Journal of Applied Physiology

Counterpoint: Artificial legs do not make artificially fast running speeds possible

Rodger Kram, Alena M. Grabowski, Craig P. McGowan, Mary Beth Brown, Hugh M. Herr

“Extraordinary claims require extraordinary evidence.”—Carl Sagan

There is insufficient evidence to conclude that modern running specific prostheses (RSP) provide physiological or biomechanical advantages over biological legs. A grand total of n = 7 metabolic running economy values for amputees using RSP have been published (1, 13). Even worse, ground reaction force (GRF) and leg swing time data at sprint speeds exist for only one amputee, Oscar Pistorius (2, 13). Until recently it would have been preposterous to consider prosthetic limbs to be advantageous, thus, the burden of proof is on those who claim that RSP are advantageous. Here, we conservatively presume neither advantage nor disadvantage as we weigh and discuss recently published scientific data. Furthermore, we propose a series of experiments that are needed to resolve the topic of this debate.

RSP do not provide a distinct advantage or disadvantage in terms of the rates of oxygen consumption at submaximal running speeds [running economy (RE)]. Brown et al. (1) compared the RE of six transtibial amputee runners (5 unilateral and 1 bilateral) to six age- and fitness-matched nonamputee runners. The mean RE was numerically worse for the amputees using RSP across all speeds (219.5 vs. 202.2 ml O2·kg−1·km−1), but the difference did not reach the criterion of significance (P < 0.05). The bilateral transtibial amputee from Brown et al. had a mean RE of 216.5 ml O2·kg−1·km−1. The only other reported RE value for a bilateral amputee is that for Oscar Pistorius, 174.9 ml O2·kg−1·km−1 (13). For good recreational runners (n = 16), Morgan et al. (9) reported a mean (SD) RE value of 190.5 (13.6) ml O2·kg−1·km−1. Thus the Brown et al. bilateral amputee's RE was 1.92 SD above that mean and Pistorius’ RE was 1.15 SD below that mean. Both athletes use the same type of prostheses. From this scant evidence, it would be foolhardy to conclude that RSP provide a metabolic advantage or disadvantage.

Since vertical GRF is the primary determinant of maximal running speed (11, 12), GRF data for amputee runners are critical to this debate. Although previous studies have characterized some aspects of the biomechanics of amputee running and sprinting (3, 4, 68, 15), there are no published GRF data for unilateral amputees at their top running speeds. GRF data for top speed running have been published for only one bilateral amputee, Oscar Pistorius. To claim that prosthetic legs provide a mechanical advantage over biological legs based on n = 1 is inherently unscientific and we are surprised that any scientists would make such a claim.

Both Brüggemann et al. (2) and Weyand et al. (13) found that Pistorius exerts lower vertical GRFs than performance matched nonamputees. Brüggemann et al. contorted this force deficiency into a supposed advantage, claiming that the smaller vertical forces and impulse allow Pistorius to perform less mechanical work than his peers. That reasoning fails to recognize that sprinting requires maximizing force and mechanical power output, not minimizing them. In their seminal work, Weyand et al. (12) concluded that “human runners reach faster top speeds . . . by applying greater support forces to the ground”. Thus it is enigmatic that Weyand and Bundle (14) in this debate can convolute the smaller GRF exerted by Pistorius into a purported advantage.

Two factors may be responsible for the GRF deficit that Pistorius exhibits: 1) his passive, elastic prostheses (and/or their interface with the residual limb) prevent him from generating high forces and/or 2) his legs are not able to generate high ground force due to relative weakness. Factor 1 is certainly plausible. Compliant prostheses are necessary for running because the forces on the residual limb-prosthesis socket interface would otherwise be intolerable. Despite the compliance of RSP, amputees uniformly report significant pain at the interface during running. Factor 2 is also possible, although Pistorius has been active and engaged in various sports for 20+ years (10). He may have learned to compensate for his force impairment by training his body to use other mechanical means to achieve fast speeds.

Although Weyand et al. (12) stated “more rapid repositioning of limbs contributes little to the faster top speeds of swifter runners,” Weyand and Bundle (14) argue that Pistorius is able to run fast because his lightweight prostheses allow him to rapidly reposition his legs during the swing phase. Brief leg swing times increase the fraction of a stride that a leg is in contact with the ground and thus reduce the vertical impulse requirement for the contact phase. But, the notion that lightweight prostheses are the only reason for Pistorius’ rapid swing times ignores that he has had many years to train and adapt his neuromuscular system to using prostheses. Weyand and Bundle (14) argue that lightweight prostheses allow Pistorius to run faster than he should for his innate strength/ability to exert vertical GRFs. An equally plausible hypothesis is that he has adopted rapid leg swing times to compensate for the force limitations imposed by his prostheses.

Pistorius’ leg swing times are not unreasonably or unnaturally fast. Nonelite runners have mean (SD) minimum leg swing times of 0.373 (0.03) s (12). Pistorius’ leg swing time of 0.284 s at 10.8 m/s is nearly 3 SD faster than that mean. However, leg swing times as low as 0.31 s for Olympic 100-m medalists at top speed have been reported (12). If elite sprinters have similar variation in leg swing times, then a leg swing time of 0.284 s is not aberrant. Furthermore, recreational athletes sprinting along small radius (1 m) circular paths exhibited mean leg swing times of just 0.234 s (5). It appears that when faced with stringent force constraints, runners with biological legs choose very short leg swing times. A thorough study of leg swing times for elite Olympic and Paralympic sprinters could provide further perspective.

Fortunately, there are simple experiments with testable hypotheses that can resolve many of the issues presented here. We propose a comprehensive biomechanical study of high-speed running by elite, unilateral amputee athletes. Studying unilateral amputees would allow direct comparisons between their affected and unaffected legs. First, we hypothesize that unilateral amputee sprinters exert greater vertical GRFs with their unaffected leg than with their affected leg. If that hypothesis is supported by data, it would indicate that RSP impose a force limitation and are thus disadvantageous. Second, we hypothesize that unilateral amputee sprinters run with equally rapid leg swing times for their affected and unaffected legs. If that hypothesis is supported, it would dispel the idea that lightweight prostheses provide a leg swing time advantage. Third, we hypothesize that adding mass to the lightweight RSP of unilateral and bilateral amputees will not increase their leg swing times or decrease their maximum running speeds. If that hypothesis is supported, then the assertion that the low inertia of RSPs provide an unnatural advantage would be discredited. Given that some Paralympic sprinters choose to add mass to their prostheses, we anticipate that added mass will not significantly slow leg swing times. Future experiments should also quantify how RSPs affect accelerations and curve running. Both require greater force and power outputs than straight-ahead steady speed running. We hope that the data needed to test these hypotheses will be forthcoming so that this debate can be elevated from a discussion of what might be to a discussion of what is known.


  1. 1.
  2. 2.
  3. 3.
  4. 4.
  5. 5.
  6. 6.
  7. 7.
  8. 8.
  9. 9.
  10. 10.
  11. 11.
  12. 12.
  13. 13.
  14. 14.
  15. 15.