analyses of various prosthetic foot components have documented that these motion inadequacies result in diminished tibial progression. Anzel et al[22] found decreased tibial speed in a group of subjects with below-knee amputations, which they attributed to the lack of normal ankle mechanics because the cushioned heel keeps body weight on the heel. On the average, the forward tibial velocity during early stance was almost half of that of individuals without amputations (100 [degrees]/s versus 189 [degrees]/s).

To compensate for decreased tibial mobility, progression is augmented through a forward trunk lean. The consequence of facilitating forward progression is a greater demand on the hip extensors. The greater correlation between hip extensor torque and fast walking compared with free-speed walking (r=. 72 versus .50) underscores the importance of this muscle group in meeting the increased demands of controlling sagittal-plane motion at a faster walking speed.

Sound-limb hip abductor torque was related to cadence at both walking speeds. Acting in its normal role, this muscle group provides pelvic stability in response to the external hip adduction moment imposed by the medial position of body weight with respect to the hip joint.[24] The relatively high loading force transferred onto the sound limb as the result of prosthetic foot limitation, however, increases the demand on the hip abductors of the sound limb.[19]

Higher cadence requires more periods of pelvic stability during single-limb support as the number of steps per minute rises. This higher cadence further increases the muscular demand on the hip abductors as they work to meet the challenge imposed by longer periods of single-limb support time. Increased number of steps per minute was associated with greater torque-producing capability in the hip abductors. Unlike walking speed, this relationship was unchanged during fast walking (r=.57 for both conditions), indicating that other factors such as the momentum of the contralateral limb may also contribute to increasing cadence during accelerated gait.

The relationship between stride length and muscle torque differed in response to changes in walking speed. During free-speed ambulation, the torque of the sound-limb knee extensors was related to stride length. For fast walking, however, the torque of the residual-limb knee extensors became most predictive of stride length. In normal gait, the knee extensors function eccentrically during the loading response to control knee flexion resulting from the heel-rocker action at the ankle.19 Functionally, the knee extensors provide dynamic shock absorption and maintain stance stability. As stride length increases, heel-only initial contact is ensured, potentially increasing the heel-rocker effect and the subsequent loading-response knee flexion. Winter[25] and Arendt-Nielsen et al[26] have documented an increase in knee flexion during initial stance at progressively faster gait speeds. A longer stride length also reduces the opportunity to compensate for weak quadriceps femoris muscles, as a more pronounced forward trunk lean would be required to move the body-weight vector anterior to the knee joint. Positioning the body weight more posterior with respect to the knee would increase the external knee flexion moment, thus placing a higher demand on the quadriceps femoris musculature.

The demands of flexed knee stance were demonstrated by Hsu et al,[27] who found progressively increasing force demands and quadriceps femoris muscle electromyographic activity as the degree of knee flexion increased. As control of the knee flexion moment during loading response is dependent on muscular torque-producing capability, inability to meet this demand results in actions that avoid knee flexion. A shortened stride length is a characteristic compromise. This compensation is consistent with the decreased knee extensor torque evident in our study and with the avoidance of knee flexion during the loading-response phase that has been reported for subjects with below-knee amputations.[23]

Force-plate analysis has revealed that individuals with amputations have higher vertical ground reaction forces on the sound side, indicating that a greater load is being placed on this limb with associated potential for the development of an increased moment and muscular demand at knee.[28] For free walking speed in our study, the ability to meet this demand on the sound limb corresponded with a greater stride length. Increasing stride length during fast walking, however, was more dependent on torque of the amputated-limb knee extensors, as evidenced by a stronger correlation (r =.67) when compared with the sound limb during free-speed walking (r =.44). To increase stride length, the foot is placed further anterior to the knee. This posture exaggerates the heel-rocker effect of the residual limb and requires increased muscular effort to provide knee stability. Although the higher demand of fast gait would be imposed on both limbs, the weaker quadriceps femoris muscles of the residual limb proved to be more sensitive according to the relative correlation values. Therefore, torque of the extensor mechanism on the amputated limb is most likely the weak link in providing an increased stride length. Interestingly, the knee extensors were the only muscle group to vary between the sound and residual limbs in our subjects.

The results of the stepwise regression procedures utilized in this study identified the best muscle torque predictors for the various stride characteristics. Although the importance of each of these muscle groups in contributing to gait function for individuals with amputations secondary to vascular disease has been discussed, other muscle groups should not be ignored. Our conclusion is supported by the relatively high correlations between some of the various muscle groups. For example, hip extensor torques of the sound and residual limbs were highly correlated (r =.86). Although hip extensor torque of the amputated limb was the only value entered into the regression equation for speed, the close relationship between the sound-limb and residual-limb hip extensors suggests that sound-limb hip extensor torque also would be an important contributor to this variable.

The relatively high collinearity between the torque variables resulted in low partial correlations. These coefficients are indicative of the relationship between the dependent variable and the remaining independent variables, given that the best predictor (independent variable) had entered the equation. These low partial correlations suggest that the overall predictability would not have improved if multiple independent variables were combined in a regression analysis. If predictability could have been improved by the addition of another independent variable, then this predictor would have entered the equation as part of the stepwise procedure.

An understanding of the relationships between functional gait speeds and the torque-producing capabilities of both lower extremities provides the clinician with valuable information to incorporate into a rehabilitation program. Although strength often is emphasized in discussions concerning the rehabilitation of persons who have had an amputation, the sound limb often is considered to be intact and is commonly overlooked. From the results of this study, however, it is apparent that adequate torque-producing capability is necessary in both the residual and sound limbs to improve gait ability in this group of individuals.

Increased emphasis during rehabilitation on the proximal muscle groups, the hip extensors and abductors, as well as the knee extensors may result in greater ambulation ability. As the length of rehabilitation programs shortens, functional outcomes may be maximized by concentrating effort on the critical muscle groups, thus optimizing the gait of the person with a dysvascular below-knee amputation.

Although poor torque-producing capability was a major limiting factor, in the gait ability of the persons with dysvascular amputations in this study, care must be taken before extrapolating these results to all individuals with below-knee amputations due to the relatively small sample size. Factors such as decreased sensation also may affect gait characteristics. Future research examining the relative impact on gait of the other complications associated with diabetes and peripheral neuropathies is warranted.

[Figures 1 to 7 ILLUSTRATION OMITTED]

References

[1] Esquenazi A. Geriatric amputee rehabilitation. Clin Geriatr Med. 1993;9:731-743. [2] Torres MM. incidence and causes of limb amputations. In: Esquenazi A, ed. Physical Medicine and Rehabilitation. Philadelphia, Pa: Hanley & Belfus Inc; 1994:1-8. [3] Sanders GT. Lower-Limb Ampulations: A Guide to Rehabilitation. Philadelphia, Pa: FA Davis Co; 1986:13-33. [4] Pecoraro RE, Reiber GE, Burgess EM. Pathways to diabetic limb amputation. Diabetes Care. 1990;13:513-521. [5] Most RS, Sinnock P. The epidemiology of lower extremity amputations in diabetic individuals. Diabetes Care. 1983;6:87-91. [6] Pinzur MS, Gold J, Schwartz D, Gross N. Energy demands for walking in dysvascular amputees as related to the level of amputation. Orthopedics. 1992;15:1033-1037. [7] Barth DG, Schumacher L, Sienko Thomas S. Gait analysis and energy cost of below-knee amputees wearing six different prosthetic feet. Journal of Prosthetics and Orthotics. 1992;4(2):63-75. [8] Waters RL, Perry J, Antonelli D, Hislop H. Energy cost of walking of amputees: the influence of level of amputation. J Bone Joint Surg [Am]. 1976;58:42-46. [9] Eisert O, Tester OW. Dynamic exercises for lower extremity amputees. Arch Phys Wed, Rehabil. 1954;35:695-704. [10] Cameron HC, Lennard-Jones JE. Amputations in the diabetic outcome and survival. Lancet. 1964;2:605-607. [11] Burgess EM, Alexander AG. The expanding role of the physical therapist on the amputee rehabilitation team. Phys Ther. 1973;53:141-143. [12] Humm W. Postoperative physical therapy. In: Anonymous. Rehabilitation of the Lower Limb. 2nd ed. London, England: Baillibre, Tindall and Cassell; 1969:6-12. [13] Renstrom P, Grimby G, Larsson E. Thigh muscle strength in below-knee amputees. Scand J Rehabil Med Suppl. 1983;9:163-173. [14] Perry J, Mulroy SJ, Renwick SE. The relationship of lower extremity strength and gait parameters in patients with post-polio syndrome. Arch Phys Med Rehabil. 1993;4:165-169. [15] Bohannon RW. Strength of lower limb related to gait velocity and cadence in stroke patients. Physiotherapy Canada. 1986;38:204-206. [16] Nakamura R, Watanabe S, Handa T, Morohashi I. The relationship between walking speed and muscle strength for knee extension in hemiparetic stroke patients: a follow-up study. Tohoku J Exp Med. 1988;154:111-113. [17] Burgess EM. Amputation surgery and postoperative care. In: Banerjee SN, ed. Rehabilitation Management of Amputees. Baltimore, Md: Williams & Wilkins; 1982:17-41. [18] Mensch G, Ellis P. Physical therapeutic management for lower-extremity amputees. In: Banerjee SN, ed. Rehabilitation Management of Amputees. Baltimore, Md: Williams & Wilkins; 1982:165-236. [19] Perry J. Gait Analysis: Normal and Pathological Function. Thorofare, NJ: Slack Inc; 1992. [20] Clarke HH. Cable Tension Strength Tests: A Manual. Springfield, Mass: Stuart E Murphy; 1953. [21] Sleivert GG, Wenger HA, Reliability of measuring isometric and isokinetic peak torque, rate of torque development, integrated electromyography, and tibial nerve conduction velocity. Arch Phys Med Rehabil. 1994;75:1315-1321. [22] Anzel SH, Perry J, Ayyappa E, et al. Prosthetic design for dysvascular below-knee amputees. Rehabilitation R&D Progress Repans. 1993;30-31: 29-30. [23] Torburn L, Perry J, Ayyappa E, Shanfield SL. Below-knee amputee gait with dynamic elastic-response prosthetic feet: a pilot study. J Rehabil Res Dev. 1990;27:369-384. [24] Soderberg GL. Kinesiology: Application to Pathological Motion. Baltimore, Md: Williams & Wilkins; 1986. [25] Winter DA. Biomechanical motor patterns in normal walking. Journal of Motor Behavior. 1983; 15:302-330. [26] Arendt-Nielsen L, Sinkjaer T, Nielsen J, Kallesoe K. Electromyographic patterns and knee joint kinematics during walking at various speeds. Journal of Electromyography and Jinesiology. 1991;1:89-95. [27] Hsu AT, Perry J, Gronley JK, Hislop HJ. Quadriceps force and myoelectric activity during flexed knee stance. Clin Orthop. 1993;288: 254-262. [28] Powers CM, Torburn L, Perry J, Ayyappa E. Influence of prosthetic foot design on sound-limb loading in adults with unilateral below-knee amputations. Arch Phys Med Rehabil. 1994;75:825-829.

(*) Loredan Biomedical Inc, 1632 Da Vinci Ct, PO Box 1154, Davis, CA 95617. ([double dagger]) BMDP Statistical Software Inc, 1440 Sepulveda Blvd, Suite 316, Los Angeles, CA 90025.

CM Powers, PT, is Doctoral Candidate, Department of Biokinesiology and Physical Therapy, University of Southern California, Los Angeles, CA 90007, and Research Physical Therapist, Pathokinesiology Laboratory, Rancho Los Amigos Medical Center, Bldg 304, 7601 E Imperial Hwy, Downey, CA 90242.

LA Boyd, PT, is Research Physical Therapist, Pathokinesiology Laboratory, Rancho Los Amigos Medical Center.

CA Fontaine, PT, was Research Physical Therapist, Pathokinesiology Laboratory, Rancho Los Amigos Medical Center, when this work was conducted.

J Perry, MD, is Chief, Pathokinesiology Laboratory, Rancho Los Amigos Medical Center, and Professor of Orthopaedics, University of Southern California. Address all correspondence to Dr Perry at Pathokinesiology Laboratory, Rancho Los Amigos Medical Center, Bldg 304, 7601 E Imperial Hwy, Downey, CA 90242 (USA).

This study was approved by the institutional review boards of Rancho Los Amigos Medical Center and Long Beach Veterans Administration Medical Center, Long Beach, CA.

Funding was provided by Veterans Administration contract #A735-RA

This article was submitted May 15, 1995, and was accepted December 13, 1995.