Gait differences between K3 and K4 persons with transfemoral amputation across level and non-level walking conditions

Abstract

Background:

A transfemoral amputee’s functional level can be classified from K-level 0 (lowest) to K-level 4 (highest). Knowledge of the biomechanical differences between K3 and K4 transfemoral amputation could help inform clinical professionals and researchers in amputee care and gait assessment.

Objectives:

Explore gait differences between K3- and K4-level transfemoral amputation across different surface conditions.

Study design:

Cross-sectional study.

Methods:

Four K3 and six K4 transfemoral amputation and 10 matched able-bodied individuals walked in a virtual environment with simulated level and non-level surfaces on a self-paced treadmill. Stability measures included medial-lateral margin of stability, step parameters, and gait variability (standard deviations for speed, temporal-spatial parameters, root-mean-square of medial-lateral trunk acceleration).

Results:

K3 walked slower than K4 with wider steps, greater root-mean-square of medial-lateral trunk acceleration, and greater medial-lateral margin of stability standard deviations, indicating their stability was further challenged. K3 participants had greater asymmetry in double support time and trunk acceleration root-mean-square in the medial-lateral direction, but similar asymmetry overall. K3 participants had larger differences from AB and in more parameters than K4, although K4 differed from AB in trunk acceleration root-mean-square in the medial-lateral direction, walking speed, and double support time standard deviations.

Conclusion:

The findings improve our understanding of K3 and K4 transfemoral amputation gait on slopes and simulated uneven surfaces.

Clinical relevance

High performing and community ambulatory transfemoral amputees cannot match the ambulatory abilities of ablebodied individuals. Understanding gait differences between these groups under conditions that challenge balance is required to develop rehabilitation protocols and prosthetic componentry targeted at improving transfemoral amputee gait and overall mobility in their chosen environment.

Keywords

Gait analysis gait rehabilitation of amputees rehabilitation K-level transfemoral amputee slopes

Background

Forward ambulation challenges stability because the center of mass (CoM) is constantly moving. For people with transfemoral amputation (TFA), additional difficulties in maintaining dynamic stability are posed by the loss of sensory feedback mechanisms and force generation in their prosthetic knee and ankle. In comparison with able-bodied (AB) populations, TFA walk slower,^1–6 with shorter, wider, and longer duration steps,¹ and more pelvic and trunk motion.^2–4 A primary goal of rehabilitation following a lower limb amputation is restoring normal gait, achieved through a combination of prosthetic technology and specific gait training. While advances in both areas greatly improve the potential for “normal” gait, even individuals who are capable of advanced levels of prosthetic mobility still do not perform as well as their AB peers.

TFA limb asymmetries include smaller ankle and knee motion and pelvic drop on the prosthetic side compared to the intact side^7–10 and increased lateral trunk bending over the prosthetic limb that can have multiple causes, such as weak hip abductor stabilizing muscles.¹¹ Double support time (DST) is also longer for the intact limb than the prosthetic limb.¹¹ Limb asymmetry evaluation is particularly important because greater joint moments and forces on the intact limb, compensating for prosthetic limb deficiencies, could lead to intact limb osteoarthritis.^12–14

TFA also have greater medial-lateral margins of stability (ML-MoS) on their prosthetic side than both their intact side and AB participants,⁵ which is considered a stability-enhancing strategy.⁶ Increased gait variability has also been associated with greater instability,^15–19 which could relate to risk of falls. In our previous study,²⁰ TFA differed from AB for root-mean-square of trunk acceleration in the medial-lateral direction (RMS_ML), walking speed, step width, ML-MoS, and gait variability.

Non-level conditions likely further exacerbate these differences in transfemoral prosthetic gait. TFA who have the ability to walk in the community are challenged on a daily basis to adapt to different walking terrains. Using K-level functional classification,²¹ individuals at K3 are community ambulators with the ability to transverse most environmental barriers and may have vocational, therapeutic, or exercise activity that demands prosthetic use beyond simple locomotion. K4 level exceeds basic ambulation skills, exhibiting high impact, stress, or energy levels. Given the higher locomotion demands faced by these individuals, it is important to understand how gait mechanics differ from AB individuals. This information could help improve clinical treatment programs and further optimize prosthetic technologies to prevent long-term complications.

The purpose of this study was to determine differences in TFA (K3 or K4 level), and AB gait over level and non-level conditions. Given the greater physical demands on K3 and K4 TFA, more biomechanical research is needed on non-level, realistic walking conditions rather than only level walking. Based on previous TFA gait research between TFA and AB, the lower functional level K3 participants should walk slower than K4, with wider steps, greater ML-MoS, greater trunk acceleration RMS_ML, and more variable and asymmetric gait patterns. We also hypothesized that K3 differs from AB for more outcome measures and more walking conditions than K4. Normalizing gait is important to prevent future complications such as osteoarthritis,²² and studying gait mechanics is an essential component. Since the literature comparing K3 and K4 populations is limited, walking biomechanical knowledge across surfaces other than level, such as up and down slopes, cross slopes, rocky, will inform practice for researchers, prosthetists, therapists, and physicians.

Methods

Participants

A convenience sample of 10 individuals with unilateral TFA (Table 1) and 10 age, sex, height, and mass-matched AB volunteers (8 males, 2 females; age = 43 ± 8.6 years; mass = 83.2 ± 10.1 kg; height = 1.78 ± 0.07 m) were recruited from The Ottawa Hospital Rehabilitation Centre, Canadian Forces Health Services, Glenrose Rehabilitation Hospital, or the community. Participant K-level was determined by clinical assessment prior to the study. All AB participants had a dominant right leg (preferred limb to kick a ball). The Ottawa Health Science Network Research Ethics Board and University of Alberta Research Ethics Board approved this study and written informed consent was obtained.

Table 1.

Participant demographics.

K-level	Sex	Age (years)	Mass (kg)	Height (m)	Prosthetic side	Etiology	Time since amputation (years)	Prosthetic foot	Prosthetic knee
K3	F	53	93.2	1.70	R	Osteosarcoma	1	C-walk	C-leg
K3	M	48	86.4	1.85	L	Osteosarcoma	27	Axtion	C-leg
K3	M	42	95.5	1.75	R	Trauma	7	Axtion	C-leg
K3	M	62	86.4	1.70	L	Trauma	34	Triton	Genium X3
K4	M	47	75.5	1.78	L	Trauma	25	Triton	Genium X2
K4	F	43	67.7	1.57	L	Trauma	15	Triton	C-leg
K4	M	43	91.8	1.83	R	Trauma	15	LP Variflex	Endolite KX06
K4	M	37	79.5	1.84	L	Osteosarcoma	0.7	Variflex	Mauch
K4	M	62	88.6	1.85	L	Vascular	6	Ossur Rotate LP	Truelife
K4	M	35	88.6	1.78	L	Trauma	4	Triton	Genium X3

Data collection

The CAREN–Extended virtual reality system (Motekforce Link, Amsterdam, NL), which includes a 12-camera motion capture system (Vicon Inc., Oxford, UK), 180° projection screen, and 6 degree-of-freedom motion platform embedded with a dual belt instrumented treadmill (Bertec Corp., Columbus, OH, USA) was used for all trials (Figure 1). The system is capable of simulating real-world environments by matching platform motion to the programmed visual surround. Participants wore a safety harness that prevented ground contact in a fall, yet permitted movement along treadmill dimensions. Three markers tracked platform motion. A 57 marker set tracked full body kinematics²³ (Figure 2), and a digitizing pointer defined 20 additional anatomical landmarks (C-Motion Inc., Germantown, MD, USA). Marker data were sampled at 100 Hz.

Figure 1.

CAREN–Extended system with Park application.

Figure 2.

Full body marker set.

Participants completed a 10 to 15-min virtual environment acclimation period that included self-paced treadmill training and a warm-up trial. Self-paced control²⁴ automatically adjusted treadmill speed to match participant walking speed, with 0.7 sensitivity scaling, 1 m/s² maximum treadmill acceleration, and equal visual scene speed (optic flow) and treadmill speed.

Using self-paced mode, participants completed two to three walking trials through a virtual park scene. Each walking condition was a 20-m section (LW: level walking, DS: downhill slope, US: uphill slope, TS: top-cross-slope, BS: bottom-cross-slope, HL: rolling-hills, MLT: medial-lateral translations, RO: simulated rocky; Table 2). These conditions represented real-world surfaces or continuous surface perturbations in different axes. Each trial started from standing, an initial level walking period, and ended with a level walking section. Intermediate sections were randomized and separated by level sections (340 m total per trial). Level walking at the beginning of the trial, after reaching steady state, was used for the level condition. Participants could take breaks between trials. Four TFA intermittently grasped one treadmill handrail; however, analyzed gait cycles were without handrail use.

Table 2.

Walking activities included in the virtual park scenario.

Walking activity		Description
Level	Level (LW)	Level treadmill platform.
Slope	Downhill (DS)	−7° decline (pitch).
	Uphill (US)	+7° incline (pitch).
	Top-cross-slope (TS)	Limb being analyzed at the top of the cross-slope. Platform tilts (roll) to the right or left with a 5° slope.
	Bottom-cross-slope (BS)	Limb being analyzed at the bottom of the cross-slope. Platform tilts (roll) to the right or left with a 5° slope.
Uneven	Medial-lateral translations (MLT)	Platform oscillates in the medial-lateral direction based on a sum of four sines with frequencies of 0.16, 0.21, 0.24, and 0.49 Hz.²⁴ Amplitude scaling, A_w = 0.015, yields a maximum range of ±4 cm.
	Rolling-hills (HL)	Platform oscillates in the sagittal plane (pitch) based on a sum of four sines with frequencies of 0.16, 0.21, 0.24, and 0.49Hz.²⁴ Amplitude scaling, A_w = 0.01, yields a maximum range of ±3°.
	Rocky (RO)	Platform oscillates in three directions simultaneously using a CAREN module with a maximum range of ±2 cm at 0.6 Hz vertically, ±1° at 1 Hz pitch, and ±1° at 1.2 Hz roll.

Data Analysis

Marker data were filtered using a fourth-order low-pass Butterworth filter (10 Hz). A 13-segment body model^25,26 was created in Visual 3D 4.96.4 (C-Motion, Inc.). Gait events were identified from foot velocity relative to the pelvis.²⁷ Thirty gait cycles were selected for each walking condition. Anterior–posterior foot marker velocity during mid-stance was used to calculate treadmill velocity for each stride.

Mean and within-subject variability (standard deviations (SD)) of walking speed, step width, DST, and ML-MoS were calculated (Table 3). RMS_ML measures acceleration and gait pattern smoothness,²⁸ stability, and variability.^29,30 SD represents within-subject variability³¹ and has identified level and non-level surface differences between AB and balance-impaired groups.^32–35

Table 3.

Measures examined to quantify stability: medial-lateral margin of stability (ML-MoS), step parameters, and gait variability.

	Outcome measures	Description
MoS	Medial-lateral margin-of-stability (ML-MoS)	Minimum medial-lateral distance between the base-of-support (BOS) boundary, defined by a lateral foot marker (fifth metatarsal), and the extrapolated center-of-mass (COM) during early stance.^{36, 37} Marker positions were defined relative to the laboratory reference frame.
Step strategies	Step width, dominant/intact limb	Medial-lateral distance (m) between heel markers at dominant/intact limb foot contact. Marker positions were resolved in the platform reference frame.
	Step width, non-dominant/prosthetic limb	Medial-lateral distance (m) between heel markers at non-dominant/prosthetic limb foot contact. Marker positions were resolved in the platform reference frame.
	Double support time, dominant/intact limb	Time (s) between prosthetic (non-dominant) foot contact and intact (dominant) toe off. Reported as percent of stride time.
	Double support time, non-dominant/prosthetic limb	Time (s) between and intact (dominant) foot contact and prosthetic (non-dominant) toe off. Reported as percent of stride time.
Variability	ML-MoS mSD	Standard deviation of ML-MoS
	Step width mSD	Standard deviation of step width
	Double support time mSD	Standard deviation of double support time
	RMS_ML	Root-mean-square of trunk acceleration in the medial-lateral (ML) directions, relative to the laboratory reference frame. Reported in units of gravity (g), by step normalized to 101 points.

mSD: mean of standard deviation.

Outcome measures were calculated using custom MATLAB programs (2010a, Mathworks, Natick, MA, USA) and R v2.14 (R Foundation for Statistical Computing, Vienna, Austria) was used for all statistical analyses. Data were examined using separate linear mixed models (lmer4 package) to examine group and limb main effects and group × limb interactions, for each walking condition. Fixed factors were group (AB, K3, K4) and limb (dominant/intact and non-dominant/prosthetic), and participant was the random factor. Analyses were also performed with a treadmill speed covariate to understand how speed affected findings.

Significant main effects were examined using multcomp package post hoc tests (p < 0.05). To minimize between group comparisons, limb comparisons were intact versus dominant and prosthetic versus non-dominant. Step width effects on RMS_ML were examined since step width can influence trunk motion.⁷ As an exploratory study with small sample size, results and discussion were reported without a statistical correction for multiple tests. All statistical results reported include speed as a covariate. The results report statistical comparisons after considering main effects and any interactions.

Results

K3 and K4 comparison

K3 participants walked slower than K4 for LW, TS, BS, and HL conditions (p < 0.031, Table 4). K3 walked with wider steps than K4 for BS and MLT (p < 0.046). K3 participants walked with wider steps than K4 when their prosthetic limb was on the TS (p < 0.003).

Table 4.

Mean of treadmill speed, temporal-spatial parameters, and margin-of-stability (ML-MoS).

		Limb	LW	DS	US	TS	BS	MLT	HL	RO
Speed (m/s)	AB		1.40 (0.13)	1.44 (0.20)	1.24 (0.18)	1.41 (0.08)	1.44 (0.14)	1.46 (0.14)	1.30 (0.20)	1.42 (0.17)
	K4		1.16 (0.29)^a	1.01 (0.33)^a	1.07 (0.29)	1.21 (0.29)^a	1.22 (0.28)^a	1.22 (0.30)^a	1.07 (0.29)^a	1.18 (0.32)^a
	K3		0.89 (0.13) ^a	0.75 (0.22)^a	0.85 (0.17)^a	0.91 (0.19) ^a	0.95 (0.17) ^a	1.02 (0.08)^a	0.76 (0.10) ^a	0.91 (0.14)^a
Step width (cm)	AB	Dominant	13.1 (2.6)	14.6 (2.4)	11.8 (2.9)	11.7 (3.5)	11.5 (3.1)	14.6 (4.3)	14.3 (4.2)	16.0 (5.5)
	AB	Non-dominant	13.1 (2.6)	14.5 (2.4)	11.9 (2.9)	11.7 (3.0)	11.5 (3.6)	15.1 (4.3)^b	14.3 (4.3)	16.1 (5.4)
	K4	Intact	15.4 (3.7)	15.9 (3.0)	19.7 (5.3)^a	15.6 (4.3)	15.2 (3.0)	16.4 (4.0)	15.8 (3.5)	15.6 (2.9)
	K4	Prosthetic	15.6 (3.5)	16.1 (3.2)	19.7 (5.4)^a	15.3 (3.0)	15.2 (3.7)	16.3 (4.4)	15.9 (3.7)	15.9 (2.8)
	K3	Intact	23.1 (6.8)^a	20.6 (0.57)	26.8 (9.4)^a	23.0 (8.1)^a	23.1 (6.1) ^a	23.4 (7.7)	24.4 (6.7)	26.0 (7.6)
	K3	Prosthetic	23.1 (6.9)^a	20.3 (1.4)	26.7 (8.7)^a	23.2 (5.9) ^a,b	22.9 (7.9) ^a	24.0 (7.13)	24.6 (6.6)	25.9 (7.3)
DST (%)	AB	Dominant	13.4 (0.9)	12.4 (1.5)	14.1 (0.9)	13.7 (1.1)	13.3 (0.6)	12.9 (0.7)	13.5 (0.8)	13.2 (0.6)
	AB	Non-dominant	13.0 (1.0)	11.9 (1.7)	13.7 (1.1)	12.6 (1.0)	13.1 (0.9)	12.2 (1.0)	13.1 (1.2)	12.5 (1.1)
	K4	Intact	15.3 (3.2)	16.0 (3.9)	16.1 (3.8)	14.2 (2.3)	14.6 (2.9)	14.8 (3.9)	15.9 (3.8)	15.8 (4.2)
	K4	Prosthetic	15.1 (1.7)	13.8 (2.3)^b	15.1 (1.0)	14.7 (1.4)	14.7 (1.9)	14.8 (1.5)	15.9 (1.5)	15.1 (1.3)
	K3	Intact	17.7 (3.4)	21.6 (0.6) ^a	17.1 (0.8)	16.0 (2.0)	17.3 (0.8)	15.7 (1.7)	19.1 (3.8)	16.7 (1.5)
	K3	Prosthetic	15.6 (1.9)	14.4 (3.2)^b	15.1 (1.1)	15.8 (1.7)	15.7 (1.2)	15.7 (1.7)	16.8 (1.3)	16.0 (1.9)
ML-MoS (cm)	AB	Dominant	7.6 (1.6)	7.9 (1.3)	7.9 (1.4)	7.2 (2.0)	7.5 (1.4)	8.0 (1.8)	8.2 (1.8)	8.5 (2.1)
	AB	Non-dominant	8.0 (1.6)	8.4 (1.4)	8.1 (1.2)	7.3 (1.5)	7.7 (1.3)	8.2 (1.6)	8.5 (1.4)	8.8 (1.7)
	K4	Intact	8.3 (1.3)	8.0 (1.2)	9.0 (1.9)	7.7 (1.2)	8.2 (1.4)	8.4 (1.7)	8.3 (1.3)	8.5 (1.2)
	K4	Prosthetic	9.3 (1.8)	9.6 (1.8)	10.5 (1.8)^a	8.6 (1.9)	10.0 (2.0)	9.6 (2.0)	9.3 (2.0)	9.2 (1.9)
	K3	Intact	10.2 (1.9)	9.2 (2.3)	10.5 (1.7)	9.8 (2.8)	9.8 (1.2)	10.5 (2.3)	10.3 (1.5)	10.8 (2.1)
	K3	Prosthetic	11.2 (2.4)	10.0 (0.5)	12.5 (4.0)^a,b	11.0 (4.0)	11.7 (2.7)	11.4 (2.6)	11.6 (2.8)	12.3 (3.1)

AB: able-bodied group; K3: K-level three transfemoral amputee group; K4: K-level four transfemoral amputee group; DST: double support time; LW: level walking; DS: downhill slope; US: uphill slope; TS: top-cross-slope; BS: bottom-cross-slope; HL: rolling-hills; MLT: medial-lateral translations; RO: simulated rocky.

Boldfaced values indicate significant difference between K3 and corresponding K4 limb.

All significant comparisons include treadmill speed as a covariate, except for the speed parameter.

Standard deviation is in parentheses.

Significantly different from corresponding AB limb.

Significant differences between limbs.

For downhill, K3 had significantly greater DST on the intact limb than the K4 intact limb (p < 0.010). However, DST SD was significantly smaller for K3 compared to K4 for US and MLT (p < 0.042; Table 5). DST SD was also smaller on the K3 prosthetic limb than on the K4 prosthetic limb for HL (p < 0.041), but was greater when walking downhill (p < 0.001).

Table 5.

Mean of the standard deviations (mSD) for treadmill speed, temporal-spatial parameters, margin-of-stability (ML-MoS), and root-mean-square for medial-lateral trunk acceleration (RMS_ML).

		Limb	LW	DS	US	TS	BS	MLT	HL	RO
Speed (m/s)	AB		0.09 (0.04)	0.10 (0.05)	0.10 (0.04)	0.08 (0.04)	0.07 (0.02)	0.08 (0.03)	0.12 (0.04)	0.10 (0.03)
	K4		0.10 (0.05)	0.11 (0.06)	0.13 (0.03)	0.06 (0.04)	0.08 (0.05)	0.06 (0.03)^a	0.11 (0.04)	0.08 (0.04)
	K3		0.10 (0.05)	0.06 (0.06)	0.08 (0.03)	0.06 (0.04)	0.04 (0.02)	0.04 (0.02)^a	0.06 (0.02)	0.08 (0.07)
Step width(cm)	AB	Dominant	2.6 (0.8)	2.7 (0.6)	2.8 (1.0)	3.2 (0.8)	3.4 (1.1)	7.6 (2.7)	3.7 (1.5)	4.3 (1.7)
	AB	Non-dominant	2.5 (0.6)	2.7 (0.6)	2.7 (0.9)	2.9 (0.6)	2.9 (0.9)	7.3 (1.8)	3.6 (1.0)	4.1 (1.1)
	K4	Intact	3.0 (0.3)	2.8 (0.7)	3.7 (0.8)^a	3.5 (0.5)	2.6 (0.7)	5.4 (1.7)	3.8 (0.8)	4.4 (0.6)
	K4	Prosthetic	2.5 (0.7)	2.9 (0.7)	3.2 (0.8)	2.5 (0.5)^b	2.6 (0.7)	4.8 (0.6)	3.3 (0.7)	3.3 (0.7)^b
	K3	Intact	3.1 (0.5)	3.4 (0.6)	4.3 (1.7)^a	3.4 (1.3)	3.3 (1)	5.1 (2.2)	3.7 (0.8)	4.5 (1.1)
	K3	Prosthetic	2.9 (0.3)	4.0 (1.8)	3.8 (1.1)^a	3.3 (1.1)	3 (0.6)	4.8 (1.3)	2.7 (0.7)	4 (1.1)
DST(%)	AB	Dominant	0.7 (0.2)	0.9 (0.2)	0.8 (0.4)	0.7 (0.1)	0.7 (0.2)	0.8 (0.2)	1.5 (0.5)	0.9 (0.2)
	AB	Non-dominant	0.6 (0.2)	1.0 (0.2)	0.9 (0.3)	0.6 (0.2)	0.7 (0.2)	0.7 (0.1)	1.4 (0.3)	0.9 (0.2)
	K4	Intact	1.4 (0.6)^a	2.2 (0.8)^a	2.1 (0.3)^a	1.1 (0.3)	1.3 (0.4)^a	1.4 (0.3)^a	2.2 (0.5)^a	1.6 (0.4)^a
	K4	Prosthetic	1.1 (0.3)^b	1.3 (0.2)^b	1.8 (0.6)^a	1.3 (0.5)^a	1.3 (0.4)^a	1.3 (0.4)^a	2.2 (0.7)^a	1.6 (0.3)^a
	K3	Intact	1.5 (0.6)	2.4 (0.3)^a	1.5 (0.4) ^a	1.3 (0.6)^a	1.6 (0.4)	1.1 (0.2)	1.6 (0.1)	2.0 (0.5)^a
	K3	Prosthetic	1.1 (0.3)	3.2 (2.0) ^a	1.5 (0.6)^a	1.2 (0.6)	1.2 (0.5)	1 (0.3)	1.6 (0.3)	1.9 (0.8)^a
ML-MoS(cm)	AB	Dominant	0.9 (0.2)	0.9 (0.2)	1.0 (0.2)	1.1 (0.3)	1.0 (0.2)	2.4 (0.7)	1.3 (0.4)	1.4 (0.2)
	AB	Non-dominant	0.9 (0.2)	0.9 (0.1)	1.0 (0.2)	1.2 (0.4)	1.0 (0.2)	2.5 (0.8)	1.3 (0.3)	1.4 (0.5)
	K4	Intact	1.0 (0.3)	1.2 (0.2)^a	1.3 (0.3)^a	1.1 (0.3)	1.0 (0.2)	1.8 (0.3)	1.4 (0.3)	1.3 (0.3)
	K4	Prosthetic	0.9 (0.1)	1 (0.2)	1.4 (0.3)^a	1 (0.3)	1.0 (0.1)	1.6 (0.4)	1.2 (0.4)	1.3 (0.2)
	K3	Intact	1.1 (0)	1.2 (0)	1.4 (0.2)^a	1.4 (0.2)	1.4 (0.1) ^a	1.7 (0.3)	1.2 (0.2)	1.6 (0.3)
	K3	Prosthetic	1.1 (0.3)	1.5 (0.8) ^a	2.0 (0.4) ^a,b	1.4 (0.2)	1.2 (0.1)	1.5 (0.4)	1.2 (0.2)	1.7 (0.6)
RMS_ML	AB	Dominant	0.085 (0.01)	0.089 (0.02)	0.085 (0.01)	0.082 (0.02)	0.080 (0.01)	0.095 (0.01)	0.087 (0.01)	0.096 (0.02)
	AB	Non-dominant	0.082 (0.01)	0.086 (0.01)	0.084 (0.01)	0.079 (0.01)	0.077 (0.02)	0.093 (0.02)	0.085 (0.01)	0.095 (0.02)
	K4	Intact	0.104 (0.02)^a	0.113 (0.03)^a	0.120 (0.03)^a	0.103 (0.03)^a	0.109 (0.03)^a	0.112 (0.02)^a	0.110 (0.03)^a	0.111 (0.02)^a
	K4	Prosthetic	0.122 (0.03)^a,b	0.135 (0.03)^a,b	0.140 (0.03)^a,b	0.128 (0.02)^a,b	0.122 (0.03)^a,b	0.133 (0.02)^a,b	0.126 (0.03)^a,b	0.128 (0.03)^b
	K3	Intact	0.119 (0.03) ^a	0.103 (0.00)^a	0.125 (0.02)^a	0.119 (0.02)^a	0.126 (0.03)^a	0.126 (0.02)^a	0.120 (0.01)^a	0.130 (0.02)^a
	K3	Prosthetic	0.150 (0.03) ^a,b	0.132 (0.01)^a,b	0.164 (0.03) ^a,b	0.168 (0.04) ^a,b	0.147 (0.03) ^a,b	0.158 (0.03) ^a,b	0.147 (0.02) ^a,b	0.160 (0.02) ^a,b

Boldfaced values indicate significant difference between K3 and corresponding K4 limb.

All significant comparisons include treadmill speed as a covariate, except for the speed parameter.

Standard deviation is in parentheses.

Significantly different from corresponding AB limb.

Significant differences between limbs,

K3 prosthetic limb had a significantly greater ML-MoS SD than K4 prosthetic limb for DS and US (p < 0.008). K3 intact limb had significantly greater ML-MoS SD than K4 intact limb for BS (p < 0.002).

K3 had significantly greater trunk acceleration RMS_ML when stepping onto their prosthetic side than K4 for all conditions except DS (p < 0.021), where K4 had greater trunk acceleration RMS_ML than K3, but not significantly. K3 also had greater trunk acceleration RMS_ML on their intact side than K4, but only for LW (p < 0.027). When a step width covariate was included in addition to the speed covariate, differences in trunk acceleration RMS_ML were no longer significant between K3 and K4, with the exception of US where K3 and K4 prosthetic limbs were significantly different (p < 0.005).

Comparison to AB

K3 participants walked significantly slower than AB for all conditions (p < 0.002, Table 4) and K4 walked significantly slower than AB for all conditions except US (p < 0.046). Both K3 and K4 had smaller walking speed variability than AB for MLT (p < 0.05). K3 participants had significantly greater step width than AB for LW, US, TS, BS, and MLT (p < 0.042); K4 participants had significantly greater step width than AB for US (p < 0.033); and K3 (p < 0.009) and K4 intact limbs (p < 0.025) had significantly greater step width SD than AB for US.

For downhill, K3 had significantly greater DST on the intact limb compared to AB (p < 0.002). K3 had significantly greater DST SD than AB for DS, US, and RO (p < 0.027), and K4 had significantly greater DST SD on both limbs than AB for US, BS, MLT, HL, and RO (p < 0.008). K4 also had significantly greater prosthetic DST SD than AB for TS (p < 0.001), and significantly greater prosthetic DST SD than AB for LW and DS (p < 0.002).

For uphill, K3 and K4 had a significantly greater ML-MoS on the prosthetic limb than AB (p < 0.039) and greater ML-MoS SD on both limbs (p < 0.01). On the intact limb, K4 had significantly greater ML-MoS SD than AB when walking downhill (p < 0.024) and K3 had greater ML-MoS SD than AB for BS (p < 0.004). K3 also had greater ML-MoS SD than AB on the prosthetic limb for DS (p < 0.016).

K3 and K4 had significantly greater trunk acceleration RMS_ML than AB for all conditions (p < 0.025) except for RO where the K4 intact limb was not significantly different from AB.

Limb asymmetries

K3 had greater step width on their prosthetic limb for TS (p < 0.005; Table 4). Both groups had significantly greater DST on the intact limb for DS (p < 0.029). K3 and K4 also had greater trunk acceleration RMS_ML when on their prosthetic side, for all conditions (p < 0.001). AB had a larger step width on the non-dominant limb for MLT (p < 0.008).

K3 had significantly greater ML-MoS and ML-MoS SD on the prosthetic side than the intact side, for US (p < 0.034). K4 had significantly greater ML-MoS on the intact side than the prosthetic for DS (p < 0.022) and greater ML-MoS SD on the prosthetic side than the intact for BS (p < 0.022).

K4 had significantly greater step width SD on the intact limb than the prosthetic limb, for TS (p < 0.001) and RO (p < 0.001) and significantly greater DST SD on the intact limb than the prosthetic limb for DS. K3 did not exhibit limb asymmetries in step width SD or DST SD.

Discussion

This study investigated TFA gait differences between K3 and K4 functional levels over level and non-level conditions, as well as differences from an AB group. The use of non-level surfaces revealed between-group gait differences that were not apparent with level walking, such as wider step width on cross-slopes and medial-lateral translations; greater ML-MoS SD for downhill, uphill, and bottom cross-slope; and differences between K3 and K4 groups on uphill and downhill walking. As hypothesized, K3 participants generally walked slower than K4 participants with wider steps, greater trunk acceleration RMS_ML, and greater ML-MoS SD. Contrary to expectations, K3 had significantly shorter DST SD than K4 for several conditions, and no significant differences were observed between K3 and K4 for step width SD or ML-MoS. Most K3 and K4 limb asymmetries were similar, but K3 participants exhibited greater asymmetry for DST and trunk acceleration RMS_ML. The outcome measures highlighted in this study could be used to quantify gait differences between TFA functional levels, and as compared to AB populations.

Trunk acceleration RMS_ML was the most consistent parameter that differed between K3 and K4, with greater results for K3 across nearly all walking conditions. Greater trunk acceleration RMS_ML is an indicator of trunk movement instability.³⁸ TFA can have difficulty positioning their trunk over the prosthetic limb, possibly from poor femur stabilization in the socket that effectively weakens the moment arm. Similarly, K3 may have less trunk control and poorer socket stabilization than their K4 counterparts. This could relate to increased trunk muscle activity that increases spinal load,^39,40 leading to low back pain^41–43 that is prevalent within the lower limb amputation population.⁴⁴

K3 participants walked with wider steps than AB in five of the eight walking conditions, but differences from K4 were only significant for cross-slopes and medial-lateral translations. Wider steps increase the base of support and could enhance stability for those with compromised balance.⁵ While step width could have been influenced by prosthetic fit or alignment, step width for level walking was similar to other results in the literature.⁸ Our K4 group had smaller step width and K3 had wider step width than published results.

Since step width has been proposed to influence trunk motion,⁷ a step width covariate was applied when analyzing trunk acceleration RMS_ML. Trunk acceleration RMS_ML differences between groups became insignificant when applying a step width covariate, suggesting that differences between K3 and K4 were related to the K3 group’s larger step width.

K3 did not differ from K4 for mean ML-MoS, but had greater ML-MoS SD for downhill, uphill, and bottom cross slope. This indicated that K3 gait was further challenged than the K4 group when walking on non-level surfaces,⁴⁵ highlighting the relevance of variability measures when assessing movement patterns.

K3 participants walked slower than K4 for all conditions, but were only significant for level, cross slope, and rolling-hills. Considering the well-documented differences in walking speed between AB and balance-impaired populations,^1,11 slower walking for K3 participants than higher functioning K4 was expected, especially over less familiar conditions such as cross slopes. K3 was also 6.8 years older and 8.4 kg heavier than the K4 group, which could contribute to their slower walking speed and overall reduced walking performance. Alternatively, slower walking speed could be related to extra time requirements to plan movements or less gait confidence.⁴⁶ Although K3 walked slower than K4 for downhill and uphill, these differences were not significant. K3 participants might not have needed to reduce speed more than K4, since they were already walking slowly.

Contrary to downhill walking, K3 participants had significantly less DST SD than K4 for the uphill, medial-lateral translations, and rolling-hills. This was unexpected since K3 gait was predicted to be more variable based on their lower functional level. However, differences were less than 1% and may not be clinically significant.

Differences between K3 and K4 in step width, DST, DST SD, and ML-MoS SD were not significant for level walking, but were significantly differently for many other conditions. This demonstrated how the challenge of navigating non-level surfaces can magnify differences between groups, and the importance of training gait on a variety of surfaces. Cross slopes had the most differences between K3 and K4 participants. This could be because, unlike level and inclined surfaces, cross-slope walking requires an asymmetrical gait pattern to keep the body vertical while adapting to different ground heights and angles.⁴⁷ This may pose a greater challenge for maintaining balance and forward progression. Thus, cross slopes should be included in prosthetic gait training.

A functional status progression was observed from K3 to K4 to AB. Overall, differences between the K3 and AB groups were larger and occurred for more parameters than differences between the K4 and AB; however, the higher functioning K4 group still exhibited significant differences from AB participants. The largest differences between K4 and AB were for walking speed, DST SD, and trunk acceleration RMS_ML, for nearly all conditions. K3 and AB differences for nearly all conditions were for walking speed, step width, and trunk acceleration RMS_ML.

Unlike K3, K4 and AB groups had similar step width, with the exception of uphill, which was the only condition where K4 walked faster than AB. Therefore, increased step width was not used as a compensation mechanism by the K4 group. This differed from other studies where TFA walked with wider steps than AB over level ground.^1,11

K4 participants had similar DST magnitudes to AB, but greater DST SD for all conditions. However, K3 only had greater DST SD than AB for four conditions. While walking speed SD and DST SD are strong fall predictors in older adults,¹⁶ no participants indicated falling during the study period (i.e. verbal report to prosthetist during prosthesis status evaluation). Additionally, walking speed SD was only significantly greater in AB participants than K4 and K3 for medial-lateral translations, possibly related to greater AB gait confidence that permitted greater walking speed fluctuations.

Future research in this topic area could expand the number of participants in the K3 and K4 groups, examine other factors that can affect gait across surface conditions, such as prosthetic components and fatigue, and verify the results in non-virtual environments.

Limitations

This study had small groups; however, significant between-group differences were found. K-level was determined by clinical assessment and, while accepted, other models based on movement assessment could be considered. Differences between participant’s prosthetic components could have affected walking performance across activities. However, all participants used prosthetic feet designed for variable speeds and terrain and mechanical knee user outcomes were within the range from microprocessor controlled knee users (no outliers with prosthetic knee).

A CAREN virtual environment was used, with conditions representing real-world scenarios and participants walking at self-selected speeds. While comparable to overground walking for persons with transtibial amputation,⁴⁸ this research was not reproduced with TFA. Therefore, some uncertainty exists for outcome transferability to the real world. However, walking in this virtual environment was similar enough to improve our understanding of how people accommodate to non-level conditions. Use of knee stance phase flexion or yielding during downslope walking was not investigated in this study involving stability-related measures. Medial-lateral translations and rocky conditions only simulated real-world uneven surfaces, because the treadmill was flat, with medial-lateral translations being similar to walking along a crowded hallway and the rocky condition provided a continuous variable platform angle. Additionally, many variables and conditions were analyzed in this exploratory study and the chance of type-I errors was increased.

Conclusion

This study examined differences between K3- and K4-level persons with TFA over level and non-level walking conditions. The most prominent difference between groups was in trunk acceleration RMS_ML, highlighting the importance of including a trunk parameter in gait assessment. K3 participants also walked slower and with wider steps than K4. Even the higher functioning K4 participants differed significantly from AB for walking speed, DST SD, and trunk acceleration RMS_ML. Cross slopes and uphill and downhill walking evoked the most significantly different outcome measures between K3 and K4, and between TFA and AB. This could have been due to the challenging nature of these conditions and potentially less experience walking on cross slopes. Gait training on these surfaces is recommended to ensure optimal mobility in the community.

Footnotes

Author contribution

All authors contributed equally in the preparation of this manuscript.

Declaration of conflicting interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

This work was supported by the Canadian Forces, Surgeon General Health Research Project Fund.

References

Highsmith

Schulz

Hart-Hughes

et al . Differences in the spatiotemporal parameters of transtibial and transfemoral amputee gait. J Prosthet Orthot 2010; 22(1): 26–30.

Sjödahl

Kinematic and kinetic gait analysis in the sagittal plane of trans-femoral amputees before and after special gait re-education. Prosthet Orthot Int 2002; 26(2): 101–112.

Cappozzo

Figura

Gazzani

et al . Angular displacements in the upper body of AK amputees during level walking. Prosthet Orthot Int 1982; 6(3): 131–138.

Darter

Wilken

JM.

Gait training with virtual reality–based real-time feedback: improving gait performance following transfemoral amputation. Phys Ther 2011; 91(9): 1385–1394.

Hof

van Bockel

Schoppen

et al . Control of lateral balance in walking. Experimental findings in normal subjects and above-knee amputees. Gait Posture 2007; 25(2): 250–258.

Hof

Gazendam

MGJ

Sinke

WE.

The condition for dynamic stability. J Biomech 2005; 38(1): 1–8.

James

Oberg

Prosthetic gait pattern in unilateral above-knee amputees. Scand J Rehabil Med 1973; 5(1): 35–50.

Vrieling

van Keeken

Schoppen

et al . Uphill and downhill walking in unilateral lower limb amputees. Gait Posture 2008; 28(2): 235–242.

McNealy

Gard

SA.

Effect of prosthetic ankle units on the gait of persons with bilateral trans-femoral amputations. Prosthet Orthot Int 2008; 32(1): 111–126.

10.

Michaud

Gard

Childress

DS.

A preliminary investigation of pelvic obliquity patterns during gait in persons with transtibial and transfemoral amputation. J Rehabil Res Dev 2000; 37(1): 1–10.

11.

Jaegers

SMHJ

Arendzen

de Jongh

. Prosthetic gait of unilateral transfemoral amputees: a kinematic study. Arch Phys Med Rehabil 1995; 76(8): 736–743.

12.

Lemaire

Fisher

FR.

Osteoarthritis and elderly amputee gait. Arch Phys Med Rehabil 1994; 75(10): 1094–1099.

13.

Radin

Parker

Pugh

et al . Response of joints to impact loading — III: relationship between trabecular microfractures and cartilage degeneration. J Biomech 1973; 6(1): 51–57.

14.

Hurwitz

Ryals

Block

et al . Knee pain and joint loading in subjects with osteoarthritis of the knee. J Orthop Res Off Publ Orthop Res Soc 2000; 18(4): 572–579.

15.

Hausdorff

Rios

Edelberg

HK.

Gait variability and fall risk in community-living older adults: a 1-year prospective study, http://www.sciencedirect.com/science/article/pii/S0003999301632155 (2001, accessed 11 November 2015).

16.

Maki

BE.

Gait changes in older adults: predictors of falls or indicators of fear?

J Am Geriatr Soc 1997; 45(3): 313–320.

17.

Dubost

Kressig

Gonthier

et al . Relationships between dual-task related changes in stride velocity and stride time variability in healthy older adults. Hum Mov Sci 2006; 25(3): 372–382.

18.

Beauchet

Dubost

Herrmann

et al . Stride-to-stride variability while backward counting among healthy young adults. J NeuroEngineering Rehabil 2005; 2(1): 26.

19.

Beauchet

Allali

Berrut

et al . Is low lower-limb kinematic variability always an index of stability? Gait Posture 2007; 26(2): 327–328.

20.

Sturk

Lemaire

Sinitksi

et al . Maintaining stable transfemoral amputee gait on level, sloped, and simulated uneven conditions in a virtual environment, Disabil Rehabil Assist Technol. 2017:1–10. doi:10.1080/17483107.2017.1420250.

21.

Gailey

Roach

Applegate

et al . The Amputee Mobility Predictor: an instrument to assess determinants of the lower-limb amputee’s ability to ambulate. Arch Phys Med Rehabil 2002; 83(5): 613–627.

22.

Gailey

Allen

Castles

et al . Review of secondary physical conditions associated with lower-limb amputation and long-term prosthesis use. J Rehabil Res Dev Wash 2008; 45(1): 15–29.

23.

Wilken

Rodriguez

Brawner

et al . Reliability and minimal detectible change values for gait kinematics and kinetics in healthy adults. Gait Posture 2012; 35(2): 301–307.

24.

Sloot

van der Krogt

Harlaar

Self-paced versus fixed speed treadmill walking. Gait Posture 2014; 39(1): 478–484.

25.

Siegler

Allard

et al . ISB recommendation on definitions of joint coordinate system of various joints for the reporting of human joint motion—part I: ankle, hip, and spine. J Biomech 2002; 35(4): 543–548.

26.

van der Helm

FCT

Veeger

HEJD

et al . ISB recommendation on definitions of joint coordinate systems of various joints for the reporting of human joint motion–Part II: shoulder, elbow, wrist and hand. J Biomech 2005; 38(5): 981–992.

27.

Zeni

JA.

Two simple methods for determining gait events during treadmill and overground walking using kinematic data. Gait Posture 2008; 27(4): 710–714.

28.

Kavanagh

Menz

HB.

Accelerometry: a technique for quantifying movement patterns during walking. Gait Posture 2008; 28(1): 1–15.

29.

Lamoth

CJC

Ainsworth

Polomski

et al . Variability and stability analysis of walking of transfemoral amputees. Med Eng Phys 2010; 32(9): 1009–1014.

30.

Iosa

Morone

Fusco

et al . Effects of walking endurance reduction on gait stability in patients with stroke. Stroke Res Treat 2011; 2012: e810415.

31.

Bland

Altman

DG.

Statistics notes: measurement error. BMJ 1996; 313(7059): 744.

32.

Gates

Scott

Wilken

et al . Frontal plane dynamic margins of stability in individuals with and without transtibial amputation walking on a loose rock surface. Gait Posture 2013; 38(4): 570–575.

33.

McAndrew

Dingwell

Wilken

JM.

Walking variability during continuous pseudo-random oscillations of the support surface and visual field. J Biomech 2010; 43(8): 1470–1475.

34.

Khandoker

Lynch

Karmakar

et al . Toe clearance and velocity profiles of young and elderly during walking on sloped surfaces. J NeuroEngineering Rehabil 2010; 7(1): 1–10.

35.

Marigold

Patla

AE.

Age-related changes in gait for multi-surface terrain. Gait Posture 2008; 27(4): 689–696.

36.

Sinitski

Terry

Wilken

et al . Effects of perturbation magnitude on dynamic stability when walking in destabilizing environments. J Biomech 2012; 45(12): 2084–2091.

37.

Hof

AL.

The “extrapolated center of mass” concept suggests a simple control of balance in walking. Hum Mov Sci 2008; 27(1): 112–125.

38.

Lamoth

CJC

van Lummel

Beek

PJ.

Athletic skill level is reflected in body sway: a test case for accelometry in combination with stochastic dynamics. Gait Posture 2009; 29(4): 546–551.

39.

Callaghan

Gunning

McGill

SM.

The relationship between lumbar spine load and muscle activity during extensor exercises. Phys Ther 1998; 78(1): 8–18.

40.

Granata

Marras

WS.

The influence of trunk muscle coactivity on dynamic spinal loads. Spine 1995; 20(8): 913–919.

41.

Adams

Burton

Bogduk

. The biomechanics of back pain. Edinburgh: Churchill Livingstone, 2006, 331 pp.

42.

McGill

Low back disorders: evidence-based prevention and rehabilitation. Champaign, Ill.: Human Kinetics, 2007, 334 pp.

43.

White

Panjabi

Clinical biomechanics of the spine, vol. 2. Philedelphia, PA: Lippincott Williams & Wilkins, 1990, http://www.lww.com/Product/9780781717069 (accessed 27 January 2016).

44.

Ehde

Smith

Czerniecki

et al . Back pain as a secondary disability in persons with lower limb amputations. Arch Phys Med Rehabil 2001; 82(6): 731–734.

45.

McAndrew Young

Wilken

Dingwell

JB.

Dynamic margins of stability during human walking in destabilizing environments. J Biomech 2012; 45(6): 1053–1059.

46.

Clark-Carter

Heyes

Howarth

CI.

The efficiency and walking speed of visually impaired people. Ergonomics 1986; 29(6): 779–789.

47.

Dixon

Pearsall

DJ.

Gait dynamics on a cross-slope walking surface. J Appl Biomech 2010; 26(1): 17–25.

48.

Gates

Darter

Dingwell

et al . Comparison of walking overground and in a Computer Assisted Rehabilitation Environment (CAREN) in individuals with and without transtibial amputation. J NeuroEngineering Rehabil 2012; 9: 81.