Journal Logo

APPLIED SCIENCES

A Limb-specific Strategy across a Range of Running Speeds in Transfemoral Amputees

SAKATA, HIROYUKI1,2; HASHIZUME, SATORU1; TAKEMURA, HIROSHI1,2; HOBARA, HIROAKI1

Author Information
Medicine & Science in Sports & Exercise: April 2020 - Volume 52 - Issue 4 - p 892-899
doi: 10.1249/MSS.0000000000002203

Abstract

The development of carbon-fiber running-specific prostheses (RSPs) has greatly improved the sprinting performance of individuals with lower extremity amputation (1). RSPs are generally characterized by C- or J-shaped design and passive materials that are lighter than the biological leg (1,2). Consequently, unlike the biological leg, RSPs cannot generate additional mechanical power during the stance phase (1,3). Because of the differences in the mechanical properties between RSPs and biological leg, runners with unilateral lower extremity amputation have larger interlimb asymmetry than do nonamputees (4). Therefore, understanding limb-specific biomechanical strategies across running speeds when using RSPs will provide insightful information to improve sprint performance and prosthesis design among individuals with unilateral lower extremity amputation.

Constant running speed is determined by the product of the stance average vertical ground reaction force (vGRF) relative to body weight (BW; Favg), step frequency (Freqstep), and contact length (Lc; the length traveled by a runner’s body during the stance phase), with the assumption that center-of-mass height at touchdown and running speed are constant (5–8). Grabowski et al. (9) found that the Favg was 9% less for the affected limb than for the unaffected limb across a range of speeds, including top speed (8.8 ± 1.0 m·s−1) in sprinters with unilateral transtibial amputation, suggesting that RSPs impair force generation and thus probably limit top speed. Furthermore, a previous study demonstrated that to attain the same top sprinting speeds (10.8 m·s−1) as able-bodied sprinters, sprinters with bilateral transtibial amputations using RSPs had 34% shorter aerial time and 16% higher Freqstep which compensated for the 22% smaller Favg of their affected limbs, as compared with the able-bodied sprinters (10). Although several studies have investigated vGRF variables and spatiotemporal parameters, including Favg, Freqstep, and Lc among runners with unilateral and bilateral transtibial amputation (5,9–11), these variables across running speeds have not been investigated among sprinters with unilateral transfemoral amputation.

According to a previous study (12), a sprinter with a unilateral transfemoral amputation had more asymmetrical joint kinematics between the unaffected and affected limbs during maximal sprinting than did able-bodied sprinters and sprinters with unilateral transtibial amputation. Interlimb asymmetry on vGRF variables and temporal parameters were larger during maximal sprinting than during walking (13). Moreover, recent studies (14,15) found that the affected limb exerted a 26% smaller peak vGRF than did the unaffected limb at the maximal sprinting speed (5.7 ± 0.7 m·s−1). However, it remains unclear how sprinters with unilateral transfemoral amputation compensate for a limitation on the magnitude of force in the affected limb in order to achieve faster running speeds.

The purpose of this study was to investigate vGRF variables and spatiotemporal parameters across a range of speeds in sprinters with unilateral transfemoral amputation using RSPs. Previous studies have suggested that RSPs impair the mass-specific vGRF during sprinting at maximal speed in sprinters with unilateral transfemoral amputation (14,15) and during running across a range of speeds in sprinters with unilateral and bilateral transtibial amputation (4,9–11). Furthermore, earlier studies demonstrated that the increase in Favg with running speed was more prominent in the unaffected limb than in the affected limb (9,11). Therefore, we firstly hypothesized that the Favg of the affected limb would be smaller than that of the unaffected limb, and the between-limb difference in Favg would increase with speed. Next, a recent finding demonstrated that contact time of the affected limb was significantly longer than that of the unaffected limb during overground maximal sprinting in unilateral transfemoral amputees (14). However, the authors also showed that vGRF impulses were similar between the affected and unaffected limbs, indicating that aerial time would be similar between the limbs. Because the Freqstep is determined from the inverse of total step time (the sum of contact time and subsequent aerial time), we secondly hypothesized that the affected limb would have a lower Freqstep than the unaffected limb across a range of running speeds. Finally, because the Lc is the product of contact time and speed (7), a longer contact time would mathematically induce a longer Lc at a given speed. Therefore, our third hypothesis was that the affected limb would have a longer Lc than the unaffected limb across a range of running speeds.

METHODS

Participants

Ten sprinters with unilateral transfemoral amputation participated in this study (Table 1). All participants had competitive athletic experience of the 100-m race and regularly trained 1–6 d·wk−1 at the time of the study. Each participant used their his/her prescribed RSPs and prosthetic knee joints (Table 1). The stiffness of the RSPs was based on the participant’s body mass and activity level (slow or fast running) (3,16). Similarly, as per the manufacturer’s guidelines, each participant used the prosthetic knee joints based on his/her activity level (walking or running) (16). The protocol was approved by the local ethics committee and was in accordance with the guidelines set out in the Declaration of Helsinki (1983). All subjects gave informed written consent before participating.

T1
TABLE 1:
Subject characteristics.

Experimental protocol

Before data collection, we instructed each participant to walk and run on the treadmill for at least 5 min (17), as a minimal familiarization period for running on an instrumented split-belt treadmill (FTMH-1244WA; Tec Gihan, Kyoto, Japan; Fig. 1). Nevertheless, all participants performed walking and running for more than 10 min. Thereafter, each participant performed running trials on the instrumented treadmill at increments of his/her 100% speed. In our study, the 100% speed for each participant was defined as the average speed of his/her 100-m personal best (PB) as recorded in the official competitions (Table 1). The running speed was increased from 30% to 70% in increments of 10%. For all trials, the treadmill belt speed was constantly accelerated at 0.84 m·s−2 to the target speed. Between the trials, we instructed the participants to take as much rest as necessary, in order to reduce the effects of fatigue. Each rest period was 1–2 min. A safety harness was used to support the participants in the event of falling and had sufficient slack so as not to impede the participants’ natural running mechanics (Fig. 1).

F1
FIGURE 1:
Illustration of the experimental setup. Two force plates were embedded in the split-belt treadmill. Each subject (N = 10) ran on the instrumented treadmill at five speeds (30%, 40%, 50%, 60%, and 70% of average speed of 100-m PB). A safety harness and handrails on the sides of the treadmill were used to support participants above the treadmill belt in the event of a fall.

Data collection

Two 6-df piezoelectric force plates (TF-40120-CL and TF-40120-CR; Tec Gihan) mounted underneath each treadmill belt were used to collect vGRF data at 1000 Hz. The GRFs measured by each force plate were converted to total GRFs using the equipped signal converter (FTMH-1244WA; Tec Gihan). The total vGRF data were filtered by a fourth-order zero-lag low-pass Butterworth filter with a cutoff frequency of 25 Hz (18,19). From the filtered vGRF data, the instants of touchdown and takeoff were detected using a threshold of 40 N (8–10,18).

Data analysis

Running speed was defined as the product of step length (Lstep) and step frequency (Freqstep), mechanically:

We calculated step frequency (Freqstep) as the inverse of the time from touchdown to contralateral touchdown (4,5,9,20), which was equal to the sum of the ground contact time (tc) and subsequent aerial time (taer; Fig. 2):

F2
FIGURE 2:
A, Illustration of a sprinter with unilateral transfemoral amputation during the contact and aerial phases for the unaffected (black) and affected (gray) limbs. B, Corresponding vGRF data for the unaffected (black) and affected (gray) limbs in a representative runner while running at 70%. The vGRF data were normalized to the BW.

Under conditions in which speed and the center-of-mass height at touchdown was constant, Lstep was equal to the product of contact length (Lc; the length the body traveled during the stance phase, determined by multiplying the tc by the treadmill belt speed) and stance average vGRF (Favg) relative to the BW, including the weight of the prostheses (7–10):

We then rearranged equation 1 by substituting equation 3 for Lstep to obtain

We calculated the variables in equations 1 through 4 and vGRF impulse (vGRI). The vGRI was calculated as time integrals of the vGRF minus BW (7,21–23), which reflected the change in the vertical velocity of the center of mass during the stance phase and determined the subsequent taer (ignoring air resistance). We analyzed 14 consecutive steps and averaged the 7 steps of each limb to determine the representative values at each speed. All vGRF variables (Favg and vGRI) were normalized to each participant’s BW, which was measured during a static standing trial on the instrumented treadmill.

Statistical analysis

Data distribution was checked using the Shapiro–Wilk test (Table, Supplemental Digital Content 1, which shows the results of the Shapiro–Wilk test, https://links.lww.com/MSS/B816). For normally distributed data (P > 0.05), two-way repeated-measure ANOVA (speed (five levels: 30%, 40%, 50%, 60%, 70%) × limb (two levels: unaffected limb, affected limb)) was performed to compare the variables between the limbs across the speeds (Table, Supplemental Digital Content 2, which shows the results of the ANOVA and nonparametric tests, https://links.lww.com/MSS/B817). Mauchly’s test of sphericity was performed to assess assumptions of variance. If the sphericity assumption was violated, Greenhouse–Geisser correction was performed to adjust the degree of freedom. Bonferroni post hoc multiple comparison was performed if significant main effects or interactions were observed (Table, Supplemental Digital Content 3, which shows the results of the post hoc tests, https://links.lww.com/MSS/B818). Effect size (ES) was estimated using partial η2. If the data were not normally distributed (P < 0.05), the nonparametric Friedman test and Wilcoxon rank sum test were used. When a significant effect of speed was observed in the Friedman test, Wilcoxon rank sum test with the Bonferroni correction was used for post hoc analyses. SPSS for Windows Version 22 (SPSS Inc., Chicago, IL) was used for all statistical analyses. Statistical significance was set to P < 0.05.

RESULTS

The average running speeds during each trial were as follows: 1.76 ± 0.22 m·s−1 for 30%, 2.35 ± 0.30 m·s−1 for 40%, 2.93 ± 0.36 m·s−1 for 50%, 3.52 ± 0.44 m·s−1 for 60%, and 4.10 ± 0.50 m·s−1 for 70% of the average speed of 100-m PB, respectively. Figures 2A and B show a typical example of a running stride and simultaneous vGRF data for the unaffected and affected limbs, recorded from a sprinter with unilateral transfemoral amputation at 70% speed (4.94 m·s−1). Figure 3 also shows typical time-course profiles of vGRF at each speed in the unaffected and affected limbs, respectively. As shown in Figures 2 and 3, the affected limb exerted a smaller peak vGRF than did the unaffected limb at relatively faster speeds.

F3
FIGURE 3:
Typical examples of vGRF during the stance phase across running speeds for the same male sprinter whose data are shown in Figure 2. Each color indicates variations in running speeds from 30% to 70%. For both the unaffected and affected limbs, peak vGRF increased and ground contact time decreased across the range of speeds. Peak vGRF of the affected limb increased markedly less than that of the unaffected limb.

Figure 4A shows a significant main effect of speed (F(1.48, 13.27) = 24.88, P < 0.01, ES = 0.73), but not limb (F(1.00, 9.00) = 2.64, P = 0.14, ES = 0.23) on Favg. However, there was a significant interaction between speed and limb (F(1.40, 12.60) = 11.56, P < 0.01, ES = 0.56). The post hoc analysis revealed that Favg increased with speed in both the unaffected and affected limbs, but Favg of the unaffected limb increased more than that of the affected limb. Although there was no significant difference between the limbs at relatively slower speed, Favg of the affected limb was significantly smaller than that of the unaffected limb at 60% (P < 0.05), but not at 70% (P = 0.05).

F4
FIGURE 4:
Average vGRF (A), contact length (B), step frequency (C), contact time (D), aerial time (E), and vGRF impulse (F) of the unaffected (black circles) and affected (white circles) limbs across five running speeds. Error bars represent 1 SD. Asterisks (*,**) indicate significant differences between limbs at each speed, at P < 0.05 and P < 0.01, respectively. Dagger (†), sharp (#), dollar ($), and section (§) indicate significant differences from 30%, 40%, 50%, and 60% at P < 0.05, respectively.

Statistical analysis also showed a significant main effect of speed (F(4.00, 36.00) = 39.22, P < 0.01, ES = 0.81) on Freqstep (Fig. 4B); however, there was no significant main effect of limb (F(1.00, 9.00) = 1.34, P = 0.28, ES = 0.13), or interaction between speed and limb (F(1.55, 13.97) = 0.25, P = 0.73, ES = 0.27). Both the unaffected and affected limbs significantly increased Freqstep with speed, but there was no significant difference between limbs at each speed.

Significant main effects of speed (F(1.58, 14.26) = 252.92, P < 0.01, ES = 0.97) and limb (F(1.00, 9.00) = 6.07, P < 0.05, ES = 0.40) were observed in Lc (Fig. 4C). Furthermore, there was a significant interaction between speed and limb (F(4.00, 36.00) = 41.30, P < 0.01, ES = 0.82). Both the unaffected and affected limbs significantly increased Lc with speed, but Lc of the affected limb increased more than that of the unaffected limb. Although the values of Lc were similar for both limbs at relatively slower speeds, the affected limb had a significantly longer Lc than did the unaffected limb at 50%, 60%, and 70% (Fig. 4C).

No significant main effect of speed (F(1.70, 15.34) = 2.39, P = 0.13, ES = 0.21) or limb (F(1.00, 9.00) = 0.00, P = 0.93, ES = 0.00) was observed in taer (Fig. 4D); however, there was a significant interaction between speed and limb (F(4.00, 36.00) = 4.28, P < 0.01, ES = 0.32). Although the unaffected limb slightly increased taer with increasing speeds, taer of the affected limb remained relatively constant across a range of speeds. Although different trends were observed for each limb, there was no significant difference between limbs at any speeds.

Significant main effects of speed (F(1.24, 12.04) = 92.42, P < 0.01, ES = 0.91) and limb (F(1.00, 9.00) = 3.42, P = 0.10, ES = 0.28) were observed in tc (Fig. 4E). Furthermore, there was a significant interaction between speed and limb (F(1.85, 16.62) = 12.12, P < 0.01, ES = 0.54). In both limbs, tc was similar at relatively slower speeds, but significantly decreased with speed. Because tc of the unaffected limb decreased more than that of the affected limb, the affected limb had the significantly longer tc than unaffected limb at relatively faster speeds (50%, 60%, and 70%).

The Friedman test showed no significant effect of speed on vGRI in both the unaffected (χ2(4) = 8.56, P = 0.07) and affected (χ2(4) = 4.00, P = 0.41) limbs; furthermore, Wilcoxon rank sum test found no significant difference in vGRI between the unaffected and affected limbs at all speeds. Although no significant difference was observed, the unaffected limb tended to increase vGRI from 30% to 60% and decrease from 60% to 70%. On the other hand, the affected limb’s vGRI remained relatively constant across a range of speeds.

DISCUSSION

The purpose of this study was to investigate vGRF variables and spatiotemporal parameters across a range of speeds in sprinters with unilateral transfemoral amputation who were using RSPs. As shown in Figure 4A, Favg increased with speed in both the unaffected and affected limbs, but the increase in Favg was more prominent in the unaffected limb. In particular, as running speed increased from 30% to 70%, Favg significantly increased by 26% (from 1.55 to 1.96 N·BW−1) and 12% (from 1.56 to 1.74 N·BW−1) in the unaffected and affected limbs, respectively. Consequently, the between-limb difference in Favg was only 1% at 30% speed, but the Favg of the affected limb was 11% smaller than that of the unaffected limb at 70% speed. These results supported our first hypothesis that Favg of the affected limb would be smaller than that of the unaffected limb, and the between-limb difference in Favg would increase with speed. These results also agreed with those of previous studies that found that peak vGRF of the affected limb was smaller than that of the unaffected limb during maximal sprinting in runners with unilateral transtibial amputation (9,11,24) or transfemoral amputation (14,15). Furthermore, previous studies suggested that each RSP has different inertial properties (25), and that the dynamic elastic response of RSPs (26,27) is different from that of the biological limbs of nonamputees. In addition, Sherk et al. (28) found that after transfemoral amputation (19.7 ± 11.1 yr, mean ± SD), muscle atrophy was prevalent in the thigh of the residual limb, where the muscle cross-sectional area was approximately 72% smaller than that in the unaffected limb and nonamputees’ thigh. Therefore, the results of the present study and those of previous studies suggest that RSPs and residual tissues of the affected limb may limit the mass-specific vGRF at faster speeds in sprinters with unilateral transfemoral amputation as well as in sprinters with unilateral transtibial amputation.

Freqstep increased with speed in both the unaffected and affected limbs without significant differences between the limbs at any speed (Fig. 4B). Based on these results, our second hypothesis, that the affected limb would have lower Freqstep than the unaffected limb across a range of running speeds, was rejected. The results of the present study were consistent with those of previous studies that found no significant differences in Freqstep between the unaffected and affected limbs during running at submaximal speed in runners with unilateral transtibial amputation (4,20,29,30). Because the Freqstep was determined from the inverse of total step time (sum of taer and tc), a higher Freqstep could be achieved by decreasing taer and/or tc. In the present study, taer of the unaffected limb increased slightly with increasing speed, but that of the affected limb remained nearly constant across a range of speeds (Fig. 4D). Consequently, a significant interaction between speed and limb was observed (P < 0.01) and the taer of the affected limb was slightly shorter than that of the unaffected limb at 50%, 60%, and 70% speed. On the other hand, tc significantly decreased with speed for both the unaffected and affected limbs, but tc of the affected limb was significantly longer than that of the unaffected limb at 50%, 60%, and 70% speed (Fig. 4E). Consequently, the symmetric Freqstep between limbs in the present study was achieved by the asymmetric combination of taer and tc in each limb across a range of speeds. The asymmetric combination of taer and tc in each limb was inconsistent with the findings of a previous study, which showed that runners with unilateral transtibial amputation showed no significant difference between the unaffected and affected limbs in both taer and tc (4). Although not describing taer, Hobara et al. (20) also reported that tc of both the unaffected and affected limbs decreased with increasing speed, without significant differences between the limbs in runners with unilateral transtibial amputation. A possible explanation for the similar Freqstep in both limbs may be minimization of the metabolic energy expenditure. Running with asymmetric Freqstep increased the rate of metabolic energy expenditure in able-bodied individuals (31). Therefore, our findings suggest that, in order to run faster, sprinters with unilateral transfemoral amputation increase Freqstep symmetrically by combining the asymmetric taer and tc of the unaffected and affected limbs, unlike sprinters with unilateral transtibial amputation.

It is noteworthy that Lc increased with speed in both the unaffected and affected limbs, but the increase was greater in the affected limb (Fig. 4C). As a result, the Lc of the affected limb was 12% longer than that of the unaffected limb at 70% speed. These results agree with our third hypotheses, but agreed with a previous study that sprinters with unilateral transtibial amputation increased Lc without significant difference between the unaffected and affected limbs across a range of speeds, including at top speed (9). Because Lc was calculated as the product of treadmill belt speed (same for both limbs during all trials) and tc, the longer Lc of the affected limb was due to the longer tc of the affected limb than that of the unaffected limb (Fig. 4E). The longer tc of the affected limb compared with the unaffected limb was also reported by previous studies that found that sprinters with unilateral transfemoral amputation had a 13% longer tc in the affected limb than in the unaffected limb at maximal sprinting (14,15). A possible explanation for the longer tc of the affected limb may be the compensatory strategy for a limitation on the magnitude of force that can be generated by the affected limb. According to a previous study (7), fast and slow runners achieved equivalent vGRI with different combinations of the vGRF and tc at the top speed for each subject. The authors found that faster runners applied greater vGRF during briefer contact periods, whereas slower runners applied lesser vGRF during longer contact periods (7). This is because vGRI must be sufficiently high in order to create enough taer to reposition the limbs in the air for the next step (7,8,24). Indeed, despite the affected limb exerting a smaller vGRF than the unaffected limb at each speed (Figs. 3, 4A), vGRI was not significantly different between the limbs at any speed (Fig. 4F). Therefore, these results suggest that, compared with the unaffected limb, the affected limb may require longer tc to exert an equivalent vGRI and may need to induce a longer Lc to compensate for a limitation on the magnitude of force in the ipsilateral limb. Considering that leg length and leg angles at touchdown and takeoff as well as running speed were associated with tc (21,22), altered touchdown kinematics due to the use of passive prosthetic knee (12,32) and/or systematic differences in the leg length (Table 1) might associated with longer Lc and tc in the affected limb.

There are several limitations in the interpretation of the present study results. First, the large variation in the participants’ training history, performance levels (100-m PB), and RSP experiences (Table 1) might have affected the results of our study. Second, our participants used their own RSPs and prosthetic knee joints for their affected limb (Table 1) and their own shoes for their unaffected limb. Previous studies have demonstrated that the RSP model and stiffness category could affect the running biomechanics of athletes with lower limb amputation (3,5,26). In particular, the use of stiffer RSPs increased Favg and Freqstep, and decreased tc and Lc in athletes with bilateral transtibial amputations (5), indicating that variation in the stiffness of the RSP may influence the current results. Furthermore, the mechanical properties of running shoes may also have an effect on running biomechanics (33–35). Although repeated-measures analyses were used to minimize these effects, these differences might have potentially led to altered running biomechanics among participants. Third, because our experiments were conducted on an instrumented treadmill, the results of this study are not completely applicable to overground running. Previous studies have identified that parameters measured with an adequate instrumented treadmill are comparable, but not directly equivalent, to those measured during overground running (36–39). Although treadmill-based analysis of running mechanics can be generalized to overground running mechanics (38), caution is needed to interpret and generalize the results of the present study.

In summary, this study revealed that all mechanical variables related to speed (Favg, Freqstep, and Lc) were similar between the unaffected and affected limbs of runners with unilateral transfemoral amputation using RSPs at relatively slower speeds, and these variables increased with increasing speed in both limbs. Although Freqstep remained similar for both limbs at relatively faster speeds, the affected limb exerted an 11% smaller Favg and showed a 12% longer Lc than the unaffected limb. Therefore, our results suggest that, in order to achieve faster running speed, runners with unilateral transfemoral amputation using RSPs likely adopt limb-specific biomechanical strategies for the unaffected and affected limbs, where the smaller Favg of the affected limb would be compensated by the longer Lc of the affected limb, not achieving the higher Freqstep.

The authors are grateful to all athletes who participated in the study. The authors also thank Mr. Kunikazu Nakamura, the technical staff at the National Institute of Advanced Industrial Science and Technology, for his great support and preparations made for the data collections. This work was supported by JSPS KAKENHI grant number 26702027. None of the authors had any conflict of interest associated with the study. The results of the study are presented clearly, honestly, and without fabrication, falsification, or inappropriate data manipulation. The results of the present study do not constitute endorsement by the American College of Sports Medicine.

REFERENCES

1. Nolan L. Carbon fibre prostheses and running in amputees: a review. Foot Ankle Surg. 2008;14(3):125–9.
2. Brüggemann GP, Arampatzis A, Emrich F, Potthast W. Biomechanics of double transtibial amputee sprinting using dedicated sprinting prostheses. Sports Tech. 2008;1(4–5):220–7.
3. Beck ON, Taboga P, Grabowski AM. Characterizing the mechanical properties of running-specific prostheses. PLoS One. 2016;11(12):e0168298.
4. Baum BS, Hobara H, Kim YH, Shim JK. Amputee locomotion: ground reaction forces during submaximal running with running-specific prostheses. J Appl Biomech. 2016;32(3):287–94.
5. Beck ON, Taboga P, Grabowski AM. How do prosthetic stiffness, height and running speed affect the biomechanics of athletes with bilateral transtibial amputations? J R Soc Interface. 2017;14(131):11–3.
6. Robilliard JJ, Wilson AM. Prediction of kinetics and kinematics of running animals using an analytical approximation to the planar spring-mass system. J Exp Biol. 2005;208(23):4377–89.
7. Weyand PG, Sternlight DB, Bellizzi MJ, Wright S. Faster top running speeds are achieved with greater ground forces not more rapid leg movements. J Appl Physiol. 2000;89(5):1991–9.
8. Weyand PG, Sandell RF, Prime DN, Bundle MW. The biological limits to running speed are imposed from the ground up. J Appl Physiol. 2010;108(4):950–61.
9. Grabowski AM, McGowan CP, McDermott WJ, Beale MT, Kram R, Herr HM. Running-specific prostheses limit ground-force during sprinting. Biol Lett. 2010;6(2):201–4.
10. Weyand PG, Bundle MW, McGowan CP, et al. The fastest runner on artificial legs: different limbs, similar function? J Appl Physiol. 2009;107(3):903–11.
11. McGowan CP, Grabowski AM, McDermott WJ, Herr HM, Kram R. Leg stiffness of sprinters using running-specific prostheses. J R Soc Interface. 2012;9(73):1975–82.
12. Buckley JG. Sprint kinematics of athletes with lower-limb amputations. Arch Phys Med Rehabil. 1999;80(5):501–8.
13. Burkett B, Smeathers J, Barker T. Walking and running inter-limb asymmetry for Paralympic trans-femoral amputees, a biomechanical analysis. Prosthet Orthot Int. 2003;27(1):36–47.
14. Makimoto A, Sano Y, Hashizume S, et al. Ground reaction forces during sprinting in unilateral transfemoral amputees. J Appl Biomech. 2017;33(6):406–9.
15. Sano Y, Makimoto A, Hashizume S, et al. Leg stiffness during sprinting in transfemoral amputees with running-specific prosthesis. Gait Posture. 2017;56:65–7.
16. Ottobock Catalog: Prosthetics Lower Limb. Duderstadt, Germany: Ottobock. pp. 69–72.
17. Zeni JA Jr, Higginson JS. Gait parameters and stride-to-stride variability during familiarization to walking on a split-belt treadmill. Clin Biomech (Bristol, Avon). 2010;25(4):383–6.
18. Clark KP, Weyand PG. Are running speeds maximized with simple-spring stance mechanics? J Appl Physiol. 2014;117(6):604–15.
19. Kram R, Griffin TM, Donelan JM, Chang YH. Force treadmill for measuring vertical and horizontal ground reaction forces. J Appl Physiol. 1998;85(2):764–9.
20. Hobara H, Baum BS, Kwon HJ, et al. Amputee locomotion: spring-like leg behavior and stiffness regulation using running-specific prostheses. J Biomech. 2013;46(14):2483–9.
21. Hunter JP, Marshall RN, McNair P. Interaction of step length and step rate during sprint running. Med Sci Sports Exerc. 2004;36(2):261–71.
22. Hunter JP, Marshall RN, McNair P. Reliability of biomechanical variables of sprint running. Med Sci Sports Exerc. 2004;36(5):850–61.
23. Hunter JP, Marshall RN, McNair PJ. Relationships between ground reaction force impulse and kinematics of sprint-running acceleration. J Appl Biomech. 2005;21(1):31–43.
24. Willwacher S, Funken J, Heinrich K, et al. Elite long jumpers with below the knee prostheses approach the board slower, but take-off more effectively than non-amputee athletes. Sci Rep. 2017;7(1):16058.
25. Baum BS, Schultz MP, Tian A, et al. Amputee locomotion: determining the inertial properties of running-specific prostheses. Arch Phys Med Rehabil. 2013;94(9):1776–83.
26. Beck ON, Taboga P, Grabowski AM. Prosthetic model, but not stiffness or height, affects the metabolic cost of running for athletes with unilateral transtibial amputations. J Appl Physiol. 2017;123(1):38–48.
27. Noroozi S, Rahman AG, Khoo SY, et al. The dynamic elastic response to impulse synchronisation of composite prosthetic energy storing and returning feet. Proc Inst Mech Eng P J Sports Eng Technol. 2014;228(1):24–32.
28. Sherk VD, Bemben MG, Bemben DA. Interlimb muscle and fat comparisons in persons with lower-limb amputation. Arch Phys Med Rehabil. 2010;91(7):1077–81.
29. Sanderson DJ, Martin PE. Joint kinetics in unilateral below-knee amputee patients during running. Arch Phys Med Rehabil. 1996;77(12):1279–85.
30. Strike SC, Arcone D, Orendurff M. Running at submaximal speeds, the role of the intact and prosthetic limbs for trans-tibial amputees. Gait Posture. 2018;62:327–32.
31. Beck ON, Azua EN, Grabowski AM. Step time asymmetry increases metabolic energy expenditure during running. Eur J Appl Physiol. 2018;118(10):2147–54.
32. Schaarschmidt M, Lipfert SW, Meier-Gratz C, Scholle HC, Seyfarth A. Functional gait asymmetry of unilateral transfemoral amputees. Hum Mov Sci. 2012;31(4):907–17.
33. Nigg BM, Anton M. Energy aspects for elastic and viscous shoe soles and playing surfaces. Med Sci Sports Exerc. 1995;27(1):92–7.
34. Nigg BM, Stergiou P, Cole G, Stefanyshyn D, Mündermann A, Humble N. Effect of shoe inserts on kinematics, center of pressure, and leg joint moments during running. Med Sci Sports Exerc. 2003;35(2):314–9.
35. Stacoff A, Reinschmidt C, Nigg BM, et al. Effects of shoe sole construction on skeletal motion during running. Med Sci Sports Exerc. 2001;33(2):311–9.
36. Kluitenberg B, Bredeweg SW, Zijlstra S, Zijlstra W, Buist I. Comparison of vertical ground reaction forces during overground and treadmill running. A validation study. BMC Musculoskelet Disord. 2012;13:235.
37. Riley PO, Paolini G, Della Croce U, Paylo KW, Kerrigan DC. A kinematic and kinetic comparison of overground and treadmill walking in healthy subjects. Gait Posture. 2007;26(1):17–24.
38. Riley PO, Dicharry J, Franz J, Della Croce U, Wilder RP, Kerrigan DC. A kinematics and kinetic comparison of overground and treadmill running. Med Sci Sports Exerc. 2008;40(6):1093–100.
39. Van Caekenberghe I, Segers V, Aerts P, Willems P, De Clercq D. Joint kinematics and kinetics of overground accelerated running versus running on an accelerated treadmill. J R Soc Interface. 2013;10(84):20130222.
Keywords:

AMPUTATION; GROUND REACTION FORCE; PROSTHETIC KNEE; RUNNING-SPECIFIC PROSTHESIS; SPATIOTEMPORAL PARAMETER

Supplemental Digital Content

Copyright © 2019 by the American College of Sports Medicine