Dual-task standing and walking in people with lower limb amputation: A structured review

Abstract

Background:

People with lower limb amputation report the need to concentrate on walking. Dual-task paradigms are often used to study such interactions between cognition and the control of balance and gait in people with lower limb amputation.

Objectives:

To summarize evidence related to dual-task standing and walking in people with lower limb amputation and discuss implications for future research.

Study design:

Structured review.

Methods:

A structured search was completed in PubMed, CINAHL, and Web of Science from database inception to May 2017. Eligible articles were in English, included participants with lower limb amputation, and assessed dual-task standing or walking. Study information was extracted by one reviewer and assessed for accuracy by a second.

Results:

A total of 12 articles met eligibility criteria. Seven examined differences in dual-task standing or walking, and five assessed dual-task walking across microprocessor-controlled and non-microprocessor-controlled prosthetic knee conditions.

Conclusion:

Results suggest that (1) dual-task interference in standing is greater for people with lower limb amputation than non-amputees and (2) the use of microprocessor-controlled knees improves dual-task performance for people with limited mobility compared to non-microprocessor-controlled knees. The small number of studies, diversity of dual-task methods, and methodological limitations challenge the synthesis of study results. Future research should assess specific clinical characteristics that can affect dual-task performance in people with lower limb amputation.

Clinical relevance

Understanding how people with lower limb amputation use increased attention in balance and gait to compensate for limb loss can inform prosthetic interventions and training. Dual-task research suggests that people with amputation use increased cognitive control for standing compared to controls and microprocessor-controlled knees improve dual-task walking in people with limited mobility.

Keywords

Prosthetics gait

Background

For people with lower limb amputation (LLA), loss of physical limb structures and peripheral motor and sensory systems contributes to profound mobility challenges. Mobility limitations commonly experienced by people with LLA include impaired postural stability,^1–3 increased energy expenditure while walking,^4–6 decreased gait speed,^5–7 and pronounced gait asymmetries.^7,8 Such limitations may result in difficulties negotiating uneven terrain,⁹ hills,¹⁰ and stairs.^11,12 It is also conceivable that these impairments contribute to restrictions in walking activity^13,14 and the high incidence of falls^15,16 reported by people with LLA.

In addition to these mobility limitations, many people with LLA report the need to “concentrate on every step.”^15,16 This perception may reflect increased reliance on cognitive resources to compensate for loss of peripheral afferent feedback and efferent control. Increased use of cognitive resources when standing and walking may limit resources available for concurrent cognitive tasks, such as social interaction or observing the environment. Furthermore, the need to concentrate on walking has been correlated with falls and fear of falling in people with LLA.¹⁷ Reducing the need to concentrate during these tasks may serve to both improve quality of life and promote safety in this clinical population. Microprocessor-controlled knees with stumble recovery have been noted to increase perceived confidence and reduce fear that the prosthetic knee will buckle,¹⁸ giving this technology the potential to reduce the need to concentrate on standing and walking with a prosthesis.

Dual-task methods assess simultaneous performance of a postural task and a cognitive or motor task to study how cognitive processes, such as concentration or executive function, interact with postural tasks, such as standing and walking.¹⁹ Declines in performance on one or both tasks (i.e. dual-task interference) can be used to infer relative contributions of cognitive resources to postural control.²⁰ Dual-task interference is often assessed in research using performance changes from single-task to dual-task conditions. Changes (and possible dual-task interference) can occur in the postural control task, the concurrent cognitive or motor task, or in both tasks.

An underpinning assumption of these experiments is that dual-task interference is caused by limited cognitive resources (i.e. capacity theory²¹) or processing (i.e. bottleneck theory²²). The capacity theory hypothesizes that, in circumstances when simultaneous performance of one or more tasks requires cognitive processing that exceeds the available capacity, performance quality of one or more tasks declines.²¹ The bottleneck theory postulates that simultaneous performance of multiple tasks can lead to processing interference where the neural system prioritizes one of the tasks and temporarily postpones action on the other, leading to performance decline on the non-prioritized task.²² Both theories assert that when increased cognitive resources or processes are required for performance of a task, remaining resources or processing channels available for the performance of a concurrent task may be insufficient and result in decline in performance on one or both tasks.

Thus, people who are required to allocate additional cognitive resources to postural tasks will experience greater dual-task interference than those for whom postural tasks are relatively automatic. However, methodological choices, such as selection of postural and cognitive tasks, use of directions for task prioritization, and inclusion of baseline (or single-task) conditions, can profoundly affect investigators’ ability to appropriately and meaningfully interpret results from dual-task studies.^20,23,24

Because dual-task paradigms are a non-invasive means to infer the extent to which cognition is employed during standing and walking, investigators^{1,2,9,25–33} have used these performance-based methods to objectively assess the perceived need to concentrate on walking in people with LLA. However, results from performance-based dual-task studies of standing and walking often do not support the hypothesis that people with LLA have increased dual-task interference compared to people without LLA. Such studies are inconsistent with survey-based studies^15–17 that note a large percentage of people with LLA report using heightened concentration for walking. A structured literature review was therefore conducted to assess research examining dual-task standing and walking in people with LLA and to offer potential explanations for discrepancies between performance-based and self-reported study results. The objectives of this review were (1) to identify, appraise, and synthesize original research that used dual-task methods to study standing and walking tasks in people with LLA and (2) to propose recommendations for future research.

Methods

Data sources and eligibility criteria

A comprehensive search of the literature was performed from database inception (earliest date: 1965) through May 2017 using three electronic databases: PubMed (1966–05/2017), CINAHL (1982–05/2017), and Web of Science (1965–05/2017). Search strategies for the three databases are provided (Table 1). References of selected articles were also searched to identify additional articles.

Table 1.

Search strategies for databases.

Search	Field Code	Search term(s)
PubMed
1	MeSH Terms	“Amputation”
2	MeSH Terms	“Amputees”
3	All fields	amput*
4	MeSH Terms	“Artificial Limbs”
5	All fields	prosthe* OR “microprocessor knee” OR “microprocessor-controlled” OR “C-leg”
6	Title/Abstract	cochlear OR “hearing loss” OR teeth OR dentures OR arthro* OR “hip replacement” OR fracture OR exoskeleton OR orthosis OR orthotic
7		(1 OR 2 OR 3 OR 4 OR 5) NOT 6
8	All fields	balance OR walking OR postur* OR gait
9	All fields	attention* OR cognitive OR dual-task OR “dual task” OR multitask* OR “reaction time” OR (mental AND load*)
10		6 AND 7 AND 8
11		Limit 9 to English language (9 AND language = English)
CINAHL
1	AB	amput* OR (artificial AND limb) OR prosthe OR “microprocessor knee” OR “microprocessor-controlled” OR “C-leg”
2	AB	cochlear OR “hearing loss” OR teeth OR dentures OR arthro* OR “hip replacement” OR fracture OR exoskeleton
3		1 NOT 2
4	AB	balance OR walking OR postur* OR gait
5	AB	attention* OR cognitive OR dual-task OR “dual task” OR multitask* OR “reaction time” OR “mental load*”
6		3 AND 4 AND 5
7		Limit 6 to English language
8		Limit 7 to academic journals
Web of Science
1	Topic	amput* OR (artificial AND limb) OR prosthe OR “microprocessor knee” OR “microprocessor-controlled” OR “C-leg”
2	Topic	cochlear OR “hearing loss” OR teeth OR dentures OR arthro* OR “hip replacement” OR fracture OR exoskeleton
3		1 NOT 2
4	Topic	balance OR walking OR postur* OR gait
5	Topic	attention* OR cognitive OR dual-task OR “dual task” OR multitask* OR “reaction time” OR “mental load*”
6		3 AND 4 AND 5
7		Limit 6 to English language

Criteria for inclusion in this review were (1) published in English; (2) adult participants with LLA of any etiology; (3) participants currently use a prosthesis; and (4) the primary or secondary outcome measures assessed dual-task performance during standing or walking tasks with a concurrent cognitive or motor task. Articles were excluded if they were an abstract, thesis, dissertation, or review.

Article selection and data collection process

Two reviewers evaluated titles and abstracts to determine the eligibility of articles identified by the literature search. Potentially eligible articles were obtained and assessed using the criteria stated above. Pertinent information, including study and sample characteristics, postural and cognitive task characteristics, and dual-task related findings, was extracted into evidence tables by a single reviewer and corroborated by a second (Tables 2 –4). Study quality was classified using the Physiotherapy Evidence Database (PEDro) Scale, which consists of 11 assessment criteria to determine the validity of study methods (Table 5). Meta-analysis of results was not possible because of the variety of study designs, methodologies, and outcomes measured.

Table 2.

Study and sample characteristics for studies of dual-task performance in people with LLA.

Author	Study design	n	Age (SD)	Sex	Amputation level	Etiology	Control/Comparison characteristics
Dual-task standing
Geurts et al.²⁵	Observational cohort	8	67.7 (18.1)	F: 3M: 5	TF: 2KD: 2TT: 4	Vascular: 5Infection: 2Trauma: 1	10 age- and sex-matched
Geurts and Mulder²	Observational cohort	12	59.4 (18.3)	F: 3M: 9	TF: 3KD: 5TT: 4	Vascular: 9Other: 3	12 age- and sex-matched8 non-matched
Vrieling et al.¹	Observational cohort	8	51.8 (12.7)	F: 2M: 6	TF: 3TT: 5	Trauma: 5Tumor: 2Vascular: 1	9 non-matched
Dual-task walking
Nakamura et al.³²	Cross-sectional	11	Not reported	Not reported	HD: 1TF: 5TT:5	Not reported	9 non-matched
Hof et al.²⁹	Cross-sectional	6	40.5 (6.6)	F: 2M: 4	TF	Not reported	6 leg-length- and sex- matched
Lamoth et al.⁹	Cross-sectional	8	43.8 (14.8)	F: 3M: 5	TF	Trauma: 5Tumor: 2Vascular: 1	8 age- and sex-matched
Morgan et al.³¹	Cross-sectional	14	53.8 (13.6)	F: 5M: 9	TF	Trauma: 8Tumor: 3Vascular: 1Infection: 2	14 age- and sex-matched
Microprocessor-controlled knees compared to non-microprocessor-controlled knees
Heller et al.²⁸	Cross-over	10	38 (not reported)	F: 3M: 7	TF	Trauma: 8Tumor: 2	MPK: Intelligent ProsthesisNMPK: Endolite Stabilized Knee
Williams et al.³³	Cross-over	8	48.5 (10.2)	F: 1M: 7	TF	Not reported	MPK: C-LegNMPK: Mauch SNS
Hafner et al.²⁷	Cross-over	17	48.4 (17.2)	F: 4M: 13	TF	Trauma: 10Tumor: 3Infection: 2Vascular: 1Other: 1	MPK: C-LegNMPK: current prosthesis
Hafner and Smith²⁶	Cross-over	17	48.4 (17.2)	F: 4M: 13	TF	Trauma: 10Tumor: 3Infection: 2Vascular: 1Other: 1	MPK: C-LegNMPK: current prosthesis
Meier et al.³⁰	Cross-over	12	46 (8.6)	F:2M: 10	TF	Trauma: 7Infection: 2Congenital: 2Vascular: 1	MPK: C-LegNMPK: Otto Bock 3R60, Mauch SNS

LLA: lower limb amputation; SD: standard deviation; F: female; M: male; HD: hip disarticulation; TF: transfemoral; KD: knee disarticulation; TT: transtibial; SNS: stance and swing; MPK: microprocessor-controlled prosthetic knee; NMPK: non-microprocessor-controlled prosthetic knee.

Table 3.

Postural and cognitive task characteristics for articles describing dual-task studies in people with LLA.

Author	Postural		Cognitive		Single-task assessment	Instructed focus
Author	Task	Outcome measure	Task	Outcome measure	Single-task assessment	Instructed focus
Dual-task standing
Geurts et al.²⁵	Quiet standing	Postural stability: root mean square AP and ML center of pressure velocity	Discrimination and decision-making: visual Stroop (n = 5)Mental tracking: serial subtractions (n = 3)	Speed (number of colors/subtractions in 15 s) and number of errors	Cognitive and postural	Postural task
Geurts and Mulder²	Quiet standing; weight shifting	Postural stability: root mean square AP and ML center of pressure displacement and velocity	Mental tracking: assess equations	Number of errors	Cognitive and postural	Not reported
Vrieling et al.¹	Standing on a moving platform	Postural stability: AP ground reaction force and center of pressure displacementAsymmetry: weight-bearing index	Discrimination and decision-making: auditory Stroop	Not reported	Postural only	Not reported
Dual-task walking
Nakamura et al.³²	Walking, details not reported	Not reported	Reaction time: response to auditory stimuli	Reaction time	Cognitive only	Not reported
Hof et al.²⁹	Walking, treadmill, standardized speed	Pace: stride timePostural stability: stride width, margin of stability Asymmetry: percent stance and double-contact time asymmetry	Discrimination and decision-making: visual Stroop	Not reported	Postural only	Not reported
Lamoth et al.⁹	Walking, indoor, SSWS	Pace: speed, stride timePostural stability: stride time variability (COV) and trunk acceleration ML and AP root mean square, sample entropy, and local stability exponent	Mental tracking: serial subtractions	Number of subtractions, absolute number of errors, accuracy	Postural only	Not reported
Morgan et al.³¹	Walking, indoor, SSWS	Pace: speedPostural stability: step widthAsymmetry: step time asymmetry	Discrimination and decision-making: auditory Stroop	Accuracy and reaction time	Cognitive and postural	Cognitive Task
Microprocessor-controlled knees compared to non-microprocessor-controlled knees
Heller et al.²⁸	Walking, treadmill, standardized speeds that changed across the trial	Postural stability: velocity of head marker movement	Discrimination and decision-making: visual Stroop	Number of errors	Postural only	Not reported
Williams et al.³³	Walking, indoor, SSWS, and controlled walking speed	Pace: speed	Verbal fluency: Controlled Oral Word Association Test, Category TestMental tracking: serial subtractions	Verbal fluency: total number of unique wordsMental tracking: number of errors	Not reported	Not reported
Hafner et al.²⁷	Walking, outdoor, two city blocks	Pace: speed	Mental tracking: verbal reverse numbers	Accuracy	Not reported	Not reported
Hafner and Smith²⁶	Walking, outdoor, two city blocks	Pace: speed	Mental tracking: verbal reverse numbers	Accuracy	Not reported	Not reported
Meier et al.³⁰	Walking, indoor obstacle course, SSWS	Pace: time to complete obstacle courseMovement efficiency: total heart beat index	Mental tracking: serial subtractions	Not reported	Postural only	Not reported

AP: anterior–posterior; COV: coefficient of variation; LLA: lower limb amputation; ML: mediolateral; SSWS: self-selected walking speed.

Table 4.

Dual-task related findings for articles describing dual-task studies in people with LLA.

Author	Postural task results	Cognitive task results	Self-report results
Dual-task standing
Geurts et al.²⁵	Postural stability: people with LLA had greater dual-task interference in root mean square of ML center of pressure velocity than controls at the start and end of rehabilitation	No differences in dual-task interference between groups	Not reported
Geurts and Mulder²	Postural stability in quiet standing: people with LLA had greater dual-task interference in root mean square of AP and ML center of pressure velocity than matched and non-matched controls. Dual-task interference in AP root mean square center of pressure velocity improved with rehabilitation	Quiet standing: people with LLA had greater dual-task interference in number of errors than non-matched controls	Not reported
Geurts and Mulder²	Postural stability in perturbed standing: no differences in dual-task interference between groups	Perturbed standing: no differences in dual-task interference between groups	Not reported
Vrieling et al.¹	Postural stability: no differences in dual-task interference between groupsAsymmetry: No differences in dual-task interference between groups	Not reported	Not reported
Dual-task walking
Nakamura et al.³²	Not reported	Reaction time: no differences in dual-task interference between groups	Not reported
Hof et al.²⁹	Pace: no differences in dual-task interference between groupsPostural stability: no differences in dual-task interference between groupsAsymmetry: no differences in dual-task interference between groups	Not reported	Not reported
Lamoth et al.⁹	Pace: no differences in dual-task interference between groupsPostural stability: controls had AP trunk acceleration sample entropy dual-task interference; people with LLA did not. No other differences in dual-task interference between groups	No differences in dual-task interference between groups	Not reported
Morgan et al.³¹	Pace: no differences in dual-task interference between groupsPostural stability: no differences in dual-task interference between groupsAsymmetry: no differences in dual-task interference between groups	No differences in dual-task interference between groups	All participants with LLA reported heightened need to pay attention to walking. Activities that require attention: icy/slippery surfaces (100%), uneven terrain (92.9%), down stairs (85.7%), and sand/snow (85.7%)
Microprocessor-controlled knees compared to non-microprocessor-controlled knees
Heller et al.²⁸	Postural stability: no differences in dual-task interference between knee conditions	No differences in dual-task interference between groups	Not reported
Williams et al.³³	Pace: no differences in walking speed between knee conditions	No differences in cognitive performance between knee conditions	Less self-reported attention required for walking in microprocessor-controlled knees, microprocessor-controlled knees reported to be less of a cognitive burden
Hafner et al.²⁷	Pace: no differences in dual-task walking speed between knee conditions	No differences in % accuracy between knee conditions	Improvement in self-report difficulty to “complete a task while walking” for microprocessor-controlled knees
Hafner and Smith²⁶	Pace: improvement in dual-task walking speed in microprocessor-controlled knees compared to non-microprocessor-controlled knees for K2 group only	No differences in % accuracy between knee conditions in either group (K2 or K3)	Improvement in self-report “mental energy expenditure” for microprocessor-controlled knees for K3 group, improvement in self-report difficulty to multitask while walking for microprocessor-controlled knees in both K2 and K3 groups
Meier et al.³⁰	Pace: no differences in dual-task interference between knee conditionsMovement efficiency: no differences in dual-task interference between knee conditions	Not reported	Not reported

AP: anterior–posterior; K2: Medicare Classification Level 2 (limited community ambulator); K3: Medicare Classification Level 3 (unlimited community ambulator); LLA: lower limb amputation; ML: mediolateral.

Table 5.

Quality assessment of reviewed articles.

Author	1	2	3	4	5	6	7	8	9	10	11
Dual-task standing
Geurts et al.²⁵	No	No	N/A	Yes	No	N/A	No	No	N/A	Yes	Yes
Geurts and Mulder²	No	No	N/A	Yes	No	N/A	No	Yes	N/A	Yes	Yes
Vrieling et al.¹	Yes	No	N/A	Yes	No	N/A	No	Yes	N/A	Yes	Yes
Dual-task walking
Nakamura et al.³²	No	No	N/A	No	No	N/A	No	Yes	N/A	No	Yes
Hof et al.²⁹	No	No	N/A	Yes	No	N/A	No	Yes	N/A	Yes	Yes
Lamoth et al.⁹	No	No	N/A	Yes	No	N/A	No	Yes	N/A	Yes	Yes
Morgan et al.³¹	Yes	No	N/A	Yes	No	N/A	No	Yes	N/A	Yes	Yes
Microprocessor-controlled knees compared to non-microprocessor-controlled knees
Heller et al.²⁸	Yes	No	N/A	Yes	No	No	No	Yes	Yes	Yes	Yes
Williams et al.³³	Yes	Yes	No	Yes	No	No	No	No	Yes	Yes	Yes
Hafner et al.²⁷	Yes	No	N/A	Yes	No	No	No	No	Yes	Yes	Yes
Hafner and Smith²⁶	Yes	No	N/A	Yes	No	No	No	No	Yes	Yes	Yes
Meier et al.³⁰	Yes	Yes	No	Yes	No	No	No	No	Yes	Yes	Yes

Criteria descriptions: (1) eligibility criteria were specified; (2) subjects were randomly allocated to groups or order; (3) allocation was concealed; (4) the groups were similar at baseline; (5) there was blinding of all subjects; (6) there was blinding of all practitioners who administered the intervention; (7) there was blinding of all assessors who measured at least one key outcome; (8) measures of at least one key outcome were obtained from more than 85% of the subjects initially allocated to groups; (9) all subjects for whom outcome measures were available received the treatment or control condition as allocated; (10) the results of between-group statistical comparisons are reported for at least one key outcome; and (11) the study provides both point measures and measures of variability for at least one key outcome.

Results

In total, 12 articles^{1,2,9,25–33} met the stated eligibility criteria (Figure 1). Results from these 12 articles (11 studies) were grouped into three categories: dual-task standing, dual-task walking, and effects of microprocessor-controlled prostheses on dual-task walking (Tables 2 –4). Seven studies compared dual-task standing and walking performance in people with and without LLA to test the hypothesis that people with LLA have greater dual-task interference (i.e. diminished performance on the postural and/or cognitive task in dual-task compared to single-task conditions) than people without LLA. Four studies (five articles) compared dual-task performance in people with LLA using microprocessor-controlled and non-microprocessor-controlled knees to test the hypothesis that microprocessor-controlled knees mitigate dual-task interference to a greater extent than non-microprocessor-controlled knees.

Figure 1.

Schematic of search strategy and results. Search was performed on 5/30/2017.

Dual-task standing in people with LLA compared to controls

Study and sample characteristics

Three observational studies^1,2,25 examined the effects of a concurrent cognitive task on postural control during quiet and perturbed standing tasks in a total of 28 people with LLA. Two^2,25 examined longitudinal changes in dual-task standing throughout a period of post-amputation rehabilitation and one¹ assessed cross-sectional differences in standing under single- and dual-task conditions between people with and without LLA. Sample characteristics varied greatly both within and across the studies. All three included small samples of 8–12 people with LLA. Samples in each study varied in age, amputation level, time since amputation, prosthetic componentry, and etiology of amputation. One study²⁵ included an age- and sex-matched control group, one included a non-matched control group,¹ and one included both matched and non-matched controls.² While participants’ cognitive status was not reported in any of the studies, individuals with significant memory and attention deficits were excluded from two studies.^2,25 Notably, investigators did not describe how memory and attention were assessed nor did they describe the number of individuals who were excluded based on these criteria.

Methods

Dual-task standing studies varied in terms of the postural control and cognitive tasks used in each. Standing postural control tasks included quiet standing on a stable surface,^2,25 weight shifts in response to visual targets,² and standing on a surface that moved in a rhythmic, predictable manner.¹ Standing task performance was quantified using measures of postural sway (e.g. root mean square of the center of pressure displacement and velocity)^1,2,25 and weight-bearing symmetry.¹ Cognitive tasks were either mental tracking tasks (e.g. mathematical functions^2,25) or discrimination and decision-making tasks (e.g. choice reaction time tests using visual²⁵ and acoustic¹ stimuli). One study²⁵ used two cognitive tasks—the majority of participants were administered the visual Stroop, a discrimination and decision-making task, but participants who were unable to discern colors were tested with serial subtraction, a mental tracking task. All three studies assessed standing postural control under both single- and dual-task conditions. Two^2,25 assessed cognitive performance under both single- and dual-task conditions and one¹ did not report cognitive task performance results. None of these studies clearly stated whether instructions to focus on the cognitive task or on the postural task were provided.

Results

Two studies^2,25 investigated the effects of a concurrent cognitive task on postural sway in quiet standing in people with LLA and controls. Both studies described significantly greater dual-task relative to single-task sway in people with LLA at the start of rehabilitation compared to controls and found that people with LLA showed significant improvements in postural sway in either the sagittal^2,25 or coronal²⁵ direction in the dual-task condition over a period of rehabilitation. However, significant changes in dual-task interference (defined as dual-task minus single-task) in coronal postural sway were only reported in one study.²⁵

Two studies^1,2 investigated the effects of a concurrent cognitive task on postural control in complex standing conditions: visually guided weight shifting² and standing on a rhythmically moving surface.¹ Neither reported statistically significant differences in single-task relative to dual-task standing between people with LLA and controls. One² described fewer weight shifts and more cognitive task errors in people with LLA compared to non-matched controls. The other¹ did not report significant group differences in dual-task relative to single-task postural control performance when standing on a moving platform.

All three studies reported that standing postural control was significantly worse in people with LLA, regardless of single-task or dual-task condition. No significant between- or within-group differences²⁵ or group-by-task interactions² were observed for cognitive task performance.

Dual-task walking in people with LLA compared to controls

Study and sample characteristics

Four cross-sectional, observational studies^9,29,31,32 examined dual-task walking in people with LLA compared to controls. Samples were small, with a total of 39 participants with LLA across all studies. Participant characteristics, such as amputation level, differed between studies. One study³² included participants with LLA at the hip, transfemoral, and transtibial levels, along with non-matched controls. The other studies^9,29,31 included participants with transfemoral amputation and controls matched by sex and either age^9,31 or leg length.²⁹ Participants in all four studies varied with respect to time since amputation and age. The majority of participants had LLA secondary to trauma, however, etiology was not reported in two studies.^29,32 Participants with LLA had statistically lower Montreal Cognitive Assessment (MoCA) scores than controls in one study.³¹ Assessment of cognitive function was not described in the other studies.^9,29,32

Methods

The four dual-task walking studies varied in their methodological approaches. One³² described the effects of walking on cognitive task performance (i.e. reaction time) to characterize the relative cognitive demand of walking in people with and without LLA. Unlike other studies in this review, the investigators did not quantify postural or walking task performance. In addition, they did not conduct statistical analyses to compare dual-task outcomes between participants with LLA and controls. The other three walking studies^9,29,31 investigated the effects of a concurrent cognitive task on walking performance (e.g. speed, gait variability). The choice of cognitive tasks varied across studies and included a mental tracking task (serial subtraction⁹) and discrimination and decision-making tasks (visual²⁹ and auditory³¹ versions of the Stroop test). Walking tasks were performed at self-selected speeds either on a level surface^9,31 or on a treadmill.²⁹ Each study included three or more walking outcomes collected with various equipment such as a triaxial ambulant accelerometer,⁹ an instrumented treadmill,²⁹ and a three-dimensional (3D) motion capture system.³¹ Walking outcomes differed across studies and included measures of pace (i.e. walking speed^9,31 and stride time^9,31), asymmetry (i.e. percent time spent in stance phase and double-contact²⁹ and step time asymmetry³¹), and postural stability (i.e. step or stride width,^29,31 stride time coefficient of variation,⁹ margin of stability,²⁹ mediolateral and anterior–posterior trunk acceleration,⁹ local stability exponent,⁹ and sample entropy⁹). Three^9,29,32 of the four studies did not describe directing participants to focus on either the walking task or the cognitive task in dual-task conditions, leaving open the possibility that participants prioritized different tasks.

Results

Walking study results were challenging to synthesize due to differences in study objectives and/or outcomes of interest. One study³² that described the effect of walking on cognitive task performance reported that both people with LLA and controls showed a significant increase in reaction times when walking compared to standing. Statistical differences in the relative cognitive demand associated with walking between people with LLA and controls were not described in this study, despite inclusion of a non-matched control group. The investigators also reported reaction time in people with LLA by level of amputation. They found that reaction times were generally longer in people with higher levels of amputation in standing and all phases of stepping. However, the investigators did not statistically compare reaction times by group and they noted that there was high variability in performance among participants with LLA.

Overall, results from the three other studies showed that a concurrent task did not affect walking in people with LLA more than in controls. Investigators noted differences in walking performance between people with and without amputation across pace, asymmetry, and postural stability outcomes, though, importantly, walking performance was not differentially affected by the addition of a concurrent task. In other words, walking performance was consistently worse in people with LLA in both single-task and dual-task conditions. When a statistically significant decline in walking performance between single-task and dual-task conditions was measured for people with LLA, a similar decline was noted in controls. For studies^9,31 that reported cognitive task outcomes, performance did not decline to a greater extent in people with LLA compared to controls, suggesting that dual-task interference was comparable between groups.

Self-report

One³¹ of the four walking studies assessed participants’ self-report of attentional demands while walking. All 14 of the participants reported the need to “pay attention” while walking. The most common situations that required attention were walking on icy or slippery surfaces, over uneven terrain, downstairs, in sand, or in snow. This study did not compare self-report data from participants with and without LLA.

Effects of microprocessor-controlled prostheses on dual-task walking in people with LLA

Study and sample characteristics

Four longitudinal intervention studies (described in five articles^{26–28,30,33}) compared the effects of microprocessor- and non-microprocessor-controlled knees on dual-task walking in people with transfemoral amputation. Each of these studies used within-participant designs, meaning that participants were exposed to all experimental conditions and served as their own controls. Three^28,30,33 studies used a before-and-after design, and the fourth study^26,27 (described in two articles) used a reversal design.

Sample sizes for these studies ranged from 8 to 17 participants, with a total of 47 participants across all studies (two articles^26,27 described the same sample of 17 people). Three^26,27,33 of the five articles reported attrition rates, which ranged from 19%^26,27 to 56%.³³ All samples were homogeneous with respect to level of amputation, with all participants having amputations at the knee disarticulation or transfemoral levels. Participants across studies were heterogeneous with respect to age and etiology, with the majority of participants experiencing amputation secondary to trauma (25 of 39 participants, 8 not reported).

Methods

Each study compared dual-task performance when participants wore microprocessor-controlled knees or non-microprocessor-controlled knees to determine if microprocessor-controlled prostheses improved dual-task walking performance in participants with transfemoral amputation (TFA). Three^26,27,30,33 of the four studies included the Otto Bock C-Leg (Ottobock, Duderstadt, Germany), a prosthetic knee with microprocessor control of both stance and swing phases of gait. The C-Leg uses sensors in the knee and pylon to inform rapid changes to the knee flexion and extension resistance.¹⁸ The remaining study²⁸ included the Intelligent Prosthesis, a single-axis hydraulic knee with microprocessor control of swing phase only. As such, the Intelligent Prosthesis knee functions similarly to most non-microprocessor-controlled knees in stance when stability is most crucial.³⁴ The types of non-microprocessor-controlled knees used for comparisons were standardized in three^28,30,33 of the four studies. Comparison knees included a single-axis knee with pneumatic swing control,²⁸ a polycentric knee with hydraulic swing control,³⁰ and a single-axis knee with hydraulic stance and swing control.^30,33 The remaining study^26,27 used participants’ current non-microprocessor-controlled knee as the comparison condition. All investigators standardized the prosthetic socket between conditions, randomized the order of knee allocation, and provided accommodation periods for each prosthetic knee condition from 4 to 32 weeks.

Walking conditions and outcomes varied across the prosthetic knee studies. The dual-task walking conditions included walking on a treadmill,²⁸ walking an indoor hallway,³³ walking an outdoor sidewalk,^26,27 and traversing a standardized obstacle course with different terrain qualities.³⁰ Walking outcomes included measures of pace (i.e. self-selected walking speed^26,27,33 or time to complete an obstacle course³⁰), postural stability (i.e. overall sway²⁸), and movement efficiency (i.e. total heart beat index³⁰). Four^26–28,33 of the five articles did not report measurement of single-task performance on the walking tasks. The choice of cognitive task also varied and included a working memory task (reading numbers on a screen²⁸), a discrimination and decision-making task (visual Stroop test²⁸), a verbal fluency task,³³ and mental tracking tasks (i.e. serial subtraction^30,33 and backward digit span^26,27). None of the articles described measurement of single-task performance on the cognitive task or provided instructions for focus in the dual-task condition. One article³⁰ did not report dual-task cognitive performance.

Results

Of the four studies, only one²⁶ found that performance of a concurrent task differentially affected walking performance when using different knee components. The study showed that walking speed significantly increased while walking in dual-task conditions using the microprocessor-controlled knee compared to non-microprocessor-controlled knees, and that this increase was most evident in people who were categorized as limited community ambulators (i.e. those classified as Medicare Functional Classification Level K2). Another study³⁰ reported that the use of a microprocessor-controlled knee resulted in lower movement efficiency in dual-task relative to single-task walking. However, this difference is likely within the error of the measure and may not be clinically meaningful. There were no other significant differences reported based on measures of pace^27,33,30 or postural stability.²⁸ No significant differences between non-microprocessor- and microprocessor-controlled knees were found in studies that reported cognitive task performance.^26–28,33

Only one³⁰ study assessed walking over challenging terrain. The obstacle course used in this study may reflect walking conditions most affected by the features of a microprocessor-controlled knee with stance control.³⁵ Surprisingly, the investigators³⁰ reported trends toward dual-task improvements in walking across all knee conditions, which potentially reflected an arousal effect. Order effects may also explain these results, as single-task walking through the obstacle course always preceded dual-task walking.

Self-report

Three of the five articles^26,27,33 used self-report surveys to assess the need to concentrate on walking and difficulty completing concurrent tasks while walking between knee conditions. Participants were asked about attention given to the walking and cognitive tasks during the research session³³ and administered questions assessing average cognitive burden in their daily lives.^26,27,33 Results showed that participants reported significant reductions in attention during walking^26,27,33 and average cognitive burden³³ in the microprocessor-controlled knee condition, relative to the non-microprocessor-controlled knee.

Discussion

The aim of this review was to examine evidence related to dual-task standing and walking in people with LLA. Based on the studies reviewed, there is some evidence that the dual-task interference in standing is disproportionately greater for people with LLA compared to non-amputees. However, current research does not support the hypothesis that dual-task interference during walking is greater in people with LLA compared to controls. Cognitive demand did not increase during walking in people with LLA compared to controls³² and the addition of a cognitive task did not differentially impact walking.^9,29,31

Evidence of heightened dual-task interference in standing, but not walking, in people with LLA is surprising given that greater interference was expected as postural tasks became more challenging.¹⁹ One potential reason for this discrepancy is the difference in participant characteristics in these studies. In the two studies that found heightened dual-task interference for people with LLA in standing postural control, 63%²⁵ and 75%² of the participants reported vascular amputation etiology and the mean ages of 67.7²⁵ and 59.4² years. In comparison, fewer than 13% of participants with LLA reported vascular etiology in studies that did not find heightened dual-task interference in walking, and participants’ mean ages were between 40 and 54 years.^1,9,31 It should be noted that the etiology of amputation^29,32 and mean age³² were not reported in several walking studies. Future studies could compare younger and older people with LLA or people with different etiologies (e.g. dysvascular and traumatic) to assess differences in dual-task interference that may be due to age and cause of amputation.

Among the four studies that examined the effects of microprocessor-controlled compared to non-microprocessor-controlled knees on dual-task walking, only one²⁶ provided evidence of improved dual-task walking performance in the C-Leg microprocessor-controlled knee compared to non-microprocessor-controlled knees. This study also provided evidence that the use of this technology may be most beneficial to people with LLA who are limited community ambulators (i.e. Medicare Functional Classification Level K2). The group of limited community ambulators who demonstrated improvements in dual-task walking were older (mean age 57.1 years) than those classified as unlimited community ambulators (i.e. Medicare Functional Classification Level K3, mean age 41.9 years). Mean age of the limited community ambulators in this study was also higher than that of participants (mean ages from 38 to 49 years) in other studies that assessed the dual-task walking performance in microprocessor- and non-microprocessor-controlled knees. Results from this study suggest that people with LLA who have heightened mobility limitations and/or older age have reduced dual-task interference when wearing microprocessor-controlled knees than when wearing non-microprocessor-controlled knees.

While there was little evidence that dual-task walking performance was worse in people with LLA or mitigated by the use of microprocessor-controlled knees, the studies^26,27,31,33 that measured self-report outcomes in addition to performance outcomes contradicted performance evidence and suggests a need for greater concentration for walking in people with LLA. One study³¹ that examined dual-task walking in people with and without LLA found that participants with transfemoral amputation reported heightened attention in walking and that this heightened attention was most commonly experienced in challenging walking conditions (e.g. icy or slippery surfaces, uneven terrain). Studies^26,27,33 that examined dual-task performance of participants with transfemoral amputation found that participants reported significant reductions in the difficulty of dual-task walking^26,27,33 and cognitive burden³³ when using a microprocessor-controlled knee outside the laboratory. These results demonstrate the value of collecting both self-report and performance-based data when assessing the role of attention in postural control and may suggest the need to assess dual-task walking in challenging or unpredictable walking conditions that better reflect home and community environments.

Recommendations for future dual-task standing and walking research in people with LLA

The small number of studies identified in this review suggests that dual-task research in people with LLA is in its infancy. Additional high-quality research is needed to understand the effects of a concurrent task on standing or walking in people with LLA, and it is critical that future research builds upon previous work to effectively inform our understanding of dual-task postural control in people with LLA.

Sample sizes and characteristics

Studies included in this review had small samples that were heterogeneous with respect to important patient characteristics (e.g. age, etiology of amputation, and mobility limitations). Sample sizes ranged from 6 to 17 participants and investigators rarely described calculating sample size a priori. Studies with small samples have the potential to be underpowered, meaning that results may not be statistically significant even in cases where a true difference between groups or conditions exists.

A second consideration related to sample selection is the ability to generalize study findings to the larger LLA population. Among people with major LLA, the estimated prevalence of amputation due to dysvascular reasons is approximately 80%.³⁶ However, people with dysvascular amputation comprised only 53% of participants in the dual-task standing studies, 9% of participants in the dual-task walking studies, and 5% of the participants in prosthetic knee studies. In addition, epidemiological studies report that 67% of people with transtibial amputation and 85% of people with transfemoral amputation use assistive devices for walking outside of their home,¹⁵ and 42% of people with LLA are over the age of 65.³⁶ In contrast, none of the reviewed studies included participants that required assistive devices during data collection, and only one study²⁵ reported mean participant ages over 65 years. Thus, participants in existing dual-task studies do not well represent the predominant etiology, functional ability, or age of the general population of people with LLA. Future dual-task research that intends to generalize results to the overall LLA population should therefore attempt to recruit samples that are representative in terms of amputation etiology, mobility limitations, and age. These is also a need for research that assesses whether dual-task walking differs in subgroups defined by these characteristics to better understand whether certain people with LLA are more affected by the addition of a concurrent task than others.

Methodological considerations specific to dual-task research

Overall quality of research methods in this area varied across studies. In addition, variations in methodological approaches across studies make interpretation of the overall evidence challenging. The lack of assessment of postural and/or cognitive task performance under both single-task and dual-task conditions makes it difficult to determine patterns of interference. Only three studies^2,25,31 included in this review reported single-task and dual-task performance for both the postural and cognitive tasks. If the postural and cognitive single-task and dual-task data are not reported, potential trade-offs between the postural task and the concurrent cognitive or motor task cannot be assessed and results may be misinterpreted.

A second key methodological consideration is the instructions provided to participants under dual-task conditions. Only two studies^25,31 noted instructions requesting that study participants prioritize either the postural or the cognitive task in dual-task conditions. When instructions for prioritization are not provided, some participants may prioritize the postural task, thus experiencing no change in the postural task and dual-task interference in the cognitive task. Others may prioritize the cognitive task and experience dual-task interference in the postural task and no change in the cognitive task. Collectively, this data set may appear to show no dual-task interference across tasks, when in actuality interference just presented differently based on participants’ task prioritization. While instructions for prioritization are not always considered to be ecologically valid, their provision mitigates the concern that study participants prioritize the postural and cognitive tasks differently.²³ Future dual-task research protocols should measure and report outcomes on both the postural and cognitive tasks in both single- and dual-task conditions. In addition, instructions for task prioritization should be provided and reported in order to adequately assess potential dual-task interference.

Postural task choice and outcome measures

Walking studies in this review examined relatively simple walking tasks performed overground or on a treadmill, but it is possible that flat, level walking is not sufficiently challenging to reveal differences in postural control between groups (e.g. people with LLA and controls, people using microprocessor- and non-microprocessor-controlled knees). People with LLA typically adapt well to simple, predictable walking tasks but they lack reliable prosthetic-side sensory inputs or motor outputs. These sensorimotor structures are crucial for adapting gait during more complex or unpredictable walking tasks, such as walking over uneven surfaces, negotiating slopes, and recovering from trips or slips.¹⁹ Because the experimental conditions (e.g. level ground walking) used in the reviewed studies do not represent the situations that are most challenging for people with LLA,³¹ it is possible that dual-task interference was minimal or not present. Future work should consider assessing dual-task walking over challenging or unpredictable surfaces that may more closely reflect a variety of daily walking tasks.

Another explanation for discrepancies between performance and self-report measures noted in this review is that outcome measures used to assess dual-task interference are not sensitive enough to detect differences in dual-task performance. There is evidence that suggests that measures such as residual standard deviation,³⁷ functional near-infrared spectroscopy (fNIRS),³⁸ or psychophysiological measures³⁹ may be more sensitive to changes associated with the addition of a concurrent task than the performance measures used most often in this body of research. Furthermore, processing speed may be more sensitive to the addition of a dual-task than postural outcomes and other cognitive task outcomes (e.g. error rate). Future research in this area could consider instructing participants to focus on the postural task while concurrently measuring processing speed on a simple reaction time task to measure difference in cognitive demand related to the postural task.

Another potential direction for measurement of dual-task interference in people with LLA is through development of improved self-report instruments. While self-report results have consistently found significant differences in microprocessor-controlled and non-microprocessor-controlled knees, instruments used to date have little evidence of psychometric testing in people with LLA.^26,27,33 Additional testing of existing instruments or development of new psychometrically sound self-report instruments that assess cognitive control of standing and walking tasks may help investigators to understand the role of cognition in the control of balance and walking in people with LLA, both within laboratory and real-world environments.

Clinical implications

Since there is little definitive evidence of differential dual-task decline in previous research, one might speculate that people with LLA do not require heightened cognitive control for postural tasks, particularly walking. There is also little evidence to support the assertion that microprocessor-controlled knees can reduce the cognitive control required for walking tasks in people with LLA. However, self-reported data from survey research^15,16 and three^26,27,31,33 studies included in this review suggest that people with LLA perceive a need to concentrate on walking. Furthermore, methodological issues such as small samples, heterogeneous study populations, variations in choice of concurrent tasks, and critical limitations in the application of dual-task methods may limit the ability to conclude that people with LLA do not require heightened cognitive control for postural tasks.

Limitations of this review

This review was limited to peer-reviewed articles. Thus, emerging studies that have been disseminated as abstracts, dissertations, or theses and are not yet available as peer-reviewed articles are not included. Furthermore, the search strategy used would not identify studies that included a dual-task, but did not discuss this aspect of the study in their abstract or keywords. Finally, few studies were identified and results could not be synthesized in a meta-analysis due to heterogeneity in research aims, sample characteristics, and measured outcomes.

Conclusion

The relatively few studies assessing dual-task standing and walking in people with LLA suggest that little is known about interactions between cognition and postural tasks in this population. Definitive conclusions about the allocation of cognitive resources during standing and walking in people with LLA cannot be drawn due to small samples, sample heterogeneity, and variation in methods used in dual-task studies to date. To build upon current evidence, future research should estimate sample sizes using data from studies included in this review. Participants should be recruited to better represent the population of people with LLA, particularly with regard to important patient characteristics (e.g. amputation etiology) that are likely to affect dual-task performance. Investigators should also consider alternative techniques like standardized self-report measures and neural imaging to assess dual-task standing and walking in people with LLA.

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

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Vrieling

van Keeken

Schoppen

et al . Balance control on a moving platform in unilateral lower limb amputees. Gait Posture 2008; 28(2): 222–228.

Geurts

Mulder

TH.

Attention demands in balance recovery following lower limb amputation. J Mot Behav 1994; 26(2): 162–170.

Buckley

O’Driscoll

Bennett

SJ.

Postural sway and active balance performance in highly active lower-limb amputees. Am J Phys Med Rehabil 2002; 81(1): 13–20.

Genin

Bastien

Franck

et al . Effect of speed on the energy cost of walking in unilateral traumatic lower limb amputees. Eur J Appl Physiol 2008; 103(6): 655–663.

Boonstra

Schrama

Fidler

et al . The gait of unilateral transfemoral amputees. Scand J Rehabil Med 1994; 26(4): 217–223.

Waters

Mulroy

The energy expenditure of normal and pathologic gait. Gait Posture 1999; 9(3): 207–231.

Jaegers

Arendzen

de Jongh

HJ.

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

Nolan

Wit

Dudzinski

et al . Adjustments in gait symmetry with walking speed in trans-femoral and trans-tibial amputees. Gait Posture 2003; 17(2): 142–151.

Lamoth

Ainsworth

Polomski

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

10.

Vrieling

van Keeken

Schoppen

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

11.

Ramstrand

Nilsson

KA.

A comparison of foot placement strategies of transtibial amputees and able-bodied subjects during stair ambulation. Prosthet Orthot Int 2009; 33(4): 348–355.

12.

Jones

Twigg

Scally

et al . The mechanics of landing when stepping down in unilateral lower-limb amputees. Clin Biomech 2006; 21(2): 184–193.

13.

Halsne

Waddingham

Hafner

BJ.

Long-term activity in and among persons with transfemoral amputation. J Rehabil Res Dev 2013; 50(4): 515–530.

14.

Stepien

Cavenett

Taylor

et al . Activity levels among lower-limb amputees: self-report versus step activity monitor. Arch Phys Med Rehabil 2007; 88(7): 896–900.

15.

Gauthier-Gagnon

Grise

Potvin

Enabling factors related to prosthetic use by people with transtibial and transfemoral amputation. Arch Phys Med Rehabil 1999; 80(6): 706–713.

16.

Miller

Deathe

Speechley

et al . The influence of falling, fear of falling, and balance confidence on prosthetic mobility and social activity among individuals with a lower extremity amputation. Arch Phys Med Rehabil 2001; 82(9): 1238–1244.

17.

Miller

Speechley

Deathe

The prevalence and risk factors of falling and fear of falling among lower extremity amputees. Arch Phys Med Rehabil 2001; 82(8): 1031–1037.

18.

Berry

Microprocessor prosthetic knees. Phys Med Rehabil Clin N Am 2006; 17(1): 91–113.

19.

Woollacott

Shumway-Cook

Attention and the control of posture and gait: a review of an emerging area of research. Gait Posture 2002; 16(1): 1–14.

20.

Abernethy

Dual-task methodology and motor skills research: some applications and methodological constraints. J Hum Movement Stud 1988; 14(3): 101–132.

21.

Tombu

Jolicoeur

A central capacity sharing model of dual-task performance. J Exp Psychol Hum Percept Perform 2003; 29(1): 3–18.

22.

Ruthruff

Pashler

Klaassen

Processing bottlenecks in dual-task performance: structural limitation or strategic postponement?

Psychon Bull Rev 2001; 8(1): 73–80.

23.

Fraizer

Mitra

Methodological and interpretive issues in posture-cognition dual-tasking in upright stance. Gait Posture 2008; 27(2): 271–279.

24.

Al-Yahya

Dawes

Smith

et al . Cognitive motor interference while walking: a systematic review and meta-analysis. Neurosci Biobehav Rev 2011; 35(3): 715–728.

25.

Geurts

Mulder

Nienhuis

et al . Dual-task assessment of reorganization of postural control in persons with lower limb amputation. Arch Phys Med Rehabil 1991; 72(13): 1059–1064.

26.

Hafner

Smith

DG.

Differences in function and safety between Medicare Functional Classification Level-2 and -3 transfemoral amputees and influence of prosthetic knee joint control. J Rehabil Res Dev 2009; 46(3): 417–433.

27.

Hafner

Willingham

Buell

et al . Evaluation of function, performance, and preference as transfemoral amputees transition from mechanical to microprocessor control of the prosthetic knee. Arch Phys Med Rehabil 2007; 88(2): 207–217.

28.

Heller

Datta

Howitt

A pilot study comparing the cognitive demand of walking for transfemoral amputees using the intelligent prosthesis with that using conventionally damped knees. Clin Rehabil 2000; 14(5): 518–522.

29.

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.

30.

Meier

Hansen

Gard

et al . Obstacle course: users’ maneuverability and movement efficiency when using Otto Bock C-Leg, Otto Bock 3R60, and CaTech SNS prosthetic knee joints. J Rehabil Res Dev 2012; 49(4): 583–596.

31.

Morgan

Hafner

Kelly

VE.

The effects of a concurrent task on walking in persons with transfemoral amputation compared to persons without limb loss. Prosthet Orthot Int 2016; 40(4): 490–496.

32.

Nakamura

Moriai

Sajiki

Reaction times of normal subjects and amputees with below-knee and above-knee prostheses during stepping. Prosthet Orthot Int 1984; 8(2): 100–102.

33.

Williams

Turner

Orendurff

et al . Does having a computerized prosthetic knee influence cognitive performance during amputee walking? Arch Phys Med Rehabil 2006; 87(7): 989–994.

34.

Michael

. Modern prosthetic knee mechanisms. Clin Orthop Relat Res 1999(361): 39–47.

35.

Sawers

Hafner

BJ.

Outcomes associated with the use of microprocessor-controlled prosthetic knees among individuals with unilateral transfemoral limb loss: a systematic review. J Rehabil Res Dev 2013; 50(3): 273–314.

36.

Ziegler-Graham

MacKenzie

Ephraim

et al . Estimating the prevalence of limb loss in the United States: 2005 to 2050. Arch Phys Med Rehabil 2008; 89(3): 422–429.

37.

Howard

Wallace

Abbas

et al . Residual standard deviation: validation of a new measure of dual-task cost in below-knee prosthesis users. Gait Posture 2017; 51: 91–96.

38.

Doi

Makizako

Shimada

et al . Brain activation during dual-task walking and executive function among older adults with mild cognitive impairment: a fNIRS study. Aging Clin Exp Res 2013; 25(5): 539–544.

39.

Knaepen

Marusic

Crea

et al . Psychophysiological response to cognitive workload during symmetrical, asymmetrical and dual-task walking. Hum Mov Sci 2015; 40: 248–263.