Age-associated differences in the gait pattern changes of older adults during fast-speed and fatigue conditions: results from the Baltimore longitudinal study of ageing

Abstract

Objective: the present study investigated the effects of walking under different challenges and kinematics and kinetics generated during these activities and how these vary with age. We hypothesised that age-associated changes in gait speed and kinetics are more pronounced during fast-speed walking and post-activity walking, compared with usual-speed walking.

Methods: investigated walking under three conditions: (i) usual speed, (ii) fast speed and (iii) post-activity in 183 Baltimore Longitudinal Study of Aging participants (mean 73 ± 9 years) who could walk unassisted.

Results: across all tasks, gait speed decreased with older age and this decline rate was exacerbated in the fast-speed walking task, compared with usual-speed walking (P < 0.001). Medial–lateral (ML) hip-generative mechanical work expenditure declined with age and the rate of decline was steeper for walking at fast speed and post-activity during hip extension (P = 0.032 and 0.027, respectively), compared with usual-speed walking.

Conclusions: these findings indicate that older adults experience exacerbated declines in gait speed and ML control of the hip, which is explicitly evident during challenging walking. Exercise programmes aimed at improving gait speed and ML joint power from hip and ankle may help reverse age-associated changes in gait pattern among older adults.

Introduction

Customary gait speed in older adults is a good measure of overall walking performance. In general, it reflects energetic efficiency, muscle strength, balance control and endurance [1]. Self-chosen usual gait speed requires the selection of a certain stride length, joint angular displacement [2, 3] and the appropriate joint torque and power [4–6]. There is evidence that the selection of these elements is aimed at maximising energetic efficiency [7]. Changes in gait patterns with ageing have been described in previous studies and a decline in gait speed appears as one of the most consistent age-associated changes [5, 6]. Slower walking of older adults was related to fear of fall [8], muscle weakness [9] and impairment of motor control [10]. However, it is still not clear whether the speed decline with ageing is a compensatory effort to improve safety [6] or it is merely the reflection of deteriorated muscle activity performance. It is reasonable to hypothesise that age-related gait changes are even more evident during challenging walks, compared with self-selected usual walking. However, age-associated changes in gait during challenging conditions have been studied only seldom.

Reduced ankle joint power during fast-speed walking among older adults was reported [5, 11]. A slower gait speed among older adults with fatigue was also previously reported [12]. It is, nonetheless, not clear whether changes in gait performance are more strongly affected by ageing under challenging conditions, such as walking at a fast speed or in a post-activity state (i.e. a possible marker of fatigue) compared with customary, self-selected walking. It is likely that these challenging conditions place a greater demand on motor control and hence may be more sensitive to age-associated declines. For example, fast-speed walking likely requires relatively rapid generation of high peak power from the joints of lower extremities.

In the present study, gait parameters in the form of spatiotemporal measurement, range of rotation and mechanical work expenditure (MWE) were investigated to determine age-associated changes in gait during challenging walks. We assessed three-dimensional (3D) gait parameters of older adults during three different tasks, namely usual-speed walking, fast-speed walking and post-activity walking. We examined the a priori hypothesis that age-related changes in gait performance were more pronounced for fast-speed walking and post-activity walking, compared with usual-speed walking.

Methods

Participants

The data reported here were collected in 183 (age 60–96 years) participants of the Baltimore Longitudinal Study of Aging (BLSA; longitudinal cohort study conducted by the Intramural Research Program of the NIH, NIA) who were studied in our gait laboratory between January and July of 2008. All participants who did not have a hip or knee joint prosthesis, severe joint osteoarthritis, pain or history of stroke or Parkinson's disease and who could follow instructions and safely complete a 4 m walk unaided were included in this study. Participants who had a body mass index (BMI, kg/m²) over 40 were not included in this study because of technical difficulty of positioning the pelvic markers needed in gait analysis. The BLSA protocol was approved by the Medstar Research Institution Review Board. Participants were given a detailed description of the study and consented to participate.

Gait measurement

The protocol for gait analysis used in our laboratory has been described elsewhere [13]. Briefly, participants were instrumented with 20 reflective markers positioned according to anatomical landmarks including: anterior and posterior superior iliac spines, medial and lateral knees, medial and lateral ankles, toe (second metatarsal head), heel and lateral wands over the mid-femur and mid-tibia. A Vicon 3D motion capture system with 10-digital cameras (Vicon 612 system, Oxford Metrics Ltd, Oxford, UK) measured the movements of the segments for lower extremity, where the reflective markers were attached (60 Hz sampling frequency). During the walking task, ground reaction forces were measured with two sequentially staggered AMTI force platforms (Advanced Mechanical Technologies, Inc., Watertown, MA, USA; 1080 Hz sampling frequency). After all markers were positioned on the skin and non-reflective firm fitting spandex short pants, participants were asked to walk across a 10 m long laboratory walkway at their usual and fast speeds. The usual-speed walking task was based on a gait test at the self-preferred customary gait speed, whereas fast-speed walking task was based on a gait test where subjects were asked to walk as fast as possible without running. After approximately 30 min of additional tasks (e.g. 10 trials of fast walking, 4 trials of walking over obstacles, 8 trials of narrow-based walking; all of these are not included in the present study), participants were asked to perform a final walking task at their customary, self-selected gait speed. This task, referred in this report as the post-activity walking task, can be considered a functional measure of typical daily life walking that occurs after a fair amount of activity, reflecting a potential marker of fatigue; here, the amount of activity was standardised across all subjects, but the specific effects on muscle and aerobic function were not controlled. Participants were not informed about the presence and location of the force platforms on the walking path. Trials were performed until at least three gait cycles of the left and right sides with complete foot landing on the force platform were obtained from usual-speed, fast-speed and post-activity walking tasks. These three measurements were taken for guaranteeing the standardised assessment of gait characteristics with property of variability (we are not reporting variability in this paper) in BLSA database. The raw coordinate data of marker positions were digitally filtered with a fourth-order zero-lag Butterworth filter with a cutoff at 6 Hz.

Data process and statistical analysis

Three dimensional kinematic and kinetic gait parameters were measured and calculated as previously reported [13]. Dominant gait periods within the stance phase, which were used in this study as the temporal measurement and the duration for MWE calculation.

Statistical analyses were performed using SAS 9.1 Statistical Package (SAS Institute, Inc., Cary, NC, USA). Because all subjects did not complete all tasks, an unbalanced design in sample sizes was handled by generalised estimating equations (GEE) [14]. In the GEE method, all non-missing pairs of tasks were used for the estimations of missing task. An advantage of this approach is that it makes maximal use of the available data set. Nonetheless, we confirmed that similar results were obtained if only the sub-set of participants who had no missing task were analysed. Multilevel regression models were used to compare age-associated gait parameters for the fast-speed walking and post-activity walking to the usual-speed walking task. These models included appropriate interaction terms to test the hypotheses that the effect of age on the different parameters varied significantly across tasks. Regression models were adjusted for gait speed, sex, weight and height. Statistical significance was defined with a P-value less than 0.05.

Results

The 183 individuals (48% women) who participated in this study were aged 60–96, with an average BMI of 26.9 (±4.2 kg/m²). Spatiotemporal gait parameters and their associations with age in the three walking tasks are summarised in Table 1. With increasing age, gait speed and stride length were significantly lower, whereas cadence was significantly higher for all walking tasks (P < 0.001, for all). Stride width and stance percent of the gait cycle (PGC) were higher with older age in the usual-speed walking and post-activity walking, but not in the fast-speed walking. PGC for hip flexion and ankle plantar flexion were higher with older age for the usual-speed walking (P < 0.001 and P = 0.003, respectively).

Table 1.

Spatiotemporal gait measures during the usual-speed, fast-speed and post-activity walking tasks and their relationship with age

Spatiotemporal parameters	Regression analysis with advanced agea									Interactions of fast-speed and post-activity conditionb
	Usual-speed (n = 183)			Fast-speed (n = 162)			Post-activity (n = 147)			Fast-speed vs. usual-speed		Fatigue-condition vs. usual-speed
	Mean	βc	P-value	Mean	βc	P-value	Mean	βc	P-value	βc	P-value	βc	P-value
Speed and distances
Gait speed (m/s)	1.11	−0.014	<0.001	1.66	−0.024	<0.001	1.27	−0.015	<0.001	−0.01	<0.001d	−0.001	0.353
Stride length (m)	1.19	−0.014	<0.001	1.42	−0.016	<0.001	1.30	−0.014	<0.001	−0.003	0.003d	−0.000	0.456
Cadence (steps/min)	111.78	0.641	<0.001	140.20	0.741	<0.001	116.77	0.615	<0.001	0.101	0.287	−0.026	0.660
Stride width (cm)	11.08	0.078	0.007	10.35	0.012	0.668	10.06	0.050	0.049	−0.066	<0.001	−0.028	0.061
PGC
Stance	63.83	0.063	<0.001	62.33	0.016	0.188	63.02	0.044	<0.001	−0.048	<0.001	−0.019	0.048
Hip extension	51.02	0.027	0.277	48.45	0.033	0.256	49.99	0.051	0.043	0.06	0.769	0.024	0.131
Hip flexion	10.87	0.050	<0.001	12.18	0.023	0.232	10.95	0.016	0.258	−0.027	0.083	−0.035	<0.001
Knee extension	27.39	0.037	0.357	25.42	−0.025	0.378	26.79	0.007	0.825	−0.062	0.029	−0.030	0.167
Knee flexion	23.19	0.030	0.441	22.57	0.009	0.762	22.74	0.015	0.640	−0.021	0.486	−0.015	0.545
Ankle dorsiflexion	38.90	0.031	0.296	29.69	0.093	0.076	36.15	0.038	0.239	0.062	0.146	0.008	0.753
Ankle plantar flexion	18.93	0.083	0.003	27.28	−0.054	0.322	20.56	0.020	0.516	−0.137	0.002	−0.063	0.023

Bold indicates significance with P < 0.05.

^aAdjusted for gait speed, height, weight and sex (gait speed: adjusted for height, weight and sex).

^bGait parameter differences in walking tasks at fast-speed and post-activity condition compared with usual-speed walking.

^cEstimated coefficient for linear regression.

^dExacerbated decline compared with the usual-speed walking (only for the case both compared walking tasks already showed negative associations with older age).

Range of rotation and MWE for the three walking tasks from the hip, knee and ankle in the anterior–posterior (AP) plane and medial–lateral (ML) plane are summarised in Table 2. The range of rotations for the hip, knee and ankle in the AP plane were lower with older age in all walking tasks. With older age, MWE during hip extension in the AP plane was higher in the usual-speed walking (P = 0.036), while MWE during hip flexion in the AP plane was higher in the usual-speed and post-activity walking tasks (P < 0.001 and P = 0.002). AP ankle-generated MWE during ankle plantar flexion in late stance was lower with older age only in the fast-speed walking (P = 0.043). AP hip absorption MWE during hip extension was lower with older age in the usual-speed walking (P < 0.001) and post-activity walking (P = 0.022).

Table 2.

Age-associated changes in gait parameters for the rotations of lower extremities in the usual-speed, fast-speed and post-activity condition

Gait parameters in kinematics and kinetics	Regression analysis with advanced age									Interactions of fast-speed and post-activity conditiona
	Usual-speed (n = 183)			Fast-speed (n = 162)			Post-activity (n = 147)			Fast speed versus usual speed		Post-activity versus usual-speed
	Mean	βb	P-value	Mean	βb	P-value	Meanc	βb	P-value	βb	P-value	βb	P-value
AP side
Range of motion (°)
Hip	39.67	−0.176	<0.001	45.39	−0.141	0.004	42.33	−0.173	<0.001	0.035	0.246	0.003	0.829
Knee	53.68	−0.254	<0.001	56.77	−0.205	<0.001	55.39	−0.248	<0.001	0.050	0.050	0.006	0.716
Ankle	23.49	−0.138	<0.001	24.19	−0.160	<0.001	24.56	−0.144	<0.001	−0.022	0.373	−0.005	0.766
Generated MWE, 1,000 J/kg for mean in
Hip extension	103.49	1.077	0.036	156.32	0.189	0.799	107.27	0.960	0.091	−0.889	0.159	−0.117	0.805
Hip flexion	90.67	1.063	<0.001	177.51	0.431	0.236	106.24	0.827	0.002	−0.632	0.048	−0.236	0.174
Knee extension	71.17	−0.119	0.769	110.17	−0.080	0.911	89.74	−0.139	0.770	0.039	0.948	−0.020	0.956
Knee flexion	7.81	−0.019	0.869	7.78	−0.133	0.347	7.45	0.004	0.974	−0.113	0.288	0.024	0.800
Ankle dorsiflexion	7.69	−0.015	0.833	10.05	0.013	0.908	10.82	0.141	0.161	0.028	0.812	0.156	0.126
Ankle plantar flexion	156.72	−0.144	0.750	219.89	−1.377	0.043	173.80	−0.856	0.115	−1.233	0.015	−0.712	0.074
Absorption MWE, 1,000 J/kg for mean in
Hip extension	235.58	−2.629	<0.001	309.39	−2.000	0.083	276.36	−2.109	0.022	0.628	0.534	0.519	0.486
Hip flexion	1.75	0.058	0.127	2.91	−0.041	0.158	1.65	−0.031	0.095	−0.099	0.094	−0.090	0.078
Knee extension	11.44	−0.172	0.117	15.61	−0.118	0.426	13.16	−0.038	0.731	0.054	0.666	0.134	0.160
Knee flexion	178.39	−0.299	0.629	243.39	0.096	0.908	199.78	−0.271	0.735	0.395	0.574	0.028	0.962
Ankle dorsiflexion	131.91	0.019	0.987	105.23	−0.135	0.820	117.57	−0.046	0.922	−0.154	0.773	−0.064	0.886
Ankle plantar flexion	5.52	0.043	0.372	5.64	−0.037	0.246	5.18	−0.010	0.757	−0.080	0.120	−0.054	0.265
ML side
Range of motion (°)
Hip	9.80	−0.064	<0.001	12.06	−0.094	<0.001	10.88	−0.082	<0.001	−0.030	0.054	−0.018	0.122
Knee	10.53	−0.072	0.031	10.62	−0.033	0.362	10.72	−0.064	0.068	0.040	0.089	0.008	0.591
Ankle	8.42	−0.095	<0.001	8.70	−0.093	<0.001	9.04	−0.097	<0.001	0.001	0.939	−0.002	0.896
Generated MWE, 1,000 J/kg for mean in
Hip extension	62.30	−1.079	<0.001	70.58	−1.434	<0.001	68.89	−1.370	<0.001	−0.356	0.032d	−0.291	0.027d
Hip flexion	14.64	−0.361	<0.001	9.54	−0.353	<0.001	15.65	−0.537	<0.001	0.008	0.879	−0.177	0.003d
Knee extension	6.57	−0.001	0.989	9.66	0.007	0.945	8.01	−0.026	0.693	0.008	0.901	−0.026	0.407
Knee flexion	3.33	−0.033	0.229	4.00	−0.116	<0.001	3.61	−0.069	0.017	−0.083	<0.001	−0.037	0.024
Ankle dorsiflexion	5.24	−0.029	0.364	5.86	−0.080	0.197	6.20	−0.006	0.896	−0.051	0.425	0.023	0.599
Ankle plantar flexion	4.45	0.061	0.009	3.95	−0.005	0.848	3.42	0.007	0.787	−0.066	0.020	−0.054	0.030
Absorption MWE, 1,000 J/kg for mean in
Hip extension	44.75	0.305	0.125	64.88	0.068	0.820	49.52	0.211	0.355	−0.238	0.266	−0.094	0.463
Hip flexion	2.47	0.122	<0.001	7.70	0.043	0.307	3.26	0.127	<0.001	−0.080	0.015	0.005	0.760
Knee extension	7.98	0.023	0.630	8.35	−0.061	0.311	7.79	−0.011	0.822	−0.085	0.024	−0.035	0.181
Knee flexion	7.81	−0.031	0.669	6.67	0.019	0.795	7.31	−0.030	0.686	0.050	0.425	0.001	0.980
Ankle dorsiflexion	8.81	0.120	0.340	8.46	0.021	0.742	8.20	−0.106	0.045	−0.099	0.446	−0.226	0.082
Ankle plantar flexion	7.13	−0.145	<0.001	10.91	−0.292	<0.001	8.63	−0.183	<0.001	−0.147	0.004d	−0.037	0.362

Bold indicates significance with P < 0.05.

^aGait parameter differences in walking tasks at fast-speed and post-activity condition compared with usual-speed walking.

^bEstimated coefficient for linear regression.

^cAdjusted for gait speed, height, weight and sex (gait speed: adjusted for height, weight and sex).

^dExacerbated decline compared with the usual-speed walking (only for the case both compared walking tasks showed negative associations with older age).

The range of rotation for the hip and ankle in the ML plane and ML hip-generated MWE during hip extension and flexion were lower with older age for all three walking tasks (P < 0.001). ML knee-generated MWE during knee flexion was lower with older age in the fast-speed walking (P < 0.001). ML ankle-generated MWE during ankle plantar flexion for the usual-speed walking was higher with older age (P = 0.009). ML hip absorption MWE during hip flexion for the usual-speed walking and post-activity walking was higher with older age (P < 0.001). ML ankle absorption MWE during ankle plantar flexion was lower with older age for all three walking tasks (P < 0.001).

Figure 1 illustrates the exacerbated declines in gait speed and ML MWEs for the fast-speed walking and post-activity walking, compared with usual-speed walking. In the fast-speed walking, age-associated declines in gait speed, stride length, ML hip-generated MWE during hip extension and ML ankle absorption MWE during ankle plantar flexion were exacerbated (P < 0.001, P = 0.003, 0.032 and 0.004, respectively) compared with usual-speed walking. Similarly, in the post-activity walking, age-associated declines in the ML hip-generated MWE during hip extension and flexion were exacerbated (P = 0.027, 0.003, respectively) compared with usual-speed walking. We also repeated the analyses on the subgroup of subjects who completed all three walking tasks (N = 141). The results were essentially identical (data not shown).

Figure 1.

Exacerbated declines in gait speed (a), ML ankle MWE in plantar flexion (b), ML hip MWE in extension (c) and ML hip MWE in flexion (d) during challenging walking tasks.

Discussion

Consistent with previous studies, slower gait speed and shorter stride length with older age were confirmed in this study in a usual-speed walking task as well as in fast-speed walking and post-activity walking [5, 15, 16]. Wider stride width with older age, already reported from previous studies [17, 18], was also confirmed in the present study in the usual-speed walking and post-activity walking tasks. Consistent with our hypotheses, we observed that the declines in gait speed and ML MWEs with older age were significantly sharper for the challenging walks (i.e. fast-speed walking and post-activity walking), compared with self-selected usual-speed walking, supporting the idea that age-associated changes in gait are especially more evident during these challenging walking conditions.

In the AP plane, all three walking tasks revealed similar age-associated patterns in the range of rotations, whereas age-associated changes in mechanical energy usages had different patterns by walking tasks (Table 2). Higher AP hip-generated MWEs during hip flexion with older age in the usual-speed walking and post-activity walking are consistent with a previously reported study [19] and can be thought of as a compensatory effort to maintain the same gait speed in spite of significantly reduced ranges in the rotational motion. Lower AP ankle-generated MWE during ankle plantar flexion of late stance for the fast-speed walking with older age may represent reduced ankle propulsion before swing that can cause shorter swing or lack of foot clearance. This result coincides with the previous reports examined by inverse dynamics [20] and electromyography (EMG) [21]. This also suggests that older adults may have difficulty in fast-speed walking because of reduced mechanical energy from ankle joint. Reduced AP hip absorption MWE during hip extension with older age, observed in the usual-speed walking and post-activity walking, may represent a substantial functional loss because this occurs during a critical weight supporting period.

Lower range of rotations from the hip and ankle joints in the ML plane for all walking tasks are likely one of the reasons for the lower ML-generated MWEs in the hip and ankle with older age. The negative associations with age of ML hip-generated MWEs during hip extension and hip flexion for all three walking tasks are probably related to lower angular speed and strength in the hip abductor muscle group. Such an age-associated decline of hip generated MWE in the ML plane may be one of the keys to mobility performance decline in older adults. Hip abductor contraction during the stance period is essential for maintaining ML stability and a robust support from the ipsilateral leg movement. At the same time, hip-generated MWE in the ML plane is important to allow enough clearance space for the contralateral foot during swing [22, 23]. In general, ML hip-generated MWE and AP hip-generated MWE are generated from hip abductors and hip flexors, respectively, and those alignments of muscle groups are relatively aligned against each other for their movements in the ML plane. Therefore, positively age-associated AP hip-generated MWE during hip flexion, which was observed in the present study in the usual-speed and post-activity walking tasks, might have affected the ML hip-generated MWE during hip flexion, which was lower with older age.

The fast-speed and post-activity conditions revealed exacerbated declines in gait performance with older age, compared with usual-speed walking task. Declines in the gait speed and ML MWE from the hip and ankle joints can be interpreted as decreased potential capacity to manoeuvre and control ML locomotion in older adults, respectively. Thus, these findings can be thought of as risk factors for the mobility limitations, falls or injuries among older adults. This hypothesis should be tested in longitudinal studies. The comparisons between different walking tasks in this study revealed elements of gait that are influenced by age under usual and different walking conditions. Causal pathways between the characteristics identified here and longitudinal age association with mobility limitation remain to be established. However, if the hypothesis raised by this study is confirmed, gait tests that focus on different walking conditions can be developed and translated into clinical practice. In addition, specific exercise programmes aimed at improving gait speed for the fast-speed walking or enhancing ML controllability from the hip and ankle joint could be developed with the aim of improving function and energetics among older adults.

This study has limitations. We did not measure the metabolic energy consumption during gait analysis in the BLSA participants, so information about efficiency in energy utilisation of older adults cannot be assessed. Also, because gait speed was observed to be faster for the post-activity walking task, when compared with usual-speed walking task, the 30 min of interim testing may not have been adequate to cause a condition of ‘fatigue’ in our participants. Repeated walking tests before the post-activity walking task may have caused different changes in other factors such as apprehension, learning (or familiarity) and motivation in older adults. To more completely describe the observed age-associated effects of fatigue, it might be helpful to also investigate these associations after controlling for changes in muscle properties as measured by EMG or maximum voluntary contraction. Still, the exacerbated age-related declines observed in the post-activity state suggest that some element in this state likely did play a role, perhaps reflecting fatigue. Finally, the cross-sectional nature of these data allows for the study of correlation but not causality. The BLSA is currently collecting longitudinal data that are designed to overcome this limitation.

Key points.

• Age-associated gait pattern change.

• Exacerbated gait speed decline.

• Decline in ML control.

Conflicts of interest

All the authors declare that non financial and personal relationships were conducted with other people or organizations that could inappropriately influence or bias this work.

Funding

This research was entirely supported by the Intramural Research Program of the NIH, National Institute on Aging.