Symmetry for Symmetry’s Sake: A Beneficial Goal in Prosthetic Gait?

By Phil Stevens, MEd, CPO, FAAOP

Nearly 30 years ago, authors David Winter, PhD, PEng, and Susan Sienko, BSc, challenged some fundamental assumptions associated with prosthetic gait that continue to this day. "Throughout all the amputee-related literature," they assert, "continuous references are made to variables that establish gait asymmetry. Subsequently, attempts are being made, without scientific justification, to force the amputee to walk more symmetrically." They continue, " cautious about gait retraining protocols which are aimed at improved symmetry based on nothing more than an idea that it would automatically be an improvement." Summarizing their position, the authors conclude, "It is safe to say that any human system with major structural asymmetries in the neuromuscular skeletal system cannot be optimal when the gait is symmetrical. Rather, a new nonsymmetrical optimal is probably being sought by the amputee with the constraints of his residual system and the mechanics of his prosthesis."1

For the thoughtful clinician, the long-held assertion of symmetry for symmetry's sake is immediately replaced by questions of just how much and which types of symmetry might constitute this "new nonsymmetrical optimal." This article reviews recent contributions to the arguments against the standard of symmetry for symmetry's sake.

Dissecting Step-length Symmetry

Observations released by Roerdink et al. in Gait & Posture in 2012 suggest that, during ambulation with a lower-limb prosthesis, step-length asymmetry can be reasonably viewed as a culmination of two related asymmetries in trunk progression and forward foot placement.2 Figure 1, adapted from Roerdink et al.'s study, illustrates one possible scenario in which an observable step-length asymmetry is a product of asymmetric trunk progression from step to step. This might result from an inability of the prosthetic leg to match the forward propulsion attained by the sound leg. Alternately, in cases of transfemoral amputations, hip flexor tightness might prevent symmetric advancement of the torso during the prosthetic step (swing phase of the sound side).

Figure 1

Figure 1: Asymmetrical step length characterized by symmetrical forward foot placement and asymmetrical trunk progression during the sound-side step (prosthetic stance). Adapted from Roerdink et al.2

Figure 2 illustrates a second scenario in which there is symmetric trunk progression but step-length asymmetry results from asymmetric forward foot placement values relative to trunk position. The authors suggest that such observations "concur with the reduced ability to swing the prosthetic leg forward during the nonprosthetic stance phase due to, for example, the loss of muscle function following amputation, altered inertial properties of the prosthetic leg, and/or socket prosthesis geometry on the affected side."2 As discussed later in this article, others have suggested that a reduced forward foot placement may also be an adaptation to preserve dynamic gait stability.

Figure 2

Figure 2: Asymmetrical step length characterized by symmetrical trunk progression and asymmetrical forward foot placement during the prosthetic step (sound-side stance). Adapted from Roerdink et al.2

Importantly, when they occur in tandem, these two asymmetries can also largely mask one another, as seen in Figure 3. This model suggests a gait that may initially appear symmetric to the patient and outside observers because of symmetric step lengths but is actually highly asymmetric.

Figure 3

Figure 3: Apparent symmetrical step length characterized by both asymmetrical forward foot placement and asymmetrical trunk progression. Adapted from Roerdink et al.2

To evaluate their theory, the authors collected gait analysis data at slow and comfortable walking speeds on a convenience sample of seven individuals with unilateral transfemoral amputations and three individuals with unilateral transtibial amputations. The cohort was diverse with respect to gender (eight males), age (29-68 years old), amputation etiology (five vascular and five traumatic), and experience with a prosthesis (nine months to 54 years).2

The authors observed inconsistent variations in both the direction and magnitude of step-length asymmetry and forward foot placement asymmetry between the prosthetic and nonprosthetic steps. By contrast, the observed trunk progression values were greater during the prosthetic step (or sound-side stance phase) than those observed during the sound-side step (or prosthetic stance phase). This appears to result from the inadequate propulsion-generating capacity of the prosthetic leg relative to the sound limb. This theory was reinforced by the observation that asymmetries in trunk progression decreased at slower walking speeds, suggesting reduced discrepancies in the propulsion provided by the sound and prosthetic limb as gait speed decreased.

Trunk progression asymmetry and forward foot placement asymmetry tended to be negatively correlated. In other words, the phenomenon illustrated in Figure 3 was frequently observed. Asymmetries in trunk progression or forward foot placement were offset by one another to produce apparent step-length symmetry. As a result, a prosthetic gait characterized by symmetrical step lengths, a characteristic generally viewed as desirable in rehabilitation for individuals with amputations, is often obtained by asymmetries in trunk progression and forward foot placement that are opposite in direction but similar in value. Obtaining symmetric step lengths in prosthetic gait appears to be an exercise in offsetting asymmetries in trunk progression due to inadequate propulsion with comparable asymmetries in forward foot placement.

Kinematics Versus Kinetics: Cycling Model

Prosthetic gait is a combination of what we see (joint angles or kinematics) and the forces and moments that produce that visible movement (kinetics). Clinicians tend to assume that when you minimize visible kinematic asymmetries you are also minimizing the underlying kinetic asymmetries. However, this assumption has been analyzed and challenged using an amputee cycling model.3

While cycling is not ambulation, it is a rhythmic motor task that allows for the isolation of locomotive propulsion. During able-bodied cycling, the ankle plantarflexes on the downstroke and dorsiflexes on the upstroke, traversing approximately 20 degrees in ankle motion. When this motion is not available, such as during prosthetic cycling, additional compensatory motion is required from the knee and hip.3 This was verified in the work of W. Lee Childers, PhD, MSPO, CP, and Géza Kogler, PhD, CO/L, BCO, LPed, in their observations of a convenience sample of eight people with unilateral amputations, in which the researchers measured the kinematics observed at the hip, knee, and ankle of both the sound limb and affected limb as subjects cycled at a fixed rate.

The need for compensatory knee and hip motion can be reduced if the crank shaft on the side of the prosthesis is shortened by approximately 1cm, obviating the need to lengthen the limb at the bottom of the peddle stroke and shorten it at the top of the stroke. This phenomenon was also confirmed when shortening the crankshaft on the side of the prosthesis restored hip and knee kinematics to those values on the sound side.3

However, the key clinical question in this laboratory trial was whether or not a restoration of normal lower-limb kinematics in a lower-limb motor task would lead to a similar restoration in sound-side kinetics, or the forces and moments used to produce the limb movements. Put more succinctly, would reductions in kinematic asymmetries correlate with reductions in kinetic asymmetries? The answer was a fairly consistent no. In fact, the kinetics observed on the affected limb became even more asymmetric. This model raises important questions about the goal of observable gait symmetries. Even when such asymmetries are reduced, the underlying kinetic asymmetries may persist or even become exaggerated.3

Kinematics Versus Kinetics: Ambulation Case Study

Under Childers' direction, this paradigm is explored with respect to ambulation symmetries in a case study about a pilot trial using an instrumented split-belt treadmill. In the pilot trial, a young subject with a unilateral transtibial amputation walked on a treadmill with the belts set to symmetrical speeds (1.2m/s). In this condition, the patient demonstrated a slightly longer sound-side step. In subsequent trials, the belt speed on the side of the amputation was successively increased to 1.25, 1.3, and 1.35m/s. Increasing the belt speed beneath the prosthesis, it was hypothesized, would require the patient to lengthen her stride on that side, producing a more symmetrical gait.4

The case subject demonstrated a trend of reduced hip range of motion and step-length asymmetry on the affected side as the belt speed was increased. However, because the treadmill was instrumented, it was also observed that the peak vertical ground reaction force experienced by the sound-side limb increased with each successive speed increase on the contralateral belt. While pilot data should always be analyzed with caution, this preliminary data suggests that the efforts made to improve gait symmetry failed to produce symmetrical limb loading.4 Indeed, in this instance, limb loading became more asymmetric, increasing the loads experienced by the sound-side limb.

Are There Functional Benefits of Asymmetry?

In addition to the aforementioned concepts and concerns, another recent publication suggests a potential functional benefit associated with step-length asymmetry. Unfortunately, it gets a bit complicated, so the reader is advised to either sit down for this next part, or simply skip to the summary at the end of the article.

Prior research has suggested that in both able-bodied and prosthetic-assisted ambulation, when people experience perturbations during ambulation, their preference is biased in favor of a forward loss of balance rather than a backward loss of balance. Restated, we would rather modify the next few steps forward through temporary increases in step length to recover our balance, than need to step backward and lose our forward momentum.5

The gait parameter that reflects this preference has been described by Hak et al. as the "backward margin of stability" (BW MoS).5 The length of the BW MoS during a single step is determined by the base of support of the leading leg (estimated by the position of the lateral malleolus) and a variable described as "the extrapolated center of mass" (eCoM). The latter represents the current horizontal position of the CoM, its forward velocity, and the pendulum length of the leg.5 For our purposes, it can be thought of as where the CoM is about to be.

The distance from the eCoM (located well in front of the leading leg) back to the base of support of the leading leg is the BW MoS, and able-bodied and prosthesis-assisted ambulators alike want that value to be fairly large. We want to know that, in the event of a gait perturbation, the forward trajectory of our CoM will still make it over our base of support. We may need to modify our next few steps to reestablish our stability but we'd rather do that than try to stop the forward momentum of the trailing limb and reach backward to find our balance.

Within this model, if an apropulsive prosthetic foot reduces the forward propulsion of the body's CoM during the sound-side step, the eCoM will be shortened (recall that eCoM is a product of forward velocity). In anticipation of this event, a shortened forward foot placement can increase the BW MoS to better ensure dynamic gait stability.5

To test this hypothesis, the proponents evaluated these gait variables in ten individuals with unilateral transtibial amputations. In contrast to the first study described in this article, step-length asymmetry for all study subjects was consistently biased toward a shortened sound-side step. This was found to be the product of a shortened sound-side forward foot placement, which the authors suggest may have occurred to increase the "backward moment of stability" in the presence of the reduced eCoM, giving the patients confidence in their ability to quickly and consistently get over the base of support of the sound-side limb in the event of gait perturbation.5


The value of step symmetry as an ideal characteristic in prosthetic gait was questioned as early as the 1980s. The arguments against symmetry have since become more nuanced and refined. When it does occur, it may do so as a consequence of additional secondary asymmetries produced to mask primary asymmetries innate to the patient's physical and prosthetic condition. Further, efforts to reduce kinematic asymmetries have not resulted in improved symmetries in the underlying kinetics. Additionally, it has now been suggested that a shortened sound-side step length may be an integral part of preserving dynamic balance during ambulation. These arguments collectively suggest that the traditional goal of symmetric gait may be overly simplified. Rather, the field may need to arrive at a more informed understanding of the ideal role of symmetry in producing the "new nonsymmetrical optimal" Winter and Sienko proposed almost 30 years ago.

Phil Stevens, MEd, CPO, FAAOP, is in clinical practice with Hanger Clinic, Salt Lake City. He can be reached at .


  1. Winter, D. A., and S. E. Sienko. 1988. Biomechanics of below-knee amputee gait. Journal of Biomechanics 21(5):361-67.
  2. Roerdink, M., S. Roeles, S. C. H. van der Pas, O. Bosboom, and P. J. Beek. 2012. Evaluating asymmetry in prosthetic gait with step-length asymmetry alone is flawed. Gait & Posture 35 (3):446-51.
  3. Childers, W. L., and G. F. Kogler. 2014. Symmetrical kinematics does not imply symmetrical kinetics in people with transtibial amputation using a cycling model. Journal of Rehabilitation and Research Development 51 (8):1243-54.
  4. Coleman, T., H. Lawrence, A. Komolafe, and W. L. Childers. 2016. Prosthetic gait analysis: A case study examining the relationship between kinematic and kinetic symmetries. Proceeds of the American Physical Therapy Association's NEXT Conference and Exposition, Nashville, TN, USA, June 8-11, 2016.
  5. Hak, L., J. H. van Dieen, P. van der Wurff, and H. Houkijk. 2014. Stepping asymmetry among individuals with unilateral transtibial limb loss might be functional in terms of gait stability. Physical Therapy 94 (10):1480-8.