Reduced adaptability to balance perturbations in older adults with probable cognitive impairment after a severe fall

Abstract

Falls in older individuals often result from unexpected balance disturbances during walking, necessitating the analysis of recovery strategies for effective falls prevention. This becomes particularly crucial for individuals with cognitive impairment, who face a higher fall risk compared to cognitively healthy adults. Hence, our study aimed to compare the recovery response to standardized walking perturbations on a treadmill between older adults with cognitive impairment and cognitively healthy older adults. 36 individuals with a recent history of a severe fall, leading to an emergency department visit without subsequent admission, were stratified into two groups (with and without probable cognitive impairment) based on scores of the Montreal Cognitive Assessment. Recovery performance was quantified using force plate data from a perturbation treadmill (M-Gait, Motek Medical B.V., Amsterdam, the Netherlands), specifically evaluating the number of steps needed to restore step length and width to pre perturbation baseline across two trials of nine different perturbations. Individuals with cognitive impairment (n = 18, mean age: 74.7) required significantly (p = 0.045, Cohen’s d = 0.69) more steps to recover total steps after perturbations compared to cognitively healthy individuals (n = 18, mean age: 69.7). While step width recovery was similar between the groups, those with probable cognitive impairment required significantly more steps to recover their step length (p = 0.039, Cohen’s d = 0.72). Thus, our findings indicate that older adults with probable cognitive impairment manifest inferior gait adaptability, especially in adapting step length, potentially underscoring a critical aspect for effective falls prevention in this population.

Introduction

Older adults with cognitive impairment (OACI) are particularly prone for falls with a two-to-three-fold higher risk than their cognitively healthy counterparts (OACH) [1]. The majority of falls occur during walking primarily induced by unanticipated disturbances in balance [2, 3], frequently stemming from slips or trips. These balance perturbations induce abrupt shifts in an individual’s center of mass, potentially resulting in falls with physical repercussions such as injuries [4, 5], as well as psychosocial consequences such as concerns of falling [6].

As task-specificity is integral to physical training, perturbation-based balance training (PBT) has become increasingly important in falls prevention efforts [7–9]. Vice versa, the measurement of an individual’s reactive dynamic balance in response to perturbations, which remains still elusive [10], may serve as an early indicator of future fall risk and a target for interventions [11], notably research in individuals with cognitive impairments is missing [12–14].

Different outcome measures such as spatiotemporal or kinetic parameters have been explored previously [10]. However, many of these methods require marker-based recordings, which are rather expensive and time consuming. Therefore, these methods may be particularly challenging to apply in clinical settings. Using the data of force plates, which are often integrated in treadmills, may present a cost- and time-efficient alternative to assess reactive dynamic balance by analyzing the time or number of steps needed to restore gait kinematics after a perturbation [15].

A previous investigation demonstrated diminished reactive balance control to stance perturbations in individuals with mild cognitive impairment compared to their cognitively healthy counterparts [16]. However, there is a lack of studies during walking. Gerards and colleagues examined the walking adaptability to perturbations in older individuals with or without a history of falls, utilizing the margin of stability to quantify the number of recovery steps. Their findings indicated that fallers exhibited inferior gait adaptability [11]. Notably, their study did not address individuals with a recent severe fall, which is defined as an incident prompting a visit to the emergency department (ED) or those with cognitive impairment. According to global guidelines for falls prevention, these groups are at an elevated risk for subsequent falls and functional decline [17].

Our objective was to evaluate the recovery performance to perturbations during walking in individuals who had experienced a severe fall requiring presentation to the ED, while accounting for their cognitive performance. This investigation aims to offer insights into the future fall risk and potential targets for interventions within this high-risk cohort for functional decline. We hypothesize that OACI will demonstrate a diminished recovery performance in response to standardized perturbations.

Material and methods

Participants

All participants were recruited for the SeFallED study, which is a mixed methods study ongoing at the Carl von Ossietzky University Oldenburg, Germany [18]. For the current exploratory sub-study, baseline data from participants, who underwent assessments in the gait laboratory between November 2021 and May 2023, were used. Recruitment for the SeFallED study started on November 15, 2021. The study is in accordance with the declaration of Helsinki (1974), has been prospectively registered in the German Clinical Trials Register (DRKS-ID: 00025949), and was approved by the Medical Ethics Committee of the University Oldenburg (number 2021–120). The recruitment procedures have been previously described [18, 19]. In summary, participants provided written consent for future contact while in the ED to a member of the study team. After being contacted via telephone, a trained member of the study team visited the participants at their home to procure their written informed consent for participation. In cases involving individuals with significant cognitive impairment or dementia, consent for participation was additionally obtained from their legal representative.

Eligible Participants were 60 years or older, presented to the ED of either the Klinikum or the Evangelisches Krankenhaus Oldenburg after a fall and were discharged within 72 hours [18]. Besides complete dependency on walking aids or subjectively not being able to walk 400m, other exclusion criteria were neurodegenerative diseases, a previous history of a stroke or an incomplete paraplegia. Furthermore, individuals, who hold on to the handrail throughout the whole trial, were not included in the analysis for recovery performance as this behavior may not reflect real world scenarios and may influence recovery responses.

Participants were stratified into OACH and OACI using the score of the Montreal Cognitive Assessment (MoCA) (maximum of 30 points) [20], setting the cutoff for probable cognitive impairment at ≤ 24 points to reduce false positives [21].

Study overview

Suitable participants were addressed in the ED to receive consent for being contacted via telephone. Within a week after discharge, a member of the study team scheduled a home visit to obtain written informed consent. During this visit, an extensive geriatric assessment was conducted. Afterwards participants were invited to visit the gait laboratory. All details of the assessment battery have been published before [18].

Reactive dynamic balance assessment

Reactive dynamic balance was assessed using the M-Gait treadmill (Motek Medical B.V., Amsterdam, the Netherlands). The participants were secured during treadmill walking using a safety harness and were instructed to not use the handrails, if possible. Due to ethical concerns, handrails were not removed from the treadmill.

After a familiarization period of six minutes, during which participants identified their individualized preferred treadmill gait speed, they underwent a perturbation protocol twice, with a two-minute break between each trial. It included nine perturbations (two slips for each leg, two trips for each leg, one emergency stop, two sways and one pitch–and—5° respectively) and a sufficient washout period of twenty to thirty seconds between the perturbations [15, 22]. Perturbations were triggered through the integrated force plates at initial foot contact through custom application as soon as the vertical force component ranged between 40 N and 200 N, and the impulse exceeded 800 Ns (D-Flow version 3.34.2, Motek Medical BV, Amsterdam, The Netherlands). Using the split belt option, each leg was perturbed in anterior-posterior direction twice, mimicking slips and trips. Slips were simulated by an acceleration of 3 m∙s^-2 to a maximum of 180% of treadmill walking speed for 0.42 s. Trips were provoked similarly with a deceleration of 3 m∙s^-2 to a minimum of 40% of treadmill walking speed, which is in line with previous research [23–26]. Besides these single leg perturbations, an emergency stop as in public transport was imitated with both belts decelerating by 9 m∙s^-2 for 0.12 s [25, 27]. Further, the protocol included two contralateral sways, perturbing each leg once by a 5-cm platform translation in 3 m∙s^-2 [25, 26]. A pitch of + and– 5 degrees respectively, with a duration of 1.0 s imitated small slope changes as present on sidewalks or at bus stops. The full protocol has been published before and a schematic illustration of the perturbations is presented in Fig 1 [18].

[Figure omitted. See PDF.]

Reactive dynamic balance assessment–outcome measures.

In addition to assessing overground and treadmill walking speed, outcome measures include the quantification of steps required to return to baseline levels. Consistent with previous research [11, 28], walking adaptability in this study was evaluated by examining recovery performance across the nine distinct perturbations conducted over two trials. Beyond the comprehensive analysis of total recovery, which incorporates both step width and step length, we conducted separate analyses for step width and step length recovery. The quantification of each participant’s recovery performance (total steps, step length, step width) adhered to the methodology outlined by Rosenblum and colleagues [15].

Data curation and outcome calculations

Ground reaction forces were processed using a custom MATLAB (The MathWorks Inc., version 9.12.0.1956245 (R2022a), Natrick, MA, USA) script. Force, momentum, and center of pressure (CoP) data were filtered using a second order lowpass Butterworth filter with a cutoff frequency of 20 Hz in line with previous approaches [29, 30]. Following previous recommendations, the CoP data was set to zero if vertical force components were below 80 N [30]. Initial contact was defined as the local maxima in CoP measured in anteroposterior direction. Similarly, toe off was defined as the local minima. To calculate step length and step width the formula published by Rosenblum and colleagues was used and modified by adding belt distance (BD) to the equation to account for the backward movement of the trailing limb from its initial contact, until the initial contact of the leading limb [15]. This modification was done to compensate for the lack of kinematic data (Fig 2).

[Figure omitted. See PDF.]

The algorithm used computes mean and standard deviation (SD) of step length and step width by using a moving window of 6 samples to generate two graphs representing mean and SD of every step [15]. To analyze recovery performance after perturbations, 20 steps prior to the perturbation were defined as baseline walking. With the help of the algorithm an overall sum of deviation (OSDev) is computed by scanning mean and SD graphs [15]. OSDev is used to determine the number of steps used to recover gait. For this purpose, the algorithm searches for the exact point, in which the change in amplitude has become sufficiently small compared to pre perturbation as this indicates that the performance returned to baseline values [15].

In contrast to Rosenblum’s approach, we used ten instead of twenty samples, which was due to differences in the perturbation protocol. Whereas Rosenblum and colleagues used at least thirty-three seconds [15], our protocol contained twenty to thirty seconds between perturbations. As a twenty samples window would be too large for individuals with a low cadence, ten samples were used instead. To avoid errors in automatic step detection, the step, which triggered the perturbation, was excluded from analysis. As it was likely that some participants would not return their gait pattern to baseline levels, a maximum of fifteen steps after perturbation was set in line with previous research [31]. The procedure was used for step length, step width and step length and step width combined (total recovery).

Statistical analysis

Participants’ baseline characteristics were analyzed using chi-square test for categorical variables, and independent t-test or, in case of non-parametric data, Mann-Whitney U test for continuous variables. Continuous variables were expressed as mean ± SD. Number of recovery steps were compared between the groups (OACH and OACI) using independent t-test. Between-group effect sizes were quantified using Cohen’s d with 0.2 representing a small effect, 0.5 a moderate sized effect, and 0.8 a large magnitude effect [32].

Given the exploratory nature of this sub-study, no a priori sample size calculation was performed. Instead, a convenience sample was analyzed in alignment with previous studies that have conducted similar investigations and included group sizes between eight to fifteen individuals [11, 15, 16].

Secondary analysis included a repeated measures of variance analysis (ANOVA). ANOVA was carried out with the within-subjects factor time (perturbation trial one and two) and the between subjects’ factor group (OACH and OACI) to explore differences between trials and groups. In case of significant time, group, or interaction effects, post hoc pairwise comparisons were conducted using Bonferroni correction for multiple pairwise comparisons. Due to the lack of an equivalent non-parametric analysis and due to ANOVA being relatively robust to violations of the normal distribution assumption [33] this analysis approach was used for all outcomes. Additionally, we investigated differences between perturbations by comparing the mean number of recovery of steps for both step length and step width averaged across groups and trials using the Kruskal-Wallis Test. In case of significant findings, we employed Bonferroni-corrected post hoc pairwise comparisons to determine differences between certain types of perturbations. Statistical significance was defined as p ≤ 0.05. Statistical analysis was conducted using SPSS for macOS (version 28.0; SPSS Inc., Chicago, IL, USA).

Results

Participant characteristics

For this exploratory sub-study full data sets of 36 participants were available, whose characteristics are reported in Table 1. Eighteen participants each were stratified into OACH and OACI group respectively. Besides cognitive function (p < 0.001), the proportion of males and females was different between the groups (p = 0.035). Further, the groups differed in treadmill walking speed with OACI walking more slowly than OACH (p = 0.012) but not in overground walking speed (p = 0.161).

[Figure omitted. See PDF.]

Recovery performance

Fig 3 shows the recovery performance of the two groups for total recovery steps as well as step length and step width across all nine perturbations. Whereas the number of recovery steps for step width did not differ between OACH (mean: 9.5, SD: 0.8) and OACI (mean: 9.7, SD: 0.8; p = 0.151), the number of total recovery steps (p = 0.045, Cohen’s d = 0.69) as well as the recovery of step length (p = 0.039, Cohen’s d = 0.72) were significantly different between the groups. OACI (total number of steps: mean: 9.8, SD: 0.9; step length: mean: 9.9, SD: 1.2) needed significantly more steps to return to their baseline than OACH (total number of steps: mean: 9.3, SD:0.6; step length: mean: 9.1, SD: 0.8).

[Figure omitted. See PDF.]

Table 2 presents a detailed breakdown of step length and step width recovery performance, categorized by perturbation type and trial. Secondary exploratory analysis unveiled significant time (p < 0.001) and between-group effects (p = 0.014) for step length recovery in the case of a trip to the right. While the overall recovery performance improved similarly in both groups between the first and second trials, individuals in the OACI group required significantly more steps than those in the OACH group to recover step length after a slip to their right foot. For step width, time effects were observed for trip right (p = 0.034), pitch up (p < 0.001), and full stop (p = 0.009), indicating a reduction in the number of recovery steps between the first and second trials across groups.

[Figure omitted. See PDF.]

S1 Table provides an overview of the average number of recovery steps required per perturbation averaged across groups and trials. Kruskal-Wallis Test revealed a significant difference for both step width (p<0.001) and step length (p<0.001). While step length differed between 8.1 steps (slip left) to 10.4 steps (trip left), step width differed between 8.5 steps for slip left and 11.0 steps for pitch up. Significant differences were evident among various perturbations as demonstrated in S1 Table.

Discussion

In this study, we disturbed gait in older adults, who had recently experienced a severe fall, walking on a perturbation treadmill with nine different types of perturbations. The participants represent a high-risk group for falls and functional decline as they were recruited after presenting to the ED after a fall. Our results support the hypothesis that OACI have a worse gait adaptability in comparison to OACH. Whereas recovery of step width did not differ between the groups, OACI needed more steps to return their step length back to pre-perturbation level than OACH.

Previous research indicated that single perturbations such as slips or trips do not cover the broad variety of perturbations that occur in everyday life [11, 28, 34, 35]. Therefore, the adequate adaptation of gait to different perturbations used in our study may present a more promising outcome to quantify a person’s reactive dynamic balance. Nonetheless, future studies are necessary to validate this approach, comparing the utilization of single or multiple types of perturbations and examining the potential association between individuals’ performance and future falls. This verification should encompass a prospective fall assessment. Moreover, future research should focus on investigating different types of perturbations to determine if certain ones pose greater challenges for individuals than others. This exploration could offer insights into specific training needs. Since perturbations were presented in a similar sequence, the observed recovery steps in our study may have been influenced by the order of their presentation, thereby limiting the generalizability of our results.

In our study, OACI had a poorer gait adaptability to balance perturbations, which was shown by the increased number of steps needed to return step length to baseline, in comparison to OACH. A poorer gait adaptability to balance perturbations in OACI is in line with previous research indicating differences in walking behavior between OACH and OACI [36–38]. As adapting step length is an important recovery mechanism to external perturbations [26, 39], immediate changes of an individual’s step length to perturbations were expected. However, the increased number of recovery steps needed by OACI may indicate less effective stepping responses, especially as Debelle and colleagues argue that a close to normal step length after a perturbation may present an important component of balance recovery [31]. However, previous results are controversial whether a close to normal, a shortened first step length (i.e., higher walking stability) or an elongated first step (i.e., better compensation for shifts in center of mass) may present better reactions to perturbations [26, 31, 39, 40]. Therefore, future research needs to target recovery mechanisms to different types and intensities of perturbations, to identify adequate recovery mechanisms and, thus, modify fall prevention efforts. According to the findings of our exploratory study, altering step length could be incorporated into falls prevention strategies to enhance gait adaptations, particularly among individuals with cognitive impairment.

The increased number of recovery steps in OACI in comparison to OACH may be due to slower cognitive processing and less cognitive flexibility, which is supported by the similarity in functional performance, as assessed with the short physical performance battery, across both groups. Impaired cognitive processing and less cognitive flexibility affect the ability to adapt behavior to external disturbances according to a previous review covering gait, cognition, and fall risks in older adults [41], which may explain findings in our study. In this context, our secondary, exploratory analysis appears promising as the observed time effects between trials one and two potentially indicate a trainability of gait adaptability to balance perturbations in both OACH and OACI. This warrants further investigation. As falls prevention interventions are not as effective in persons with cognitive impairment or dementia, although risk factors such as muscular strength can be adequately modified [42], dynamic reactive balance control might be an important target for falls prevention especially in these high-risk groups.

Strengths and limitations

The decision to exclusively use data from force plates was motivated by clinical practice due to the approach being less time and cost consuming. Furthermore, force plates are well-established to measure ground reaction forces and, thus, provide kinetic information during walking [43]. While force plates offer an efficient and straightforward way of data collection, they lack the ability to provide accurate kinematic information. Step length, being a key kinematic parameter, is not directly measured by force plates. As a result, the insights gained from this method may not be as detailed and accurate as those obtained from marker-based systems. Thus, the local maximum of the CoP in anterior-posterior direction was defined as initial contact as it was more robust in regard to an automatic detection of steps. Nevertheless, it is likely that a minor time deviation to the true initial contact occurred. As this deviation has been similarly for all participants and across all step detection measures, it is unlikely to have affected results of the present study.

Due to not using a marker-based system, participants had to keep walking after the perturbations. To better reflect real life scenarios, self-paced walking approaches, which automatically and continuously adjust speed in real-time to an individual’s walking speed, may be used in future research studies.

Sample size was in line with previous studies. However, longitudinal approaches and bigger samples are warranted to analyze the relationship between recovery steps and future falls in individuals with and without cognitive impairment. In such studies sex differences in gait adaptability to perturbations may be further analyzed, as differences in group distribution could not be looked at in our study. Large scale studies would further benefit from an extensive neuropsychological test battery to confirm a formal diagnosis of mild cognitive impairment as well as an a priori sample size calculation.

Conclusion

This study found moderate differences in gait adaptability to different perturbations in individuals with probable cognitive impairment when compared to cognitively healthy older adults using data from treadmill integrated force plates. Adaptation of step length was particularly challenging for individuals with probable cognitive impairment, which might be a link to providing effective falls prevention also for this group. As all participants had experienced an injurious fall about four weeks before their gait analysis, the approach used in this study may be valid to determine reactive balance performance in frail populations as well as in clinical practice. However, long-term observation and prospective fall assessments, as well as intervention trials are needed to support this notion.

Supporting information

S1 Table. Overview of the average number of recovery steps required per perturbation averaged across groups and trials.

https://doi.org/10.1371/journal.pone.0305067.s001

(DOCX)

Acknowledgments

The authors would like to thank all participants who have been involved and dedicated their precious time to this study. A special thanks goes to the participatory research team of the SeFallED study.

They would also like to acknowledge the “Klinikum Oldenburg” and the “Evangelisches Krankenhaus Oldenburg” in providing a dedicated environment to inform participants about the study and offer a chance for conducting clinical research.

Moreover, the authors would like to express their gratitude to all study nurses, particularly Sabine Ehlers and Ute von Varel, and student assistants, particularly Johanna Michel, involved in recruiting participants at the EDs and supporting data collection.

Citation: Voß M, Zieschang T, Schmidt L, Hackbarth M, Koschate J, Stuckenschneider T (2024) Reduced adaptability to balance perturbations in older adults with probable cognitive impairment after a severe fall. PLoS ONE 19(7): e0305067. https://doi.org/10.1371/journal.pone.0305067

About the Authors:

Malte Voß

Roles: Data curation, Formal analysis, Methodology, Writing – original draft

Affiliation: Department for Health Services Research, Geriatric Medicine, School of Medicine and Health Services, Carl von Ossietzky University, Oldenburg, Lower Saxony, Germany

Tania Zieschang

Roles: Conceptualization, Funding acquisition, Methodology, Writing – review & editing

Affiliation: Department for Health Services Research, Geriatric Medicine, School of Medicine and Health Services, Carl von Ossietzky University, Oldenburg, Lower Saxony, Germany

Laura Schmidt

Roles: Conceptualization, Methodology, Project administration, Writing – original draft, Writing – review & editing

Affiliation: Department for Health Services Research, Geriatric Medicine, School of Medicine and Health Services, Carl von Ossietzky University, Oldenburg, Lower Saxony, Germany

Michel Hackbarth

Roles: Data curation, Methodology, Software, Writing – review & editing

Affiliation: Department for Health Services Research, Geriatric Medicine, School of Medicine and Health Services, Carl von Ossietzky University, Oldenburg, Lower Saxony, Germany

ORICD: https://orcid.org/0000-0002-6367-5342

Jessica Koschate

Roles: Funding acquisition, Writing – review & editing

Affiliation: Department for Health Services Research, Geriatric Medicine, School of Medicine and Health Services, Carl von Ossietzky University, Oldenburg, Lower Saxony, Germany

Tim Stuckenschneider

Roles: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Project administration, Writing – original draft

E-mail: [email protected]

Affiliation: Department for Health Services Research, Geriatric Medicine, School of Medicine and Health Services, Carl von Ossietzky University, Oldenburg, Lower Saxony, Germany

ORICD: https://orcid.org/0000-0002-0972-4377

References

1. Tinetti ME, Speechley M, Ginter SF. Risk factors for falls among elderly persons living in the community. N Engl J Med. 1988;319(26):1701–7. pmid:3205267.

2. Berg WP, Alessio HM, Mills EM, Tong C. Circumstances and consequences of falls in independent community-dwelling older adults. Age Ageing. 1997;26(4):261–8. pmid:9271288.

3. Crenshaw JR, Bernhardt KA, Achenbach SJ, Atkinson EJ, Khosla S, Kaufman KR, et al. The circumstances, orientations, and impact locations of falls in community-dwelling older women. Arch Gerontol Geriatr. 2017;73:240–7. Epub 20170721. pmid:28863352; PubMed Central PMCID: PMC5858880.

4. Rupp M, Walter N, Pfeifer C, Lang S, Kerschbaum M, Krutsch W, et al. The Incidence of Fractures Among the Adult Population of Germany—and Analysis From 2009 through 2019. Dtsch Arztebl Int. 2021;(Forthcoming). Epub 20211008. pmid:34140088.

5. Thompson HJ, McCormick WC, Kagan SH. Traumatic brain injury in older adults: epidemiology, outcomes, and future implications. Journal of the American Geriatrics Society. 2006;54(10):1590–5. pmid:17038079; PubMed Central PMCID: PMC2367127.

6. Schoene D, Heller C, Aung YN, Sieber CC, Kemmler W, Freiberger E. A systematic review on the influence of fear of falling on quality of life in older people: is there a role for falls? Clin Interv Aging. 2019;14:701–19. Epub 2019/06/14. pmid:31190764; PubMed Central PMCID: PMC6514257.

7. Maki BE, Cheng KC, Mansfield A, Scovil CY, Perry SD, Peters AL, et al. Preventing falls in older adults: new interventions to promote more effective change-in-support balance reactions. J Electromyogr Kinesiol. 2008;18(2):243–54. Epub 20070904. pmid:17766146.

8. McCrum C, Bhatt TS, Gerards MHG, Karamanidis K, Rogers MW, Lord SR, et al. Perturbation-based balance training: Principles, mechanisms and implementation in clinical practice. Front Sports Act Living. 2022;4:1015394. Epub 20221006. pmid:36275443; PubMed Central PMCID: PMC9583884.

9. Oddsson LIE, Boissy P, Melzer I. How to improve gait and balance function in elderly individuals—compliance with principles of training. European Review of Aging and Physical Activity. 2007;4(1):15–23.

10. Taylor Z, Walsh GS, Hawkins H, Inacio M, Esser P. Perturbations during Gait: A Systematic Review of Methodologies and Outcomes. Sensors (Basel). 2022;22(15). Epub 20220808. pmid:35957484; PubMed Central PMCID: PMC9371403.

11. Gerards MHG, Meijer K, Karamanidis K, Grevendonk L, Hoeks J, Lenssen AF, et al. Adaptability to Balance Perturbations During Walking as a Potential Marker of Falls History in Older Adults. Front Sports Act Living. 2021;3:682861. Epub 20210519. pmid:34095828; PubMed Central PMCID: PMC8169998.

12. Konig M, Epro G, Seeley J, Potthast W, Karamanidis K. Retention and generalizability of balance recovery response adaptations from trip perturbations across the adult life span. J Neurophysiol. 2019;122(5):1884–93. Epub 20190911. pmid:31509470.

13. Handelzalts S, Kenner-Furman M, Gray G, Soroker N, Shani G, Melzer I. Effects of Perturbation-Based Balance Training in Subacute Persons With Stroke: A Randomized Controlled Trial. Neurorehabilitation and neural repair. 2019;33(3):213–24. Epub 20190215. pmid:30767613.

14. Barajas JS, Peterson DS. First-trial protective step performance before and after short-term perturbation practice in people with Parkinson’s disease. J Neurol. 2018;265(5):1138–44. Epub 20180308. pmid:29520471.

15. Rosenblum U, Kribus-Shmiel L, Zeilig G, Bahat Y, Kimel-Naor S, Melzer I, et al. Novel methodology for assessing total recovery time in response to unexpected perturbations while walking. PLoS One. 2020;15(6):e0233510. Epub 2020/06/04. pmid:32492029; PubMed Central PMCID: PMC7269230.

16. Kannan LN, Bhatt TS. Perturbation-based balance assessment: Examining reactive balance control in older adults with mild cognitive impairments. Physiol Int. 2021;108(3):353–70. Epub 20210916. pmid:34529584.

17. Montero-Odasso M, van der Velde N, Martin FC, Petrovic M, Tan MP, Ryg J, et al. World guidelines for falls prevention and management for older adults: a global initiative. Age Ageing. 2022;51(9). pmid:36178003; PubMed Central PMCID: PMC9523684.

18. Stuckenschneider T, Koschate J, Dunker E, Reeck N, Hackbarth M, Hellmers S, et al. Sentinel fall presenting to the emergency department (SeFallED)—protocol of a complex study including long-term observation of functional trajectories after a fall, exploration of specific fall risk factors, and patients’ views on falls prevention. BMC Geriatr. 2022;22(1):594. Epub 20220718. pmid:35850739; PubMed Central PMCID: PMC9289928.

19. Stuckenschneider T, Schmidt L, Speckmann EM, Koschate J, Zieschang T. Recruiting patients for falls prevention in the emergency department—worth the challenge. BMC Geriatr. 2023;23(1):880. Epub 20231221. pmid:38129767; PubMed Central PMCID: PMC10740331.

20. Nasreddine ZS, Phillips NA, Bédirian V, Charbonneau S, Whitehead V, Collin I, et al. The Montreal Cognitive Assessment, MoCA: a brief screening tool for mild cognitive impairment. Journal of the American Geriatrics Society. 2005;53(4):695–9. pmid:15817019

21. Thomann AE, Berres M, Goettel N, Steiner LA, Monsch AU. Enhanced diagnostic accuracy for neurocognitive disorders: a revised cut-off approach for the Montreal Cognitive Assessment. Alzheimer’s research & therapy. 2020;12(1):1–10. pmid:32264975

22. Kurz I, Gimmon Y, Shapiro A, Debi R, Snir Y, Melzer I. Unexpected perturbations training improves balance control and voluntary stepping times in older adults—a double blind randomized control trial. BMC Geriatr. 2016;16:58. Epub 2016/03/06. pmid:26944706; PubMed Central PMCID: PMC4778347.

23. McCrum C, Karamanidis K, Grevendonk L, Zijlstra W, Meijer K. Older adults demonstrate interlimb transfer of reactive gait adaptations to repeated unpredictable gait perturbations. Geroscience. 2020;42(1):39–49. Epub 2019/11/30. pmid:31776885; PubMed Central PMCID: PMC7031170.

24. Rieger MM, Papegaaij S, Pijnappels M, Steenbrink F, van Dieen JH. Transfer and retention effects of gait training with anterior-posterior perturbations to postural responses after medio-lateral gait perturbations in older adults. Clin Biomech (Bristol, Avon). 2020;75:104988. Epub 2020/03/17. pmid:32174482.

25. Rieger MM, Papegaaij S, Steenbrink F, Pijnappels M, van Dieen JH. Development of a Balance Recovery Performance Measure for Gait Perturbation Training Based on the Center of Pressure. Front Sports Act Living. 2021;3:617430. Epub 2021/03/05. pmid:33659894; PubMed Central PMCID: PMC7917114.

26. Roeles S, Rowe PJ, Bruijn SM, Childs CR, Tarfali GD, Steenbrink F, et al. Gait stability in response to platform, belt, and sensory perturbations in young and older adults. Med Biol Eng Comput. 2018;56(12):2325–35. Epub 2018/06/28. pmid:29946955; PubMed Central PMCID: PMC6245003.

27. Kirchner M, Schubert P, Haas CT. Characterisation of Real-World Bus Acceleration and Deceleration Signals. Journal of Signal and Information Processing. 2014;05(01):8–13.

28. Geerse DJ, Roerdink M, Marinus J, van Hilten JJ. Walking adaptability for targeted fall-risk assessments. Gait Posture. 2019;70:203–10. Epub 20190220. pmid:30901621.

29. Yu H, Sarkodie-Gyan T. Towards a Methodology for the Differential Analysis in Human Locomotion: A Pilot Study on the Participation of Individuals with Multiple Sclerosis. Engineering. 2012;04(10):20–6.

30. Horsak B, Slijepcevic D, Raberger AM, Schwab C, Worisch M, Zeppelzauer M. GaiTRec, a large-scale ground reaction force dataset of healthy and impaired gait. Sci Data. 2020;7(1):143. Epub 20200512. pmid:32398644; PubMed Central PMCID: PMC7217853.

31. Debelle H, Maganaris CN, O’Brien TD. Biomechanical Mechanisms of Improved Balance Recovery to Repeated Backward Slips Simulated by Treadmill Belt Accelerations in Young and Older Adults. Front Sports Act Living. 2021;3:708929. Epub 20210921. pmid:34622205; PubMed Central PMCID: PMC8490723.

32. Cohen J. The earth is round (p<. 05): Rejoinder. 1995.

33. Schmider E, Ziegler M, Danay E, Beyer L, Bühner M. Is It Really Robust? Methodology. 2010;6(4):147–51.

34. Te B, Komisar V, Aguiar OMG, Shishov N, Robinovitch SN. Compensatory stepping responses during real-life falls in older adults. Gait & Posture. 2023;100:276–83. pmid:36689855

35. Talbot LA, Musiol RJ, Witham EK, Metter EJ. Falls in young, middle-aged and older community dwelling adults: perceived cause, environmental factors and injury. BMC public health. 2005;5:86. Epub 20050818. pmid:16109159; PubMed Central PMCID: PMC1208908.

36. Montero-Odasso M, Verghese J, Beauchet O, Hausdorff JM. Gait and cognition: a complementary approach to understanding brain function and the risk of falling. Journal of the American Geriatrics Society. 2012;60(11):2127–36. Epub 20121030. pmid:23110433; PubMed Central PMCID: PMC3498517.

37. Bovonsunthonchai S, Vachalathiti R, Hiengkaew V, Bryant MS, Richards J, Senanarong V. Quantitative gait analysis in mild cognitive impairment, dementia, and cognitively intact individuals: a cross-sectional case-control study. BMC Geriatr. 2022;22(1):767. Epub 20220923. pmid:36151524; PubMed Central PMCID: PMC9502583.

38. Lindh-Rengifo M, Jonasson SB, Ullen S, Stomrud E, Palmqvist S, Mattsson-Carlgren N, et al. Components of gait in people with and without mild cognitive impairment. Gait Posture. 2022;93:83–9. Epub 20220117. pmid:35101749.

39. Ren X, Lutter C, Kebbach M, Bruhn S, Bader R, Tischer T. Lower extremity joint compensatory effects during the first recovery step following slipping and stumbling perturbations in young and older subjects. BMC Geriatr. 2022;22(1):656. Epub 20220810. pmid:35948887; PubMed Central PMCID: PMC9367084.

40. Yang F, Kim J, Munoz J. Adaptive gait responses to awareness of an impending slip during treadmill walking. Gait Posture. 2016;50:175–9. Epub 20160913. pmid:27632061.

41. Zhang W, Low LF, Schwenk M, Mills N, Gwynn JD, Clemson L. Review of Gait, Cognition, and Fall Risks with Implications for Fall Prevention in Older Adults with Dementia. Dement Geriatr Cogn Disord. 2019;48(1–2):17–29. Epub 20191119. pmid:31743907.

42. Racey M, Markle-Reid M, Fitzpatrick-Lewis D, Ali MU, Gagne H, Hunter S, et al. Fall prevention in community-dwelling adults with mild to moderate cognitive impairment: a systematic review and meta-analysis. BMC Geriatr. 2021;21(1):689. Epub 20211210. pmid:34893027; PubMed Central PMCID: PMC8665555.

43. Hong CY, Guo LY, Song R, Nagurka ML, Sung JL, Yen CW. Assessing postural stability via the correlation patterns of vertical ground reaction force components. Biomed Eng Online. 2016;15(1):90. Epub 20160802. pmid:27485525; PubMed Central PMCID: PMC4969977.