Comparing the mechanical energetics of walking among individuals with unilateral transfemoral limb loss using socket and osseointegrated prosthetic interfaces

Abstract

Osseointegration (OI), or bone-anchoring, of a prosthesis is a transformative procedure for addressing issues of socket fit among individuals with lower limb loss. Mechanically, the removal of the socket interface substantially alters the transmission of load and the flow of energy through the prosthetic limb. Here, we compared the mechanical energetics of walking between socket and OI interfaces using biomechanical data and custom models of 8 service members pre- and ~24-months post-OI. Relative to a socket interface, an OI interface shifted loads toward the intact limb, which increased collision losses, while the net mechanical work of both the prosthetic and intact limbs remained minimal for both interfaces. At the joint level, despite the removal of the socket interface potentially saving ~ 6 J of work per stride, these reduced collision losses were transmitted to the center of mass instead of altering joint work. The principal change in prosthetic limb joint mechanics with an OI interface was a decrease in negative prosthetic limb hip work during late stance, driven by a decreased hip flexion moment and prosthetic limb offloading. Our findings suggest that despite previously reported improvements in walking economy, after OI individuals likely walk with increased mechanical energetic asymmetry.

Keywords: Biomechanics; Bone-anchored prostheses; Ground reaction forces; Lower limb amputation; Mechanical work.

Fig. 1

Custom musculoskeletal models of individuals with unilateral transfemoral amputation (A) using socket suspension, with a translational degree of freedom between the socket and residual femur directed along the long axis of the femur, and (B) with an osseointegrated interface, where the residual femur is rigidly attached to the abutment and prosthesis. This figure was generated using OpenSim 4.4 (https://simtk.org/projects/opensim) and Adobe Illustrator 2024 (https://www.adobe.com/products/illustrator.html).

Fig. 2

(A) Across-participant ensemble averaged limb powers and (B) mean limb work values across slow (0.6–1.1 m s^− 1) and fast (1.1–1.6 m s^− 1) walking speeds for socket and OI interfaces. Shaded regions and error bars represent ± 1 s.d. Values were normalized to each participant’s body mass. Significant effects (P < 0.050) of interface, speed, and interface-speed interaction were determined using two-way repeated measures ANOVAs and are graphically represented as an osseointegrated femur, speedometer, and their interaction. Statistical results of the effect of limb within each interface and pairwise comparisons are reported in text.

Fig. 3

Across-participant (A) ensemble averaged vertical ground reaction forces (vGRFs) and (B) maximum vGRFs across slow (0.6–1.1 m s^− 1) and fast (1.1–1.6 m s^− 1) walking speeds for socket and OI interfaces. Shaded regions and error bars represent ± 1 s.d. Values were normalized to each participant’s body mass. Significant effects (P < 0.050) of interface, speed, and interface-speed interaction were determined using two-way repeated measures ANOVAs and are graphically represented as an osseointegrated femur, speedometer, and their interaction. Statistical results of pairwise comparisons are reported in text.

Fig. 4

Across-participant ensemble averaged (A) socket and (C) hip powers with (B, D) associated positive and negative mean work values across slow (0.6–1.1 m s^− 1) and fast (1.1–1.6 m s^− 1) walking speeds for socket and OI interfaces, where applicable. Hip power was summed across all 3 degrees of freedom of the hip joint. Shaded regions and error bars represent ± 1 s.d. Values were normalized to each participant’s body mass. Significant effects (P < 0.050) of interface, speed, and interface-speed interaction were determined using two-way repeated measures ANOVAs and are graphically represented as an osseointegrated femur, speedometer, and their interaction. Statistical results of pairwise comparisons are reported in text. Socket work values were compared between speeds using pairwise t-tests.

Fig. 5

Across-participant prosthetic side ensemble averaged (A) sagittal hip moments and (C) sagittal hip angular velocities with (B, D) associated maxima at slow (0.6–1.1 m s^− 1) and fast (1.1–1.6 m s^− 1) walking speeds for socket and OI interfaces. Shaded regions and error bars represent ± 1 s.d. Moments were normalized to each participant’s body mass. Significant effects (P < 0.050) of interface, speed, and interface-speed interaction were determined using two-way repeated measures ANOVAs and are graphically represented as an osseointegrated femur, speedometer, and their interaction. Statistical results of pairwise comparisons are reported in text.