A scoping review of voluntary gait adaptability tasks requiring cognitive demands in older adults

Article information

Phys Act Nutr. 2023;27(1):30-40
Publication date (electronic) : 2023 March 31
doi : https://doi.org/10.20463/pan.2023.0004
1Institute of Movement and Sport Gerontology, German Sport University Cologne, Cologne, Germany
*Corresponding author : Kyungwan Kim Institute of Movement and Sport Gerontology, German Sport University Cologne, Cologne, Germany, Am Sportpark Muengersdorf 6, 50933 Cologne, Germany. Tel: +49-221-4982-7160 E-mail: k.kim@dshs-koeln.de
Received 2023 February 6; Revised 2023 March 13; Accepted 2023 March 14.



Voluntary gait adaptability is a complex construct that requires cognitive demands and dynamic balance control; it also has implications for the daily lives of older adults. This ability has been extensively studied, however, a comprehensive overview of appropriate tasks for measuring voluntary gait adaptability in older adults is lacking. Our scoping review aimed to identify existing voluntary gait adaptability tasks for older adults, summarize the specific methodological features requiring cognitive demands found in previous studies, and categorize these tasks according to experimental procedure and setup.


A comprehensive literature search was performed using six databases (PubMed, SPORTDiscus, Web of Science, CINAHL, MEDLINE, and Embase). This included studies that investigated voluntary gait adaptability in older adults (≥ 65 years old) with and without neurological disorders, with a focus on experimental tasks requiring cognitive function (e.g., response to visual or auditive stimuli) while walking.


Sixteen studies were included; most involved visual stimuli, such as obstacles, stairs, or colored cues, and few required auditory stimuli. The studies were categorized according to the experimental procedure, for example, ascent/descent of obstacles (n = 3), inconsistent surfaces (n = 1), lateral gait adjustment (n = 4), obstacle avoidance (n = 6), and stepping tasks (n = 2), as well as experimental setup, including instrumented treadmills (n = 3), stairs (n = 3), and walkways (n = 10).


The results show wide heterogeneity between studies regarding experimental procedures and setup. Our scoping review highlights the need for additional experimental studies and systematic reviews on voluntary gait adaptability in older adults.


Everyday activities require constant interactions between cognitive and physical resources. Most importantly, human walking is highly adaptive, allowing the gait to be adjusted to different environmental features, such as avoiding obstacles or walking on slippery surfaces. However, in old age, cognitive deficits and unstable dynamic balance are prominent, which considerably impair stable gait patterns and adaptability to unfamiliar and ever-changing environments [1]. Older adults (OA) often suffer from walking problems, including reduced adaptability to environmental changes and hazards, which can increase their fall risk. These age-related changes are due to the progressive degradation of cognitive function and motor control abilities [2,3]

The ability to adapt to environmental circumstances, that is gait adaptability (GA), is a prerequisite for avoiding falls and other severe injuries in everyday life. GA is considered an essential indicator of neurological diseases, such as dementia [4,5], stroke [6-8], and Parkinson’s disease (PD) [9]. Recent studies on OA with PD showed impaired GA in terms of adaptation to unexpected targets [10,11]. In this context, cognitive function and motor control mechanisms are essential for appropriately executing complex gait adaptations. As gait and cognition interact closely, both domains are crucial for reacting to both unpredictable and predictable objects, as well as appropriately adjusting gait patterns [12]. For instance, when someone detects a pothole on a sidewalk, it is necessary to proactively change their gait to avoid tripping. In this case, one should visually perceive the pothole, cognitively plan alternative motor actions, and execute the planned motor actions. Depending on whether or not cognitive demands are required, GA can be divided into two types of adaptation, voluntary gait adaptability (VGA) and involuntary gait adaptability (IGA).

VGA refers to the ability to proactively adjust the gait and cope with both internal and external constraints when unexpected environmental changes occur. An example of this is gait adaptation when avoiding obstacles or other pedestrians, or specific stepping strategies when walking on uneven terrain. VGA requires continuous interactions between sensory, cognitive, and motor processes to effectively adapt to changing environments and maintain safe, goal-oriented locomotion [13]. In contrast, IGA involves instantaneous reactive motor actions in the lower and upper extremities in response to external mechanical perturbations. Examples of IGA include reactive balance control when tripping over a doorstep or slipping on a wet sidewalk [14]. As the present scoping review examines the role of cognitive function in VGA, IGA is not covered. The need to promote safe community ambulation and prevent falls has been receiving continuous attention; thus, novel training devices and intervention methods have been developed for this purpose. These devices and methods are based on improving gait and/or balance disorders in OA with or without neurological disorders.

Cognitive function is crucial for enabling individuals to adapt their behavior to environmental events, such as avoiding other pedestrians or hazards on the street, and can be considered part of VGA. One of the key components of cognitive function is motor planning, which occurs in the premotor cortex and supplementary motor areas [15]. A recent study focusing on general GA demonstrated that visual perturbations while walking overground while wearing a virtual reality headset resulted in shorter and slower strides in OA [16]. Weerdesteyn et al. [17] investigated an obstacle avoidance strategy for assessing avoidance reactions to suddenly dropping wooden obstacles during walking. After the participants had the choice to set either a shortened or lengthened step for obstacle avoidance, the OA set a lengthened step, even when a shortened step would have been more favorable. The researchers hypothesized that these findings were associated with age-related changes in decision-making. However, in the aforementioned studies, both sensory and mechanical perturbations were tested. These tasks are related to IGA, focus on motor responses, and do not have cognitive demands.

VGA in OA decreases with increasing cognitive demand. Many studies have included participants with neurological impairments, such as PD or stroke, in order to study VGA [10,11,18,19]. In an experimental study on OA aged 65 years and older, VGA was recorded in terms of a transition from a wide to a narrower path. Other studies have used dual-task paradigms during stair climbing to assess VGA [20,21]. There is a considerable body of literature on cognitive tasks affecting VGA [22-24]. The experimental setup, such as a staircase or walkway, differs between studies, as well as the GA tasks, for example, descending stairs while subtracting backward or adapting to a narrower path. In addition, some studies cannot be allocated absolutely to either VGA or IGA, such as speed-adaptation tasks on a split-belt treadmill or adaptive walking on terrains with different slopes [25,26]. To the best of our knowledge, no previous study has conducted a comprehensive literature review of VGA in OA. Moreover, earlier studies did not perform more specific clustering of VGA tasks. At this point, it remains unclear which VGA tasks are suitable for investigating the VGA of OA, and which cognitive demands are embedded in the various VGA tasks. These considerations motivated us to conduct a comprehensive literature review on the role of cognitive function in VGA tasks in OA with and without neurological impairment.

Thus, the present scoping review aimed to: a) identify the VGA tasks that were used among OA with or without neurological disorders, b) summarize the specific methodological features requiring cognitive function during walking, and c) categorize the experimental tools in terms of experimental procedure and setup. Such a summary and categorization can improve our understanding of VGA and the role of cognitive function in its tasks may have implications for future fall prevention strategies.


Our scoping review was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR) to systematically review and summarize relevant literature [27].

Literature search strategy

After a few preliminary searches to scope the literature and identify specific keywords and synonyms, six databases, PubMed, SPORTDiscus, Web of Science, CINAHL, MEDLINE, and Embase, were accessed and searched on January 15th, 2022. As fewer hits were obtained using medical subject headings (MeSHs), MeSH terms were not used to exclude potentially suitable studies. To ensure a comprehensive literature search, the search terms were entered in “All Fields” and no further restrictions were applied. The exception was the database CINAHL, where only a search in the “Title/Abstract” field could be performed. Table 1 presents an overview of the search terms used for PubMed. The same search strategy was used for the other electronic databases.

PubMed search strategy.

Study selection

The study selection process was conducted independently by two researchers. After duplicates were removed, titles and abstracts were screened for eligibility. Predetermined inclusion and exclusion criteria were applied (see Table 2). Suitable study participants were required to be over 65 years of age without any physical impairments. As it is the primary aim of this review, studies had to investigate VGA tasks overground or on a treadmill. The most important point is whether the tasks required cognitive demand while walking (e.g., visual and auditory stimuli) so that the participants had to adapt their gait pattern. Subsequently, full texts of the included studies were screened according to these criteria. After two researchers independently conducted the screening, discrepancies were identified and resolved through discussion.

Inclusion and exclusion criteria.

Analysis of available studies

The screening process was performed using the collaboration tool Rayyan [28]. After full-text screening, relevant data from the included studies were summarized in a table using Microsoft Excel (version 15.82) to conduct a collaborative abstraction of the selected literature. For studies that examined additional outcomes and parameters, only data relevant to VGA and the aim of this review were extracted. A second Excel table was used to build categories within the experimental tools. A maximum of five categories was set for each tool. Using more than five categories was not considered useful because it would not ensure a simplified overview.


Study selection procedure

The database search returned 5712 records (Fig. 1). No additional records could be identified through manual searching of the gray literature or common search engines. After removing duplicates and screening titles and abstracts, 158 articles remained. The full texts of these reports were assessed for eligibility in compliance with the inclusion criteria (see Table 2). Various reasons, such as insufficient population age (n = 59) or incorrect intervention tasks (n = 36), led to the exclusion of 142 studies. Therefore, 16 reports were included in this scoping review.

Figure 1.

Flowchart of the study selection process.

Overview of selected studies

An overview of the 16 included studies is presented in Table 3. It summarizes the content that is relevant to our research aims by referring to ‘participants’, ‘experimental setup,’ experimental procedure’, ‘outcome measures,’ and ‘specific features regarding voluntary gait adaptability in older adults.’ The publication years ranged from 2006 to 2021, with half of the studies published after 2018. All studies had a clearly described experimental design. Nine of the 16 included trials were conducted in North America, three in Europe, two in Asia, one in Australia, and one in South America.

Descriptive summary of included studies sorted by year of most recent publication. The column ‘Experimental setup’ details the walking surface used in each study (e.g., overground, treadmill, or walking matt), ‘Experimental procedure’ describes the study design (e.g., number of trials, randomization, etc.), ‘Outcome measures’ includes the obtained gait and balance parameters, and ‘Specific features regarding VGA in OA’ shows which cognitive demands were made of the participants while walking and how these affected their dynamic balance control.

Study population

The number of participants ranged from 10 [37] to 127 [4]. The study population included in this scoping review had a broad age range; the lowest average age was 67.0 ± 1.0 years [21], and the highest was 82.7 ± 6.4 years [24]. One study did not indicate an average age but had an age span consistent with the minimum age inclusion criteria of the review [31]. All articles investigated GA in both women and men, except for Zhang et al. [21], who only included women. Two studies did not provide information on sex [36,38]. Following the inclusion criteria, only healthy or neurologically impaired OA were included. All studies involving participants with any physical impairment or gait abnormalities, such as freezing of gait in Parkinson’s disease, were excluded. Two experiments focused on healthy OA [21,34], while one focused on patients with Alzheimer’s [5]. Mazaheri et al. [35] compared younger (22.7 ± 2.5) and older adults, with older individuals classified as having low (74.5 ± 7.5) or high (76.0 ± 6.6) executive function. Eight additional studies compared younger and older groups to detect age differences in the investigated outcomes. Muir et al. [3]1 studied a middle-aged population. Three research groups focused on fallers and non-fallers with a high or low risk of falling [4,24,36]. Various measurement tools have been used to classify the health status of the observed populations. No population classification of the 16 included studies was possible.

Experimental setup and procedure

We formed five categories for the experimental procedures and three categories for the experimental setups of the included studies. A summary is shown in Table 4, which provides more detailed insights into the experimental procedures used to study VGA in OA and the experimental setup in which VGA was measured. Such categorization is meaningful to scope specific VGA tasks for OA and to understand which aspects of VGA, as well as which training or measurement tools, should be considered in further experimental studies.

Categorization of included studies according to methodological characteristics.

We identified three categories of experimental setups, walkway, stairs, and instrumented treadmill. Of the 16 studies, 10 were executed on the ground using different types of walkways. As shown in Table 3, VGA was measured using instrumented walkways, such as GAITRite, electronic walkways with embedded sensors (Zeno/Protokinetics), force plates (AccuGait), and diverse surfaces. The lengths of the walkways varied from 5–15 m. Three studies used stairs or curbs for adaptability tasks. Zhang et al. [21] used a standardized staircase with five steps, whereas Di Giulio et al. [32] assessed VGA using a 2 m walkway with a stair ascent and descent. The remaining three studies used three different types of instrumented treadmills (Bertec, M-Gait, and C-Mill). Instructed walking speed was measured in 13 self-selected trials by the participants. One study individually adjusted the speed to the gait parameters of the population [35], while the other used a fixed walking speed of 0.75 m/s to match the optical flow of the virtual reality system [29]. Muir et al. [31] did not provide information regarding speed in their experimental setup.

To investigate VGA, no mechanical perturbations were included in the experimental procedures. All tasks in the included studies required cognitive effort while walking; therefore, participants had to adapt their gait patterns; however, heterogeneous tasks were used. We identified five categories of experimental procedures (i.e., types of task). The most frequently used task (six of 16) contained different types of obstacle avoidance. Four of these studies required participants to avoid and cross obstacles positioned at a fixed spot on the walkway. While avoiding obstacles, the study population of Orcioli-Silva et al. [5] was instructed to count down from 20 to one. One research group also included stationary obstacles, auditory stimuli, and moving obstacles [38]. The second category was lateral gait adjustment, which was found in four out of 16 studies. This category can be divided into two subtasks: transition from a wide to a narrow pathway [20] and lateral stepping adjustments to cues [30,36]. Kazanski and Dingwell [29] tested both conditions: walking on a gradually narrowing path and lateral maneuvering to an adjacent path. Third, three studies included the ascent/descent of obstacles in the VGA task. Two experiments were conducted on the use of stairs to measure adaptability [21,32]. Furthermore, Zhang et al. [21] approached the dual-tasking of the cognitive flexibility of participants during the ascent and descent of stairs. In the remaining experiments, a curb was used instead of stairs [33]. The classification of stepping tasks was identified in two studies [24,35]. Finally, a task category for inconsistent surfaces was formed. Rogers et al. [37] investigated adaptive changes in gait on different surfaces with supplementary obscured vision. Overall, a wide range of trials was conducted. There were differences in the number of diverse conditions in the experiments and the number of repetitions.

Outcome measures and task-specific features

Task outcomes were measured using different instruments. The results of the experiments were mainly expressed by kinematic data, for example, spatiotemporal parameters, such as step length and width, or gait velocity. The reported features of the VGA tasks, as summarized in Table 3, were determined from group differences between the YA and OA or within-group differences between trials. In almost all included studies, OA showed poorer GA. Conditions that require cognitive functioning result in reduced gait velocity, motor performance, attentional cost, and gait variability. Thus, not all results were directly attributed to age but showed, for example, important mechanisms for the indirect influences of age on the navigation of competing lateral balance tasks [29]. Cui et al. [33] did not find any age differences in stability indices when stepping down from a curb. Some research groups reported a potentially higher risk of falls for OA with affected VGA [4,23,24,34,36] or an increased cognitive load for neurologically-impaired OA [5].


The objective of this scoping review was to evaluate studies addressing VGA in OA to a) identify the VGA tasks that were used; b) categorize the experimental tools in terms of population, experimental setup, and procedure; and c) summarize the outcome measures and specific features of VGA. Our search strategy resulted in 16 studies that could be used to investigate VGA tasks in OA. The analysis of the results showed wide heterogeneity among the studies. Variations between studies can be observed in the targeted OA group, experimental setup, specific VGA tasks, and selected outcome measures. Despite this, the findings allowed the formation of categories for experimental setup and procedure. The experimental setup can be categorized into walkway, instrumented treadmill, and stairs. The experimental procedure categories included obstacle avoidance, lateral gait adjustment, ascent/descent of obstacles, stepping tasks, and inconsistent surfaces. The study population varied widely; therefore, no further categorization could be considered. These findings show that, given the current literature, it is not yet possible to conduct a systematic meta-analysis. However, we have attempted to outline indicative trends. Based on our results, we have highlighted different experimental tools for assessing VGA in detail and clarify the categorizations made. In the final section, we discuss the limitations of this scoping review and present our conclusions.

The most common reason for exclusion from our full-text screening was a young average age (n = 59). Furthermore, we included only one study on patients with neurological diseases [5]. All other records of populations with neurological diseases that were screened for the full text did not meet the criteria because of accompanying physical impairments. Many studies that investigated VGA, for example, in Parkinson’s disease, had to be excluded for bodily impairments, such as freezing of gait, or stroke patients with severe hemiparesis. Future studies should investigate tasks that require cognitive functioning in neurologically-impaired populations with no physical limitations in order to determine possible fall prevention strategies.

The measurement of health status and motor and cognitive functioning varied among the 16 included studies. Heterogeneous inclusion criteria were used to classify participants as healthy or not cognitively impaired. For instance, Lowrey et al. [23] used self-reported measures to quantify OA as healthy, whereas others measured OA objectively according to their criteria. Additionally, one experiment did not provide information regarding average age [31], while Zhang et al. [21] included an all-female population. The latter limited comparison with the other studies that included both women and men. All other research groups included both women and men but did not consider sex differences. In contrast, Raffegeau et al. [34] investigated these differences and found that women stepped closer to obstacles without increasing trail toe clearance or decreasing step length in backward walking than men. This shows a conservative strategy used by older women and indicates a difference in GA between women and men. Although tendencies have emerged, for example, most of the included studies compared younger and older adults, not all studies could be clearly placed into either category. For example, Muir et al. [31] included a middle-aged group, and Mazaheri et al. [35] conducted an age comparison of two OA groups with lower and higher executive function. Therefore, the latter study could be either be included in age comparison or impaired OA. Collectively, there is a lack of comprehensive participant characteristics to compare studies and obtain a broad understanding of the investigated population. In addition, we would have had to form more than five categories to appropriately classify all studies.

We clustered the following study setups: walkway, stairs, and instrumented treadmill. One study could not be definitively categorized as either walkway or stairs [33] as it included stepping onto and down from a curb, which could be interpreted as a single step or stepping onto an obstacle. We assigned this to the stairs classification, as a curb would imply a single stair step. Although we set up categories for the experimental settings, this illustrates the potential problems of allocating different setups to these categories. The lengths of the walkways ranged between 5–15 m. This might lead to altered VGA results, as a shorter pathway might not allow the detection of all relevant changes in GA. Thirteen out of 16 studies indicated self-selected comfortable walking speeds, which allowed extensive comparison, however, Muir et al. [31] did not report walking speed. As the latter was executed on a walkway, we could include the study in terms of VGA; however, in the absence of walking speed on a treadmill, it is possible that there was manipulation of walking velocity related to motor control and motor adaptation rather than to VGA. In Selgrade et al. [30], no harness was used on the treadmill; the other two instrumented treadmill studies used harnesses. In addition to safety reasons, participants with a harness likely felt more secure about completing the task and might have shown more risk-taking behaviors, which may have changed the outcome. Overall, there were adequate reports on experimental setup. It is possible to categorize the three terms based on a few uncertainties. This helps us to better understand VGA and the effect of setup.

For the experimental procedures, five categories were identified: obstacle avoidance, lateral gait adjustment, ascent/descent of obstacles, stepping tasks, and inconsistent surfaces. The strength of almost all the included studies was the randomized order of trials and tasks to prevent bias. We aimed to classify the GA tasks more precisely; however, not all tasks fit completely into the assigned categories. Furthermore, even within the same category, tasks differed. Within the obstacle avoidance category, all procedures primarily focused on avoiding obstacles, but the heights and sizes of the obstacles, for example, varied. Raffegeau et al. [34] described the height of the curb used, which could have been categorized as either obstacle avoidance or ascent/descent of obstacles. Upon closer examination of the study methods, the experiment was assigned to the obstacle avoidance category; however, the decisions and interpretations of category allocation remain subjective. None of the studies used individualized obstacles based on the height of the participants, which might have prevented unequal preconditions. Some studies included additional tasks. For example, Muir et al. [31] requested that participants step over an obstacle while walking, but participants in another investigation had to count numbers in addition to avoiding obstacles [5]. Both studies can be included in the same category regarding experimental procedure, but vary in measured outcomes, population, and additional tasks. The category of lateral gait adjustment describes two subtasks: adaptation of gait to a gradually narrowing path and lateral maneuvering to an adjacent path. We decided to combine these subtasks because they require both lateral precision and adaptability. Considering these aspects, the five categories allow us to form a better under tanding of the VGA tasks used and have implications for future fall prevention strategies.

Although the experimental results were mainly expressed using kinematic data, the outcome measures varied between studies. Therefore, more homogeneous measurements are required to obtain an appropriate summary. Despite the different outcome measures, OA showed poorer GA in comparison to younger adults in almost all of the included studies. Conditions that require cognitive functioning resulted in reduced gait and motor performance, which supports the findings of Conradsson et al. [18]. Indeed, GA appears to be affected by competition for cognitive tasks. Furthermore, Chapman and Hollands [36] and Lowrey et al. [23] reported that an obstacle-avoidance strategy could increase the risk of falling. These findings suggest that age-related changes, such as cognitive decline, have an impact on VGA. However, Cui et al. [33] provide evidence that healthy aging may not contribute to impaired GA. Cui et al. [33] found no age-related differences in stability indices. Taking a more detailed view, outcome measures, such as the center of mass position, differed in comparison with other stair setup studies measuring step width and gait speed [21,32]. Another aspect to consider is the different analyses of outcome measures. It is particularly important to analyze only the GA step before the obstacle or all the set steps when investigating the GA. These findings confirm that there are still controversial aspects that preclude further comparison of outcome measures and task-specific features.

Although a comprehensive literature search was conducted using six databases, we could not identify all the relevant studies. As we did not find any matching articles through manual searching, we assume that the number of omitted studies was limited. No quality assessment was conducted in this review because the main aim was to widely scope the existing literature and identify available VGA tasks. Therefore, we did not exclude studies that used a suitable task but did not meet all the quality requirements.

This scoping review was conducted to identify different VGA tasks based on the analyzed studies. The 16 studies that emerged from the existing literature were heterogeneous in some experimental tools. Nevertheless, we were able to categorize the VGA tasks in terms of experimental setup and procedure. The study population did not allow further clustering. The applied VGA tasks indicated that OA had poorer GA. However, these findings do not imply generalizability because of the heterogeneity of the experimental tools. Therefore, systematic meta-analysis is not possible at this time. Although only tentative implications for future fall prevention strategies can be drawn, the results of this review provide a preliminary overview of existing VGA studies and can be understood as the first step into further research. Future studies should aim to replicate the results of existing VGA studies in order to validate the established categories. Further studies should investigate the association between different categories, such as the study population, task-specific features, and outcomes.


All authors contributed to the conception and design of this study. Kim and Zijlstra fine-tuned the initial study plan. Vinent and Deller collected and analyzed the data. Kyungwan Kim, Marie Vinent and Lena Deller wrote the first draft of this manuscript. Wiebren Zijlstra commented on previous versions of this manuscript. All the authors have read and approved the final version of the manuscript.


1. Bueno GAS, Gervásio FM, Ribeiro DM, Martins AC, Lemos TV, de Menezes RL. Fear of falling contributing to cautious gait pattern in women exposed to a fictional disturbing factor: a non-randomized clinical trial. Front Neurol 2019;10:283.
2. Perrot A, Bertsch J. Role of age in relation between two kinds of abilities and performance in acquisition of new motor skill. Percept Mot Skills 2007;104:91–101.
3. Voelcker-Rehage C. Motor-skill learning in older adults-a review of studies on age-related differences. Eur Rev Aging Phys Act 2008;5:5–16.
4. Pieruccini-Faria F, Montero-Odasso M. Obstacle negotiation, gait variability, and risk of falling: results from the “gait and brain study”. J Gerontol A Biol Sci Med Sci 2019;74:1422–8.
5. Orcioli-Silva D, Simieli L, Barbieri FA, Stella F, Gobbi LTB. Adaptive walking in Alzheimer’s disease. Int J Alzheimers Dis 2012;2012:674589.
6. Heeren A, Van Ooijen MW, Geurts ACH, Day BL, Janssen TWJ, Beek PJ, Roerdink M, Weerdesteyn V. Step by step: a proof of concept study of C-Mill gait adaptability training in the chronic phase after stroke. J Rehabil Med 2013;45:616–22.
7. Moore JL, Roth EJ, Killian C, Hornby TG. Locomotor training improves daily stepping activity and gait efficiency in individuals poststroke who have reached a “plateau” in recovery. Stroke 2010;41:129–35.
8. van Ooijen MW, Heeren A, Smulders K, Geurts ACH, Janssen TWJ, Beek PJ, Weerdesteyn V, Roerdink M. Improved gait adjustments after gait adaptability training are associated with reduced attentional demands in persons with stroke. Exp Brain Res 2015;233:1007–18.
9. Song J, Paul SS, Caetano MJD, Smith S, Dibble LE, Love R, Schoene D, Menant JC, Sherrington C, Lord SR, Canning CG, Allen NE. Home-based step training using videogame technology in people with Parkinson’s disease: a single-blinded randomised controlled trial. Clin Rehabil 2018;32:299–311.
10. Caetano MJD, Lord SR, Allen NE, Brodie MA, Song J, Paul SS, Canning CG, Menant JC. Stepping reaction time and gait adaptability are significantly impaired in people with Parkinson’s disease: implications for fall risk. Park Relat Disord 2018;47:32–8.
11. Yiou E, Caderby T, Delafontaine A, Fourcade P, Honeine JL. Balance control during gait initiation: state-of-the-art and research perspectives. World J Orthop 2017;8:815–28.
12. Shimada H, Ishii K, Ishiwata K, Oda K, Suzukawa M, Makizako H, Doi T, Suzuki T. Gait adaptability and brain activity during unaccustomed treadmill walking in healthy elderly females. Gait Posture 2013;38:203–8.
13. Caetano MJD, Menant JC, Schoene D, Pelicioni PHS, Sturnieks DL, Lord SR. Sensorimotor and cognitive predictors of impaired gait adaptability in older people. J Gerontol A Biol Sci Med Sci 2017;72:1257–63.
14. Weerdesteyn V, Hollands KL, Hollands MA. Gait adaptability. Handb Clin Neurol 2018;159:135–46.
15. Pelicioni PHS, Lord SR, Okubo Y, Menant JC. Cortical activation during gait adaptability in people with Parkinson’s disease. Gait Posture 2022;91:247–53.
16. Osaba MY, Martelli D, Prado A, Agrawal SK, Lalwani AK. Age-related differences in gait adaptations during overground walking with and without visual perturbations using a virtual reality headset. Sci Rep 2020;10:15376.
17. Weerdesteyn V, Nienhuis B, Duysens J. Advancing age progressively affects obstacle avoidance skills in the elderly. Hum Mov Sci 2005;24:865–80.
18. Conradsson D, Hinton DC, Paquette C. The effects of dual-tasking on temporal gait adaptation and de-adaptation to the split-belt treadmill in older adults. Exp Gerontol 2019;125:110655.
19. Pieruccini-Faria F, Ehgoetz Martens KA, Silveira CRA, Jones JA, Almeida QJ. Interactions between cognitive and sensory load while planning and controlling complex gait adaptations in Parkinson’s disease. BMC Neurol 2014;14:250.
20. Shkuratova N, Taylor N. The influence of age on gait parameters during the transition from a wide to a narrow pathway. Physiother Res Int 2008;13:75–83.
21. Zhang C, Sun W, Song Q, Gu H, Mao D. Performance of older adults under dual task during stair descent. J Exerc Sci Fit 2018;16:99–105.
22. Geerse DJ, Roerdink M, Marinus J, van Hilten JJ. Assessing walking adaptability in Parkinson’s disease: “The Interactive Walkway. ” Front Neurol 2018;9:1096.
23. Lowrey CR, Watson A, Vallis LA. Age-related changes in avoidance strategies when negotiating single and multiple obstacles. Exp Brain Res 2007;182:289–99.
24. Yamada M, Higuchi T, Tanaka B, Nagai K, Uemura K, Aoyama T, Ichihashi N. Measurements of stepping accuracy in a multitarget stepping task as a potential indicator of fall risk in elderly individuals. J Gerontol A Biol Sci Med Sci 2011;66:994–1000.
25. Yokoyama H, Sato K, Ogawa T, Yamamoto SI, Nakazawa K, Kawashima N. Characteristics of the gait adaptation process due to split-belt treadmill walking under a wide range of right-left speed ratios in humans. PLoS One 2018;13e0194875.
26. Zijlstra W, Rutgers AWF, Van Weerden TW. Voluntary and involuntary adaptation of gait in Parkinson’s disease. Gait Posture 1998;7:53–63.
27. Tricco AC, Lillie E, Zarin W, O’Brien KK, Colquhoun H, Levac D, Moher D, Peters MDJ, Horsley T, Weeks L, Hempel S, Akl EA, Chang C, McGowan J, Stewart L, Hartling L, Aldcroft A, Wilson MG, Garritty C, Lewin S, Godfrey CM, Macdonald MT, Langlois EV, Soares-Weiser K, Moriarty J, Clifford T, Tunçalp Ö, Straus SE. PRISMA extension for scoping reviews (PRISMA-ScR): checklist and explanation. Ann Intern Med 2018;169:467–73.
28. Ouzzani M, Hammady H, Fedorowicz Z, Elmagarmid A. Rayyan-a web and mobile app for systematic reviews. Syst Rev 2016;5:210.
29. Kazanski ME, Dingwell JB. Effects of age, physical and self-perceived balance abilities on lateral stepping adjustments during competing lateral balance tasks. Gait Posture 2021;88:311–7.
30. Selgrade BP, Childs ME, Franz JR. Effects of aging and target location on reactioin time and accuracy of lateral precision stepping during walking. J Biomech 2020;104:109710.
31. Muir BC, Bodratti LA, Morris CE, Haddad JM, van Emmerik REA, Rietdyk S. Gait characteristics during inadvertent obstacle contacts in young, middle-aged and older adults. Gait Posture 2020;77:100–4.
32. Di Giulio I, Reeves ND, Roys M, Buckley JG, Jones DA, Gavin JP, Baltzopoulos V, Maganaris CN. Stair gait in older adults worsens with smaller step treads and when transitioning between level and stair walking. Front Sport Act Living 2020;2:63.
33. Cui C, Kulkarni A, Rietdyk S, Barbieri FA, Ambike S. Synergies in the ground reaction forces and moments during double support in curb negotiation in young and older adults. J Biomech 2020;106:109837.
34. Raffegeau TE, Kellaher GK, Terza MJ, Roper JA, Altmann LJ, Hass CJ. Older women take shorter steps during backwards walking and obstacle crossing. Exp Gerontol 2019;122:60–6.
35. Mazaheri M, Roerdink M, Bood RJ, Duysens J, Beek PJ, Peper CLE. Attentional costs of visually guided walking: effects of age, executive function and stepping-task demands. Gait Posture 2014;40:182–6.
36. Chapman GJ, Hollands MA. Age-related differences in visual sampling requirements during adaptive locomotion. Exp Brain Res 2010;201:467–78.
37. Rogers HL, Cromwell RL, Grady JL. Adaptive changes in gait of older and younger adults as responses to challenges to dynamic balance. J Aging Phys Act 2008;16:85–96.
38. Gérin-Lajoie M, Richards CL, McFadyen BJ. The circumvention of obstacles during walking in different environmental contexts: a comparison between older and younger adults. Gait Posture 2006;24:364–9.

Article information Continued

Figure 1.

Flowchart of the study selection process.

Table 1.

PubMed search strategy.

Search level Search term
Key terms #1 “old*” OR “older adults” OR “senior*” OR “elder*” OR “aged” OR “aging” OR “ageing” OR “community-dwelling”
Key terms #2 “instrument* treadmill*” OR “treadmill*” OR “C-Mill*” OR “gait* analysis*” OR “interactive walk* way” OR “hand-made set*”
Key terms #3 “proactive adapt*” OR “proactive step*” OR “adaptive” OR “feedforward control*” OR “gait adapt*” OR “adaptive walk*” OR “voluntary step*” OR “target* step*” OR “step* adjustment” OR “visual* cued walk*”

Table 2.

Inclusion and exclusion criteria.

Inclusion criteria Exclusion criteria
Population OA (>65 years old) All other diseases (suffers from severe walking difficulties such as paraplegic patients, freezing of gait)
Healthy OA
Neurologically diseased (e.g., fallers, Parkinson’s disease, stroke, dementia, traumatic brain injury, multiple sclerosis)
Task(s) Tasks that require cognitive demands during walking: Visual stimuli such as obstacles, stepping targets, stairs, curbs, pedestrians, or corner, and auditive stimuli such as numerical or nominal objects Response to external mechanical disturbances, and perturbation.
No manipulation of the walking speed of the treadmill
Outcome Voluntary gait adaptability Involuntary gait adaptability
Setting On the overground or treadmill Other settings
Study design Cross-sectional studies, experimental studies, RCT, observational studies, interventional studies Literature reviews, non-peer-reviewed articles, book chapters, conference papers, theses, case reports
Other Language English or German Other languages; full text not available

RCT, randomized controlled trial.

Table 3.

Descriptive summary of included studies sorted by year of most recent publication. The column ‘Experimental setup’ details the walking surface used in each study (e.g., overground, treadmill, or walking matt), ‘Experimental procedure’ describes the study design (e.g., number of trials, randomization, etc.), ‘Outcome measures’ includes the obtained gait and balance parameters, and ‘Specific features regarding VGA in OA’ shows which cognitive demands were made of the participants while walking and how these affected their dynamic balance control.

Study Subjects Experimental setup Experimental procedure Outcome measures Specific features regarding VGA in OA
Effects of age, physical and self-perceived balance abilities on lateral stepping adjustments during competing lateral balance tasks Kazanski and Dingwell (2021) [29] 20 YA (w=11, m=9, 21.7±2.6) Instrumented treadmill in a virtual-reality system (M-Gait) 5 trials on a virtual reality path navigation task on the treadmill Lateral stepping adjustment during two competing lateral balance sub-tasks that manipulated either path width or location Spatial thresholds for stepping error onset and sub-task exchange OA made more subtask A stepping errors on wider paths and exchanged sub-tasks earlier than YA
18 OA (w=13, m=5, 71.6±6.0) The fixed walking speed at 0.75 m/s (optic flow matched) 1. Sub-task A: Walking on a gradually narrowing path Results were not directly attributed to age but important mechanisms for indirect influences of age on the navigation of competing for lateral balance tasks
2. Sub-task B: Deciding when/how to laterally maneuver to an adjacent path
Effects of aging and target location on reaction time and accuracy of lateral precision stepping during walking Selgrade et al. (2020) [30] 11 YA (w=4, m=7, 23.4±2.3) Instrumented treadmill (Bertec) Lateral precision stepping on a laser beam target lines on the side of the treadmill: 3 targets on 4 locations (12 targets in total) Stepping accuracy to the target Divergence time (time after toe-off at which the foot diverged from its normal swing trajectory to move forward to the target) OA showed larger errors and later divergence times than YA
11 OA (w=9, m=2, 77.5±6.2) Self-selected comfortable walking speed EMG (Trigno) YA increased early swing gluteus medius activity with lateral target distance, but OA did not
Neuromuscular evidence for precision stepping deficits in OA
Gait characteristics during inadvertent obstacle contacts in young, middle-aged, and older adults Muir et al. (2020) [31] 20 YA (w=9, m=11, 2035 years) 15-m walkway embedding three force plates (AccuGait) 10 trials of stepping over the obstacle while walking Foot trajectories and head angles Based on the contact/non-contact with the obstacle with the leading/trail leg (comparison between both conditions) OA had fewer obstacle contact trials primarily with the lead limb OA showed inappropriate foot placement before the obstacle: impaired ability to gather obstacle position information
14 MA (w=10, m=4, 5064 years) Required walking speed unknown Lower contact rates compared to YA were due to the cautious strategies adopted during obstacle crossing
19 OA (w=8, m=11, 6579 years) Obstacle on the pathway (height of 26 cm) 3D motion capture analysis (Vicon)
Stair gait in older adults worsens with smaller step treads and when transitioning between level and stair walking Di Giulio et al. (2020) [32] 21 YA (w=8, m=13, 29.0±1.0) 2-m walkway and stair ascent and descent 4 trials in a randomized order: Gait velocity OA showed more handrail use, slower walking speed, smaller foot clearance, larger trunk rotation, smaller lead foot horizontal and vertical clearance
1. Ascent from standing start Trunk orientation
20 OA (w=10, m=10, 74.0±1.0) 2. Descent from standing start Foot overhang
Self-selected comfortable walking speed 3. Ascent preceded and followed by 2 m walking Foot clearance Larger going and pausing before negotiating stairs are needed to improve stair safety for OA
4. Descent preceded and followed by 2 m walking
Synergies in the ground reaction forces and moments during double support in curb negotiation in young and older adults Cui et al. (2020) [33] 10 YA (w=7, m=3, 22.9±6.7) 8-m walkway included an elevated curb (4 m long, 1 m wide, 15 cm high) with two embedded force plates (AccuGait) 15 trials of stepping up and down a curb in an alternating sequence Starting positions were adjusted to ensure foot contact with the force plate and the right foot crossed the curb edge first COM (center of mass) position COM velocity No age differences in the stability indices
GRVs (ground reaction variables) Healthy aging may not contribute to the diminished stability of the resultant force and moments
10 OA (w=8, m=2, 73.5±5.0) Self-selected comfortable walking speed 3D motion capture analysis (Vicon) COP (center of pressure)
Twisting moment about the vertical axis
Older women take shorter steps during backwards walking and obstacle crossing Raffegeau et al. (2019) [34] 54 OA (w=27, m=27, 72.0±5.0) 8-m walkway Self-selected comfortable walking speed 5 trials on 3 conditions in a fixed order: Gait speed No sex differences in gait speed and step width
Step width Decreased step length in BW only in women OA: conservative strategy
Obstacle (a wooden dowel fixed at 10 cm to simulate curb height) 1. Forward walking (FW) Step length Women OA stepped closer to the obstacle without increasing trail toe-clearance
2. Backwards walking (BW) Approach to obstacle
3D motion capture analysis (Vicon) 3. Obstacle crossing Toe clearance Shorter steps and closer obstacle approach in women OA indicate the predisposing to tripping or falling
Landing distance
Obstacle Negotiation, Gait Variability, and Risk of Falling: Results From the “Gait and Brain Study” Pieruccini-Faria et al. (2019) [4] 110 non-fallers (w=65, m=45, 72.3±5.3) Electronic walkway with embedded sensors (Zeno/Pro-tokinetics, 6.0 x 0.9 m) Two walking trials Walking speed Fallers showed higher step time variability and step length variability when negotiating the obstacle than non-fallers
27 fallers (w=18, m=9, 71.8±4.5) Self-selected comfortable walking speed 1) Unobstructed Step time variability
2) Obstructed: Obstacle on the walkway at 85 cm from the end of the walkway (30 cm height and 70 cm width) Step length variability High gait variability before crossing obstacles may be a risk factor for falls
Performance of older adults under dual task during stair descent Zhang et al. (2018) [21] 30 OA (w=30, m=0, 67.0±1.0) A standardized staircase (five steps in total, 17 cm riser x 29 cm tread x 150 cm width) Three trials on two tasks in a randomized order: Gait speed Gait speed, foot clearance, and hip flexion angle decreased, and step width increased under the dual-task condition
1. Single task: Stair descent without additional task Single support
2. Dual task: Stair descent while performing subtracting in a series of three from randomly selected numbers between 100 and 300 Step width
Self-selected comfortable walking speed 3D motion capture analysis (Vicon) Foot clearance Gait performance and postural control of women OA decreased by the second cognitive task, and they predisposed to implement a compensatory strategy for enhancing postural stability
Range of motion
Attentional costs of visually guided walking: effects of age, executive function and stepping-task demands Mazaheri et al. (2014) [35] 15 YA (w=8, m=7, 22.7±2.5) Instrumented treadmill (C-Mill) Stepping targets on the treadmill Stepping accuracy in stepping targets OA with high EF showed more attentional demand in uncued walking than the other groups
10 OA with low executive function (EF) (w=5, m=5, 74.5±7.5) Two trials for each condition
1. Uncued walking Reaction time to vibration stimuli OA with low EF showed no differences between walking conditions, whereas the attentional costs of OA with high
Individually adjusted walking speed, step length, and step width 2. Regularly cued walking
3. Irregularly cued walking EF increased from regularly to irregularly cued walking
15 OA with high EF (w=11, m=4, 76.0±6.6) + 11 vibration stimuli in each trial in that participants pressed the button as soon as possible Decreased flexibility of OA with low EF may be attributed to the already increased attentional costs of uncued walking
Adaptive walking in Alzheimer's Disease Orcioli-Silva et al. (2012) [5] 19 OA with AD (w=14, m=5, 79±6.1) 8-m walkway Self-selected comfortable walking speed Obstacles on the pathway 5 trials on 4 tasks in a randomized order: Stride length OA with AD walked more slowly when confronted either with an obstacle or a secondary task
1. Free gait without DT Step width
2. Free gait with DT Singe support AD patients used anticipatory and online adjustments for stability during the task
3. Adaptive gait without DT Double support
3D motion capture analysis (Optotrak) 4. Adaptive gait with DT Stride duration Increasing task complexity enhances cognitive load and the risk of falls for AD patients
*DT: Counting down from 20 to 1 Stride velocity
*Adaptive gait: Obstacle avoidance
Measurements of stepping accuracy in a multitarget stepping task as a potential indicator of fall risk in elderly individuals Yamada et al. (2011) [24] 31 OA with a low risk of falling (% of men=25.80%, 82.7±6.4) A black elastic mat walkway (10 m long and 1 m wide) with 45 pieces of a 10 x 10 cm square in three different colors (red, blue, and yellow) Participants were instructed: Stepping failure (failure to step on the footfall target) OA with high risk in mobility showed a higher rate of stepping and avoidance failure than with low risk
Self-selected comfortable walking speed 1) to step on a footfall target with either side of the foot and any part of the sole
87 OA with a high risk of falling (% of men=26.40, 80.7±7.9) men=25.80%, Classification of the high risk in falling: fall experiences and TUG over 13.5 sec 2) to take as many steps as necessary while walking between the lines to comfortably walk toward the next footfall target Avoidance failure (failure to avoid distracters) Measuring stepping accuracy may contribute to identifying OA with a high risk of falling
3) not to step on the distractors
Age-related differences in visual sampling requirements during adaptive locomotion Chapman and Hollands (2010) [36] 10 YA (w=unknown, m=unknown, 25.1±2.68) 8-m walkway Self-selected comfortable walking speed 3D motion capture analysis (Vicon) A total of 70 trials: 10 no targets, 30 central targets, and 30 lateral targets Gaze parameters (e.g., saccadic latency, fixation numbers) High-risk OA required more time to plan and execute mediolateral stepping adjustments than low-risk OA and YA
48 LEDs placed under the walkway Walking velocity
10 OA with a low risk of falling (w=unknown, m=unknown, 69.9±4.53) Gaze tracking (ASL 500 mobile) 24 were placed in a central strip (central targets) and the other Path of progression
Step width Impaired ability to make rapid sideway stepping adjustments in avoiding obstacles may contribute to the risk of trips or falls in OA
Two sets of opto-electric timing gates (Metrodyne) Number of steps
10 OA with a high risk of falling (w=unknown, m=unknown, 76.1±5.32) 24 were placed in a more lateral location at approximately 125% of the normal step width Mean mediolateral stepping error
Stepping errors
Step variability
The influence of age on gait parameters during the transition from a wide to a narrow pathway Shkuratova and Taylor (2008) [20] 20 YA (w=10, m=10, 26.6±6.1) 8.2-m instrumented walkway (GAITRite, 732 cm long and 61 cm wide active area) 3 walking trials: The transition from a wide to a narrow pathway preferred walking speed Step length OA made a successful transition from a wide to a narrow pathway: no aging effects
Step time
Step width
20 OA (w=10, m=10, 74.3±7.2) Self-selected comfortable walking speed Base of support OA decreased their base of support when approaching the transition step
Double support time
Adaptive changes in gait of older and younger adults as responses to challenges to dynamic balance Rogers et al. (2008) [37] 5 YA (w=3, m=2, 27.20±4.09) 8-m walkway in two conditions: 3 trials on 4 gait conditions in a randomized order: Balance adaptations in: Age differences for GSR on all conditions: OA adopted a more stable gait pattern than YA
1. Noncompliant and consistent surface consisted of a linoleum floor 1. Gait on a non-compliant and consistent surface with normal vision Gait velocity
2. Compliant and inconsistent surface constructed from four panels 2. Gait on a compliant and inconsistent surface with normal vision Cadence OA and YA exhibited similar adaptive balance strategies while showing and increasing steps/s, under proprioceptive and proprioceptive-visual challenges to dynamic balance
5 OA (w=3, m=2, 68.0±1.87) 3. Gait on a non-compliant surface with vision obscured Gait-stability ratio(GSR)
Self-selected comfortable walking speed 4. Gait on a compliant and inconsistent surface with vision obscured
Age-related changes in avoidance strategies when negotiating single and multiple obstacles Lowrey et al. (2007) [23] 8 YA (w=4, m=4, 23.1±2.0) 5-m instrumented walkway (GAITRite) 6 trials on an unobstructed walking (control trial) Step length OA crossed the obstacles with a reduced step velocity and stepped closer to the trailing edge
Self-selected comfortable walking speed 6 trials on each obstructed walking in a randomized order: Step width
8 OA (w=4, m=4, 76.1±4.3) Obstacle on the pathway p.5 cm x 5 cm piece of wood) 1. Single obstacle crossing Walking velocity OA used smaller step lengths and widths using a narrower base of support than YA
3D motion capture analysis (Optotrak) 2. Double obstacle crossing Horizontal takeoff and landing distances from the obstacles OA’s avoidance strategy could increase their risk of falling or imbalance when stepping over an obstacle
The circumvention of obstacles during walking in different environmental contexts: a comparison between older and younger adults Gerin-Lajoie et al. (2006) [38] 9 YA (22-35 years, 24.6±4.1) 10-m walkway Self-selected comfortable walking speed A total of 90 trials were performed 45 of 90 trials with auditory messages through headphones (numerical, nominal, spatial, prices, product name, store department) Gait speed The greater effect of auditory distractions in OA than in YA
3D motion capture analysis (Optotrak) 3 conditions, each condition was repeated 5 times: Unobstructed step length
A mo ving obstacle system using a mannequin that was mounted on an overhanging rail (3.66 m long) crossing a 10 m long walkway at a 45° 1. Walking and no obstacles included Personal space More increased personal space in OA than in YA
9 OA (66-75, 69.7±3.2) 2. Walking and a stationary obstacle The error rate on answering questions related to the massages
3. Walking and a moving obstacle Increased environmental information processing affected both the motor and cognitive performance of OA more than in YA
Additional: catch trials to make the initial prediction of the mannequin‘ s more difficult
Task: avoid the mannequin and walk 10m straight ahead at natural speed without stopping

AD, Alzheimer’s disease; DT, dual-task; EMG, electromyography; MA, middle-aged adults; OA, older adults; TUG, timed up-and-go test; YA, young adults.

Table 4.

Categorization of included studies according to methodological characteristics.

Study Experimental procedure
Experimental setup
Ascent/descent of obstacles (n = 3) Inconsistent surface (n = 1) Lateral gait adjustment (n = 4) Obstacle avoidance (n = 6) Stepping task (n = 2) Instrumented treadmill (n = 3) Stairs (n = 3) Walkway (n = 10)
Chapman and Hollands (2010) [36] x x
Cui et al. (2020) [33] x x
Di Giulio et al. (2020) [32] x x
Gerin-Lajoie et al. (2006) [38] x x
Kazanski and Dingwell (2021) [29] x x
Lowrey et al. (2007) [23] x x
Mazaheri et al. (2014) [35] x x
Muir et al. (2020) [31] x x
Orcioli-Silva et al. (2012) [5] x x
Pieruccini-Faria et al. (2019) [4] x x
Raffegeau et al. (2019) [34] x x
Rogers et al. (2008) [37] x x
Selgrade et al. (2020) [30] x x
Shkuratova and Taylor (2008) [20] x x
Yamada et al. (2011) [24] x x
Zhang et al. (2018) [21] x x