- Journal List
- Karger Author's Choice
- PMC11239142
As a library, NLM provides access to scientific literature. Inclusion in an NLM database does not imply endorsem*nt of, or agreement with, the contents by NLM or the National Institutes of Health.
Learn more: PMC Disclaimer | PMC Copyright Notice
Gerontology. 2024 Jul; 70(7): 689–700.
Published online 2024 Apr 24. doi:10.1159/000535968
PMCID: PMC11239142
PMID: 38657580
Shani Batcir,a Yuliya Berdichevsky,a Yaacov G. Bachner,b Omri Lubovsky,c Ronen Debi,c and Itshak Melzer
a
Author information Article notes Copyright and License information PMC Disclaimer
Associated Data
- Supplementary Materials
- Data Availability Statement
Abstract
Introduction
An effective reactive step response to an unexpected balance loss is an important factor that determines if a fall will happen. We investigated reactive step strategies and kinematics of unsuccessful balance recovery responses that ended with falls in older adults.
Methods
We compared the strategies and kinematics of reactive stepping after a lateral loss of balance, i.e., perturbations, between 49 older female adults who were able to successfully recover from perturbations (perturbation-related non-fallers, PNFs) and 10 female older adults who failed to recover (perturbation-related fallers, PFs). In addition, we compared the successful versus unsuccessful recovery responses of PFs matched to perturbation magnitude.
Results
The kinematics of the first reactive step response were significantly different between PFs and PNFs, i.e., longer initiation time, step time, swing time, and time to peak swing-leg velocity, larger first-step length, and center-of-mass displacement. Incomplete crossover stepping and leg collision were significant causes of falls among PFs. Similar findings were found when we compared the successful versus unsuccessful recovery responses of PFs.
Conclusions
The crossover step, which requires a complex coordinated leg movement, resulted in difficulty in controlling and decelerating the moving center of mass following a lateral perturbation, affecting the kinematics of the stepping response, leading to a fall.
Keywords: Older adults, Falls, Perturbations, Loss of balance, Reactive step strategies, Reactive step kinematics
Introduction
Falls among older adults result in serious injury in 20–30% of the time [1], and 2–5% of falls lead to hospital visits [2]. One of the tremendous consequences of falls is hip fractures [3], mainly resulting from sideway falls that result in landing on the greater trochanter [4]. Hip fractures rank as the leading cause of trauma-related hospitalizations and are a top ten cause of death [5]. If balance is challenged unexpectedly, i.e., by a postural perturbation, a recovery step response is the most effective strategy for preventing a fall [6–10]. These step responses are very common reactions to the loss of balance in daily life in older adults [11–13]. An ineffective reactive stepping response may result in a fall, which is a considerable health problem among older adults as it compromises their well-being and health [5]. While falls impact nearly one-third of the elderly population [1, 5], they pose a challenging subject for investigation due to their infrequent occurrence, typically happening only once or twice a year for most older individuals [5].”
An important question to answer in fall and injury prevention is what goes wrong during an unsuccessful balance recovery, leading to an unexpected loss of balance and fall. Past investigations of how older adults fall are usually based on witness reports or self-reports of fall events, which may be imprecise [14]. Another way to investigate falls are by studies that utilize video-cameras that capture fall events in public areas at long-term care sites or nursing homes [13, 15, 16]. Such studies help investigate how older adults fall in long-term care, i.e., of relatively frail older adults. These observations are a practical method to better understand which people fall, what the people actually did when they fell, and how they fell within a specific environment [13, 15, 16]. Video capture of falls has been used to describe the cause of imbalance, activity at the time of falling, protective responses, and the location where impact occurs [11, 13]. Although video-based investigations improve researchers’ capacity to comprehend the origins of falls, there are limitations associated with this method in fall studies. The primary constraints are particularly regarding the control over the environment and the examination of balance recovery kinematics during a fall and that video-based investigations where conducted among older adults who live in long-term care settings and not independent community-dwelling older adults.
In the current investigation, we targeted laboratory-induced sideway fall events. We aimed to compare: (1) the strategies and kinematic characteristics of unsuccessful balance recovery versus successful recovery among older adults who failed to respond effectively (PFs) versus older adults who were able to effectively prevent a fall (PNFs); and (2) the strategies and kinematic characteristics of the PFs’ unsuccessful balance recoveries versus their own successful balance recoveries. Our experimental setup involved perturbation in standing since a sizable proportion of falls occur during static or near-static conditions engagements and activities (40–50%) [11, 12], e.g., standing quietly (13% falls) and sitting down (12% falls) [13]. Moreover, a loss of balance during standing shares many fundamental control subtasks with gait (e.g., step initiation, fast swing-phase duration, appropriate placement of the swing foot, stabilization of the CoM during swing, etc). We hypothesized that compared to successful balance recovery trials, the unsuccessful balance recovery trials would present a less effective stepping strategy, especially in terms of lower abilities to perform a fast first reactive step and the ability to control the center-of mass (CoM) motion over the base of support (BoS) provided by the feet. This may offer a potential rationale for the biomechanical factors contributing to falls, potentially enhancing the development of treatment strategies.
Methods
Participants
A convenience sample of 84 older adults (59 females and 25 males) volunteered to participate in the study. Volunteers were community-dwelling older adults who were recruited using flyers and advertisem*nts for participation in a one-time perturbation session evaluation. They were 70 years of age or older, independent in daily living activities and were excluded if they had any of the following: (1) blindness or serious vestibular impairments (Meniere’s disease, dizziness); (2) an inability to ambulate independently; (3) a score <24 on the Mini-Mental State Examination (MMSE); (4) symptomatic severe cardiovascular disease; (5) neurological disorders such as stroke, Parkinson’s, or Multiple sclerosis; (6) orthopedic acute disorders requiring total hip or knee replacement; (7) severe rheumatoid arthritis; and/or (8) cancer (metastatic or under active treatment).
Assessment Protocol
After approval by the Helsinki Ethics Committee of Barzilai Medical Center in Ashkelon, Israel (ClinicalTrials.gov registration number #NCT01439451, initial release September 23, 2011), participants signed a consent form. They then stood with their heels and toes touching on a motor-driven perturbation treadmill device, i.e., the Balance Measure and Perturbation system, BaMPer [17]. The participants were exposed to a total of 13 right and 13 left unannounced surface translation perturbations that systematically increased from low magnitude (No. 1) to high magnitude (No. 13). The timing and direction of the unannounced perturbation were randomized. The instructions given to the participants were to: “React naturally and try to avoid falling.” Participants were able to stop the experiment or rest at any point. The magnitudes of the perturbations were specified in terms of transverse motion in centimeters (cm) of the BaMPer, timing in milliseconds (ms), velocity in cm/ms, and acceleration in cm/ms2 (see online suppl. Table 1; for all online suppl. material, see https://doi.org/10.1159/000535968). The participants wore a safety harness that did not hamper arm or leg movements and prevented them from falling on the ground. Of 59 females, 10 female participants were unable to successfully recover from laboratory-induced loss of balance during the experiment (perturbation-related fallers, PFs); thus, we decided to exclude the 25 male older adults from the analyses and to include only the 49 females that did not fell during the experiment as a control group (perturbation-related non-fallers, PNFs). A laboratory-induced fall was defined observationally in the following cases: if the safety harness stretched and restrained the participants to prevent a fall, the participant grasped the research assistant, or was caught by the assistant before falling. Later, we verified each fall incident by 3D-kinematic data (details below) showing that the participant’s CoM traveled out of the boundaries of the feet, i.e., the BoS, at the end of the balance response (selected cases in Fig. 1).
Fig. 1.
Examples and kinematic analysis of the unsuccessful balance recovery falls for older adults following unexpected lateral perturbations: (a) grasping the research assistance; (b) stretched harnesses restrained the reaction; (c) caught by the research assistant; and (d) falling into harness. Top graphs of a, b, c, and d – the black solid line represents the platform perturbation displacement in cm in the mediolateral direction. Bottom graphs of a, b, c, and d – the green solid line represents movement of the left foot, the red solid line represents movement of the right foot, and the solid purple line represents movement of the body’s center of mass in the mediolateral direction; Dotted vertical lines represent start or end time points. The following events are marked: (A) platform perturbation onset – detected as the initiation of horizontal platform motion in the mediolateral direction (dotted black line); (B) end of platform perturbation (2nd dotted black line); (C) first step initiation (e.g., foot off the ground) – defined as the initial vertical displacement of the marker that was placed on the swinging ankle (dotted blue line); (D) end of first step – defined as the displacement of the swinging ankle marker, resulting from placing the foot on the ground completing the first step (2nd dotted blue line).
Data Analysis
Step Strategies
We identified the presence of first-step strategies using Windows Media Player which allowed image pauses, slow motion, and the video to be run back and forth, as described in detail in [9, 10]. The following are the classifications of the first-step recovery strategies: loaded-leg sideways stepping – performing the first step with the loaded leg; unloaded-leg sideways stepping (ULSS) – performing the first step with the unloaded leg; crossover stepping (CoS) – stepping with the unloaded leg, swinging the leg over the loaded leg; incomplete CoS – stepping with the unloaded leg, but unable to swing the leg over the loaded leg to complete the crossover step; leg collisions (Col) – collision between the swinging leg and the loaded leg during CoS; and hip abduction (Abd) – abducting the unloaded leg. Interobserver reliability was excellent, as described in detail [9, 10].
Kinematic Analyses of Stepping
The following kinematic parameters of the first-recovery step were extracted by 3D-kinematic data that were acquired through the Ariel Performance Analysis System (APAS, Ariel Dynamics Inc.; CA, USA) a video-based 3D motion analysis system that can capture video from two cameras simultaneously and perform a biomechanical analysis automatically. The two cameras were mounted at a 45° angle between each camera and the subject’s standing position, at a height of 2.5 m and about 7 m in front of the perturbation system. The two cameras simultaneously recorded the motion of 8 reflective markers with a sampling frequency of 60 Hz. The markers were placed at (1) the anterior midpoint of the ankle joints, (2) anterior superior iliac spines, (3) acromion processes, and (4) radial styloid processes. Views from both cameras were mapped onto a 3D coordinate system using an internal direct linear transformation algorithm. The data were also grabbed and a full-test video clip recorded each subject for observational analysis using Windows Media Player, allowing image pauses, slow motion, and running it back and forth. For kinematic analysis, perturbation trials that were followed by execution of recovery stepping were also digitized, transformed, and smoothed using low-pass filter (Butterworth second order forward and backward passes) with a cut-off frequency of 5 Hz. This approach was shown to be valid and reliable, i.e., mean point estimate error of less than 3.5 mm, 1.4 mm mean linear error, and 0.26° mean angular error (Klein and DeHaven, 1995). Than we used a program written especially for this project in C# (Microsoft 2000) that were able to extract the following events (more details [9, 10]): (1) first-step initiation time (ms) – the time from the perturbation to lifting the stepping foot off the ground; (2) first-step duration (ms) – the time from the perturbation to completion of the first-recovery step; (3) first step length (cm) – the distance traveled by the foot from foot off the ground to foot contact; (4) swing-phase duration (ms) – the time from lifting the stepping foot off the ground to foot contact; (5) time to peak velocity of the swinging leg that performed the first stepping response to recover balance (ms); and (6) estimated CoM (eCoM) path displacement (cm) – the distance the CoM traveled from the initial point prior to stepping to the point that the foot contacted the ground, completing the first-recovery step. A large eCoM displacement path reflects less ability to control the CoM.
Since participants performed more than one step in an attempt to recover their balance during both the successful and unsuccessful trials, the following kinematic parameters of the extra steps were extracted: (1) the total balance recovery response time – the time from perturbation onset to either a fall into the harness system/the harness system stretched or until foot contact of the ground completing a successful balance recovery when extra steps were needed, (2) total step path length – the Euclidian distance in cm that the foot displaced from first-step initiation to either a fall into the harness system/harness system stretched or foot contact of the ground even when extra steps were needed completing a successful balance recovery, and (3) total eCoM displacement – the distance in cm that the CoM traveled from perturbation onset to either the time point of a fall into the harness system/harness system stretched or foot contact the ground completing a successful balance recovery. Age, MMSE [18], height, weight, body mass index (BMI), medical and fall history, Fall Efficacy Scale (FES-I) [19], and the Late Life Function and Disability Instrument (LLFDI) [20] were also assessed.
Sample Size Estimation
Our prior data [10] indicate that the difference in the first step initiation duration, the first step duration, and the total time to recover balance between older adults who reported 2 falls or more versus older adults who reported no falls in the past year. The data were normally distributed with standard deviation 95 ms, 147 ms, and 323 ms, respectively, with a difference in the mean the step initiation duration is 104 ms, 150 ms, and 324 ms, respectively. Based on the data, we calculated that we would need to study 9–10 pairs of subjects to be able to reject the null hypothesis that this response difference is zero with probability (power) 0.8. The type I error probability associated with this test of this null hypothesis is 0.05.
Statistical Analysis
Data were analyzed using PASW, version 26.0 (Somers, NY, USA). The Shapiro-Wilk statistic was first calculated to test for normality of distributions. Characteristics of PFs and PNFs were compared using t tests (age, weight, height, BMI), a Mann-Whitney U test (number of diagnosed diseases, number of medications per day, falls in the last year, and MMSE, FES-I, and LLFDI scores), and a χ2 test (yes/no falls in the last year).
We performed χ2 and Fisher’s exact tests to compare the frequencies of first-step strategies of PFs’ unsuccessful balance recovery trials to the matched successful trials of PNFs. In addition, we performed Mann-Whitney U tests to compare the kinematics of unsuccessful balance recovery trials of PFs to the matched successful trials of PNFs. The matching method of the PNFs’ successful trials were based on our previous findings showing the first-step leg strategies change with increasing perturbation magnitudes, and the kinematics of reactive stepping are highly affected, respectively, to the size of the effect, by perturbation magnitude, whether a single- or multiple-step response occurred, and also by the first-step leg strategy was performed [9, 10, 21]. Therefore, to investigate potential biomechanical factors contributing to falls we control these factors by matching. We firstly characterized the unsuccessful trial of PFs in terms of perturbation magnitude, step response type (single- or multiple-step), and the first-step strategy. Then we matched them with PNFs’ successful recovery trials performed in the same perturbation magnitude, same step response type (single- or multiple-step), and in most cases with the same first-step leg strategy was taken. In few unsuccessful-trial cases, where leg collision during CoS and incomplete CoS were performed, due to the absence of these first-step leg strategies among PNFs under the same specific conditions, we matched them with successful CoS trails. Four successful recovery trials of PNFs matched to each unsuccessful trial of PFs in a ratio of 4:1. The p value was set to 0.05 and adjusted for multiple comparisons (0.05/9 = 0.0056).
We also performed within-subjects analysis to compare the kinematics of unsuccessful balance recovery trials (i.e., fall trials) versus perturbation magnitude-matched successful trials among the PFs using Wilcoxon signed-rank tests. χ2 were used to compare the first-step strategies and number of steps to complete balance recovery. The p value was set to 0.05 and adjusted for multiple comparisons (0.05/9 = 0.0056).
Results
Table 1 shows that there were no significant differences between PFs and PNFs in age, cognitive function (MMSE), health, height, weight, BMI, and upper extremity function scores. However, PFs made up a higher proportion of those who reported past falls (p = 0.017), and they had higher FES-I scores (p = 0.048), and lower self-reported overall function scores (p = 0.026) and self-reported basic lower extremity function scores (p = 0.014).
Table 1.
Demographic characteristics of perturbation-related female fallers (PFs) versus perturbation-related female non-fallers (PNFs)
| Characteristic | PF (N = 10) | PNF (N = 49) | p value | |
|---|---|---|---|---|
| Age, years | 79.35±4.35 | 78.71±5.40 | t = −0.35 | 0.728 |
| Reported a fall in last year, % (n) | 80.0 (n = 8) | 38.7 (n = 19) | χ2 = 5.67 | 0.017 |
| Mini-Mental State Examination | 29 (26, 30) | 29 (25, 30) | z = −1.34 | 0.178 |
| Number of diseases diagnosed | 2 (0, 4) | 1 (0, 8) | z = 0.37 | 0.465 |
| Number medications/day | 3.5 (1, 7) | 4 (1, 10) | z = −0.30 | 0.759 |
| Height, cm | 153.40±6.09 | 156.36±8.04 | t = 1.10 | 0.276 |
| Weight, kg | 61.90±8.87 | 65.95±13.67 | t = 0.89 | 0.373 |
| BMI, kg/m2 | 26.26±3.01 | 29.94±4.44 | t = 0.46 | 0.644 |
| Fall Efficacy Scale (FES-I) | 29 (16, 58) | 19 (16, 46) | z = 1.97 | 0.048 |
| Late Life Function | ||||
| Overall function | 53.65 (50.32, 100) | 65.02 (45.22, 79.35) | z = −2.21 | 0.026 |
| Upper extremity function | 71.19 (57.64, 100) | 77.50 (51.27, 100) | z = −1.00 | 0.324 |
| Basic lower extremity function | 61.84 (54.05, 100) | 81.17 (53.35, 100) | z = −2.43 | 0.014 |
| Advanced lower extremity function | 47.25 (33.12, 100) | 57.00 (34.59, 81.36) | z = −1.92 | 0.054 |
Open in a separate window
Values are means ± 1 SD where a variable was normally distributed using an independent t test (age, weight, height, and BMI), and medians (Min, Max) where a variable was not normally distributed using a Mann-Whitney U test (Mini-Mental State Examination score, number of diagnosed diseases, number of medications per day, Fall Efficacy Scale, and Late Life Function scores), and χ2 for number of retro-fallers. Significant differences between PFs and PNFs (p value <0.05).
t indicates group comparison based on t test; z indicates group comparison based on Mann-Whitney U test, and χ2 indicates group comparison based on χ2 testing.
A total of 1,389 balance recovery trials were performed by 59 female participants (i.e., 1,171 recovery trials by 49 PNFs and 218 by 10 PFs). In 637 trials, participants performed a change of support strategy to recover their balance. PFs performed 111 stepping responses (50.9% of their perturbation trials) versus 526 of PNFs (44.9% of their perturbation trials; χ2 = 2.66, p = 0.102). We identified 18 trials (16.2% of their perturbation trials) where the 10 participants (PFs) were unable to recover balance and fell (4 PFs fell twice or more during the experiment). The following analyses are based on a comparison between the 18 unsuccessful trials of 10 PFs versus 72 successful balance recovery trials of the 49 PNFs, matched by perturbation magnitudes, step response type (single- or multiple-step response) and stepping strategy.
Step Strategies in Balance Recovery of PFs versus PNFs
Table 2 shows the differences between strategies used by the PFs and PNFs. The unloaded leg strategies (i.e., ULSS, CoS, and leg Abd) were performed in all of the unsuccessful recovery trials and successful-matched trials. Both PFs and PNFs initiated a CoS response an equal number of times (61.1% vs. 54%, respectively, χ2 = 0.29, p = 0.589). However, 11 out of 18 (61.1%) of the first-step strategies during the unsuccessful recovery trials were ineffective CoS (i.e., 9 incomplete CoS and 2 leg collisions) compared to 22 (30.5%) ineffective CoS balance recovery steps taken during the successful recovery trials (χ2 = 5.74, p = 0.016). Effective complete CoS was seen only among PNFs (23.5% vs. 0% in PFs, χ2 = 5.15, p = 0.023). Leg collisions were not observed during the successful trials of PNFs.
Table 2.
Strategies of the first-recovery stepping response of 18 trials of 10 PFs during their unsuccessful balance recovery trials versus the 72 matched successful balance recovery trials of 49 PNFs
| First-step strategies | Unsuccessful recovery trials (N = 18) | Successful recovery trials (N = 72) | p value |
|---|---|---|---|
| strategy/18 (%) | strategy/72 (%) | ||
| ULSS | 6 (33.3) | 29 (40.5) | 0.577 |
| LLSS | – | – | 1.000 |
| Complete CoS | 0 (0) | 17 (23.5) | 0.023 |
| Leg collisions | 2 (11) | 0 (0) | 0.004 |
| Incomplete CoS | 9 (50) | 22 (30.5) | 0.121 |
| Ineffective CoS (i.e., incomplete CoS and leg collisions) | 11 (61.1) | 22 (30.5) | 0.016 |
| Leg abduction | 1 (5.5) | 4 (5.5) | 1.000 |
Open in a separate window
Comparisons of frequencies of step recovery strategies between PFs and PNFs were made using χ2 and Fisher’s exact tests.
LLSS, loaded leg sidestep; ULSS, unloaded-leg sidestep; CoS, crossover stepping.
Kinematic Analysis of Recovery Step Responses of PFs versus PNFs
Table 3 shows a comparison between the unsuccessful balance recovery trials of PFs and the successful balance recovery ones of PNFs matched by perturbation magnitude, stepping strategy, and step response type (single- or multiple-step). Table 3A shows that compared to PNFs, PFs show a longer first-step initiation time (283 ms vs. 350 ms, p = 0.004), longer swing time duration (367 ms vs. 559 ms, p < 0.001), longer time to complete the first step (651 ms vs. 976, p < 0.001), longer time to peak velocity of the swinging foot (467 ms vs. 709 ms, p < 0.001), larger displacement of the eCoM (7.52 cm vs. 16.87 cm, p < 0.001), and a tendency to take a larger first-recovery step (9.27 cm vs. 16.71 cm, p = 0.008). In addition, the kinematics of the total balance recovery response when extra steps were needed to recover balance significantly differs between the unsuccessful recovery trials of PFs and the successful recovery trials of PNFs (Table 3B).
Table 3.
Between-group kinematic differences of 18 falling trials of perturbation-related fallers (PFs) versus 72 matched successful recovery trials of perturbation-related non-fallers (PNFs)
| Spatiotemporal parameters | Unsuccessful recovery trials of PF | Successful recovery trials of PNF | p value |
|---|---|---|---|
| A. First-recovery step | |||
| Step-initiation duration, ms | 350 (233, 868) | 283 (133, 701) | 0.004* |
| Swing-phase duration, ms | 559 (317, 1,436) | 367 (116, 734) | <0.001* |
| Step duration, ms | 976 (567, 1,753) | 651 (317, 1,436) | <0.001* |
| Step length, cm | 16.71 (3.17, 47.87) | 9.27 (1.39, 92.42) | 0.008 |
| Time to peak swinging-foot velocity, ms | 709 (450, 1,133) | 467 (233, 1,518) | <0.001* |
| eCoM path displacement, cm | 16.87 (6.36, 51.91) | 7.52 (4.18, 23.68) | <0.001* |
| B. Total stepping response | |||
| Total step duration, ms | 1,837 (918, 5394) | 868 (517, 2338) | <0.001 |
| Total step length, cm | 37.43 (5.55,103.7) | 18.01 (1.56, 96.57) | 0.006 |
| Total eCoM path displacement, cm | 32.42 (10.84, 66.65) | 10.56 (4.18, 50.58) | <0.001 |
Open in a separate window
Values are median (Min, Max).
Comparisons based on Mann-Whitney U tests.
eCoM, estimated center of mass.
*Significant differences for the recovery step parameters set on p = 0.05/9 = 0.0056.
Step Strategies of Unsuccessful Balance Recovery versus Successful Balance Recovery of PFs
Table 4 shows the strategies performed in the 18 unsuccessful balance recovery trials compared to 18 successful perturbation magnitude and strategies-matched balance recovery trials among the PFs. The number of recovery steps performed during the unsuccessful trials were significantly higher compared to the successful trials (2 vs. 1, p = 0.018). The unloaded-leg strategies (i.e., ULSS, CoS, and Abd) were performed in 100% of the unsuccessful recovery trials and in 88.8% of the successful-matched trials (χ2 = 2.059, p = 0.151). Eleven out of 18 (61.1%) of the first-step strategies during the unsuccessful recovery trials were incomplete CoS, including two events of leg collisions, compared to four complete CoS balance recovery steps (22.2%) during the successful recovery trials (χ2 = 5.448, p = 0.019). Leg collisions were not observed during the successful trials.
Table 4.
Strategies of the first-recovery stepping response of 18 unsuccessful balance recovery trials versus the 18 successful perturbation magnitude-matched balance recovery trials in PFs
| Subject number | Unsuccessful recovery trials (N = 18) | Successful recovery trials (N = 18) | ||||||
|---|---|---|---|---|---|---|---|---|
| perturbation magnitude | direction | first step strategy | steps, n | perturbation magnitude | direction | first step strategy | steps, n | |
| 4 | 10 | Left | ULSS | 1 | 10 | Right | ULSS | 1 |
| 11 | Left | Incomplete CoS | 2 | 11 | Right | ULSS | 1 | |
| 12 | Right | ULSS | 1 | 11 | Right | ULSS | 1 | |
| 12 | Left | Incomplete CoS | 2 | 11 | Right | ULSS | 1 | |
| 6 | 9 | Left | Incomplete CoS | 1 | 9 | Right | ULSS | 1 |
| 13 | Right | Collision in CoS | 5 | 13 | Left | COS | 1 | |
| 13 | 6 | Right | Incomplete CoS | 1 | 5 | Right | ULSS | 1 |
| 7 | Left | ULSS | 1 | 6 | Left | ULSS | 1 | |
| 7 | Right | ULSS | 2 | 6 | Left | ULSS | 1 | |
| 18 | 12 | Left | ULSS | 5 | 12 | Right | ULSS | 1 |
| 30 | 9 | Left | ULSS | 5 | 9 | Right | ULSS | 1 |
| 47 | 8 | Left | Incomplete CoS | 2 | 8 | Right | COS | 3 |
| 52 | 10 | Right | Abd | 2 | 10 | Left | Abd | 1 |
| 54 | 9 | Right | Incomplete CoS | 4 | 9 | Left | LLSS | 2 |
| 62 | 7 | Left | Incomplete CoS | 2 | 8 | Left | COS | 2 |
| 81 | 9 | Right | Incomplete CoS | 5 | 7 | Left | COS | 2 |
| 9 | Left | Incomplete CoS | 2 | 11 | Right | LLSS | 3 | |
| 10 | Left | Collision CoS | 2 | 10 | Right | ULSS | 2 | |
| Median (min, max) | 2 (1, 5) | 1 (1, 3) | ||||||
Open in a separate window
Number of steps defined as the number of steps needed to recover balance from perturbation onset to either a fall into the harness system/harness system stretched or until foot contact of the last successful reactive step.
PFs, perturbation-related fallers; ULSS, unloaded sidestep; LLSS, loaded-leg sidestep; CoS, crossover stepping; Abd, leg abduction.
Kinematic Analysis of Recovery Step Responses in Unsuccessful Balance Recovery versus Successful Balance Recovery of PFs
Table 5A shows no statistically significant difference in the first-step initiation duration between unsuccessful and successful step recovery trials of PFs (350 ms vs. 367 ms, p = 0.532). The swing-phase duration, the first-step length, and first-step eCoM displacement were significantly higher in the unsuccessful recovery trials than in the successful ones (559 ms vs. 400 ms, p = 0.004; 16.71 cm vs. 8.62 cm, p = 0.003; and 16.87 cm vs. 7.56 cm, p < 0.001, respectively). The first-step duration and the time to peak velocity of the swinging foot tended to be longer in the unsuccessful recovery trials compared to the successful trials (976 ms vs. 743 ms, p = 0.010; 709 ms vs. 617 ms, p = 0.011, respectively). Table 5B shows statistically significant differences in the kinematics of the total balance response between the unsuccessful and successful recovery trials, i.e., the total duration of the stepping response and total eCoM path displacement (1,837 ms vs. 818 ms, p = 0.010; 37.43 cm vs. 10.60 cm, p < 0.001; and 32.42 cm vs. 8.11 cm, p < 0.001, respectively), while total step length tended to be longer (37.43 cm vs. 10.60 cm, p < 0.001).
Table 5.
Kinematics of the first-recovery step (A) and the total stepping response (B) of the 18 unsuccessful recovery trials versus the 18 perturbation-magnitude matched successful balance recovery trials of PFs
| Spatiotemporal parameters | Unsuccessful recovery trials | Successful recovery trials | p value |
|---|---|---|---|
| A. First-recovery step | |||
| Step-initiation duration, ms | 350 (233, 868) | 367 (250, 868) | 0.532 |
| Swing-phase duration, ms | 559 (317, 1,436) | 400 (183, 1,402) | 0.004* |
| Step duration, ms | 976 (567, 1,753) | 743 (551, 1,770) | 0.010 |
| Step length, cm | 16.71 (3.17, 47.87) | 8.62 (1.90, 30.65) | 0.003* |
| Time to peak swinging-foot velocity, ms | 709 (450, 1,133) | 617 (300, 1,051) | 0.011 |
| eCoM path displacement, cm | 16.87 (6.36, 51.91) | 7.56 (5.09, 15.36) | <0.001* |
| B. Total stepping response | |||
| Total step duration, ms | 1,837 (918, 5,394) | 818 (551, 2,638) | 0.010 |
| Total extra-step path length, cm | 37.43 (5.55, 103.7) | 10.60 (1.90, 53.92) | <0.001* |
| Total eCoM path displacement, cm | 32.42 (10.84, 66.65) | 8.11 (5.09, 23.54) | <0.001* |
Open in a separate window
Values are median (Min, Max). Comparisons between the 18 falling trials and 18 perturbation-level non-falling matched trials within the PFs based on Wilcoxon signed-rank tests.
PFs, perturbation-related fallers; eCoM, estimated center of mass.
*Significant differences for the stepping response parameters set on p = 0.05/9 = 0.0056.
Discussion
In the present investigation, we compared the kinematics and strategies of the unsuccessful balance recovery of PFs to the successful balance recovery of PNFs matched to perturbation magnitude, stepping strategies, and single-/multiple-step response. Additionally, we compared the successful balance recovery of PFs versus their own unsuccessful balance recovery events at a similar perturbation magnitude. This provided an opportunity to explore what went wrong in the kinematics and strategies of laboratory-induced falls.
Comparison of Balance Recovery in PFs versus PNFs
The almost 24% slower step initiation duration, i.e., step reaction time, in PFs versus PNFs (350 ms vs. 283 ms, Table 3A) can be explained by several physiological factors that include: (a) a higher sensory detection level, (b) reduced sensory and motor nerve conduction velocity, and (c) longer CNS processing time. A recent study found a higher sensory detection level among fallers who showed a poorer foot tactile sensory threshold than non-fallers [22]. Tuerk et al. [23] found fall concern among older adults to be negatively associated with brain volumes in areas important for motor and emotional control, which suggests that PFs have a lower ability to process information quickly. These two findings suggest that a reduced sensory detection level and slower CNS processing time may be associated with slower step initiation time. The PFs in our study also showed a higher fall concern (i.e., higher FES-I) compared to the PNFs. Based on Tuerk et al.’s [23] findings, this may further explain why PFs exhibited a delay in step initiation time.
In regard to the ability to execute the first-recovery step quickly, during the unsuccessful balance trials, compared to the PNFs, the PFs demonstrated a significantly longer swing-phase duration (52%), as well as a 52% longer time to peak velocity of the swinging leg, resulting in 50% longer time to complete the first-recovery step (651 ms vs. 976 ms). This suggests that during the unsuccessful trials, the PFs showed a lower ability to recruit lower limb muscle motor units quickly, i.e., they exhibited less muscle power. We define muscle power as the ability to produce muscle strength quickly, indicating the rate of motor performance [24]. Because loss of balance occurs immediately after a perturbation, recovery stepping must be performed quickly in order to capture and reverse a fall. The ability to produce this force and execute a rapid step is advantageous. Several investigations support this concept [25–30]. Past investigations also found associations between muscle power and falls [31]. For example, the 5-Sit-to-Stand test quantifies leg muscle power [32, 33] and predicts fall risk [26, 28, 29, 34, 35], and it can predict older recurrent fallers [34, 35]. There is evidence showing that the rate of torque development in non-fallers is greater than in fallers [36]. A different study showed that fallers demonstrated 19% lower peak torque and 29% longer motor time [37]. Our results suggest that these effects on muscle physiology could likely explain most, if not all, of the reduction in the step execution performance seen in PFs. Additionally, a result that supports the notion that PFs have less ability to quickly recruit muscle force is the larger eCoM path displacement in the first-recovery steps observed in the unsuccessful balance recovery trials. The less ability to recruit muscle force fast enough is attributed to the inability of the PFs to decelerate the moving CoM over the BoS. This was accompanied with a significantly larger first reactive noneffective step observed among this group versus the PNFs (16.71 cm vs. 9.27 cm, p = 0.006), which does not represent a large, effective recovery step, but rather a difficulty in “catching” the CoM while moving over the BoS. This spatiotemporal inaccuracy led to the execution of more extra steps taken by the PFs to try to avoid an imminent fall (Table 3B). Several past studies also suggest that the characteristics of reactive lateral steps depend on afferent inputs properly informing on both the CoM location and speed of its displacement [38, 39]. In accordance with this view, we assume that the higher eCoM path displacement during the unsuccessful versus successful balance recovery trials (16.71 cm vs. 8.62 cm, Table 3) represents the difficulty PFs have in monitoring their CoM movements after unexpected lateral perturbations, which leads to maladjusted motor responses, i.e., extra steps.
Comparison of Unsuccessful Balance Recovery Events versus Successful Balance Recovery Events of PFs
Since a comparison was made between the unsuccessful balance recovery trials of PFs versus their own successful balance recovery trials, it was not surprising to discover that while the initiation time of the first-recovery step was identical (Table 5A), and step execution differed. This shows that during the unsuccessful recovery trials, PFs performed ineffective balance response and/or had difficulty in recruiting muscle force rapidly enough, i.e., sufficient muscle power, and the inability to decelerate and control CoM movement after a perturbation.
When trying to understand what causes this unsuccessful reactive balance response in PFs, we should take into consideration that the kinematic behavior of the entire balance recovery response is influenced by the first-recovery step strategy. In both the unsuccessful and successful trials of PFs, the responses started most frequently with activation of the leg that became unloaded by the surface translation (i.e., ULSS, CoS, and leg Abd; Tables 2, ,4).4). These strategies were performed because fall events are provoked during high perturbation magnitudes that induce a fast CoM displacement, rapidly loading the stance leg and unloading the leg performing the step, thereby allowing a faster foot-lift of the unloaded leg. These unloaded-leg strategies are suggested to be a more rapid way of enlarging the BoS during exposure to high perturbation intensities [7, 10, 21]. In the current investigation, we found that balance recovery needed to prevent a fall was ineffective mostly due to incomplete CoS (9 out of 18 events) and leg collisions (2 out of 18 events) between the swinging and stance legs during the performance of CoS. This may highlight the disadvantage of performing the CoS strategy during unexpected lateral perturbations. During the CoS strategy, there is a need to coordinate a complex lateral swing leg trajectory to move the stepping unloaded leg across the loaded leg, but there was an increment shown in reactive step time, swing time, and step length, along with a significantly larger eCoM path displacement when using the CoS strategy (Tables 2, ,4).4). This also resulted in a greater number of multiple steps being performed, which the kinematics of the total balance response during the unsuccessful balance recovery trials, resulting in larger total eCoM displacement.
In this study, we acknowledge several limitations. First, we compared a relatively small sample of 10 older female adults who failed to recover their balance in response to lateral surface translations on 18 occasions to 49 PNFs. Thus, the results are based on limiting any generalization of the conclusions. Nevertheless, as we also tested within subjects’ differences (i.e., comparing successful vs. unsuccessful trials of the same participant), this may actually be a strength of the current investigation since all extraneous factors other than those of interest could be controlled or accounted for. Also, since a similar result (a part of step initiation time) was found when we compared the PFs to PNFs, this helps strengthen our conclusions. Second, the perturbation magnitudes were set from low to high magnitudes. Although some may argue that the magnitude of the perturbations should be randomized, we selected this experimental setup because exposing older adults to high perturbation magnitudes at an early stage of the experiment induced a stepping response even at very low perturbation magnitudes, which harmed our ability to identify reactive step thresholds. Third, in our experimental setup, we chose to use unconstrained instructions, i.e., “React naturally to avoid falling,” since these are likely to be more ecologically valid in exploring the ability to respond to a loss of balance in real-life. Some researchers may argue that instructions such as “Try not to take a step” or “Try to step as rapidly as possible” might be more appropriate. Fourth, in the current study detection level (i.e., two-point discrimination and filaments) was not examined, this may influence the results since there is considerable evidence that lower limb somatosensation contributes to the control of upright balance. For example, Meyer et al. [40] found that reduced plantar sensitivity produced a relative shift in compensatory torque production from the ankles and trunk to the hips. Their findings demonstrate that plantar cutaneous afferents play an important role in the shaping of dynamic postural responses. Furthermore, the results suggest that loss of plantar sensation may be an important contributor to the dynamic balance deficits and increased risk of falls associated with peripheral neuropathies. Thus, we suggest that in future studies sensory detection and polyneuropathy will be tested.
Conclusions
This study expands current knowledge by exploring reactive balance responses in cases where older adults failed to recover balance following an unexpected lateral loss of balance. Our results suggest that the CoS strategy performed in response to a loss of balance may be particularly challenging for older adults, resulting in an incomplete step or leg collision preceding a fall. CoS strategy specifically affects the kinematics of the execution of the first-recovery step (in ms), which becomes slower due to a longer time to peak velocity (cm/ms) and slower swinging-leg duration (cm), resulting in an ineffective recovery step, than extra-ineffective step resulting a fall. Our findings highlight the importance of improving specifically recovery step execution phase capabilities to prevent falls and suggest practicing effective stepping strategies as an important part of perturbation-based balance training as a fall prevention intervention.
Acknowledgments
The authors would like to thank all the volunteers who participated in the study.
Statement of Ethics
The study protocol was reviewed and approved by the Helsinki Committee of Barzilai Medical Center in Ashkelon, Israel (ClinicalTrials.gov registration number #NCT01439451, initial release September 23, 2011). A written informed consent was obtained from all those who participated in the study.
Conflict of Interest Statement
IM developed and built the BaMPer perturbation system that was used in this project and holds a patent on some of the technology used in the BaMPer system.
Funding Sources
The study was partially supported by a grant from the Ministry of Health, Israel and by the Helmsley Charitable Trust through the Agricultural, Biological, and Cognitive Robotics Initiative of Ben-Gurion University of the Negev.
Author Contributions
I.M. and S.B. were involved in planning the experiments, analyzing, and interpreting the data, and drafting the manuscript. Y.B. was involved in data analysis, code writing and data interpretation. Y.G.B. was involved in experiment design, data interpretation, and drafting of the manuscript. O.L. and R.D. were involved in subject recruitment, medical screening, and drafting the manuscript.
Funding Statement
The study was partially supported by a grant from the Ministry of Health, Israel and by the Helmsley Charitable Trust through the Agricultural, Biological, and Cognitive Robotics Initiative of Ben-Gurion University of the Negev.
Data Availability Statement
All data generated or analyzed during this study are included in this article. The research data are not publicly available on legal or ethical grounds, and thus, data are available upon request. Further inquiries can be directed to the corresponding author.
Supplementary Material
Click here to view.(17K, docx)
References
1. Krueger PD, Brazil K, Lohfeld LH. Risk factors for falls and injuries in a long-term care facility in Ontario. Can J Public Health. 2001;92(2):117–20. [PMC free article] [PubMed] [Google Scholar]
2. Becker C, Kron M, Lindemann U, Sturm E, Eichner B, Walter-Jung B, et al.. Effectiveness of a multifaceted intervention on falls in nursing home residents. J Am Geriatr Soc. 2003;51(3):306–13. [PubMed] [Google Scholar]
3. Norton R, Campbell AJ, Lee-Joe T, Robinson E, Butler M. Circ*mstances of falls resulting in hip fractures among older people. J Am Geriatr Soc. 1997;45(9):1108–12. [PubMed] [Google Scholar]
4. Greenspan SL, Myers ER, Kiel DP, Parker RA, Hayes WC, Resnick NM. Fall direction, bone mineral density, and function: risk factors for hip fracture in frail nursing home elderly. Am J Med. 1998;104(6):539–45. [PubMed] [Google Scholar]
5. Centers for Disease Control and Prevention CDC . Fatalities and injuries from falls among older adults-United States, 1993-2003 and 2001-2005. MMWR Morb Mortal Wkly Rep. 2006;55(45):1221–4. [published correction appears in MMWR Morb Mortal Wkly Rep. 2006 Dec 8;55(48):1303]. [PubMed] [Google Scholar]
6. Luchies CW, Alexander NB, Schultz AB, Ashton-Miller J. Stepping responses of young and old adults to postural disturbances: kinematics. J Am Geriatr Soc. 1994;42(5):506–12. [PubMed] [Google Scholar]
7. Maki BE, McIlroy WE. The role of limb movements in maintaining upright stance: the “change-in-support” strategy. Phys Ther. 1997;77(5):488–507. [PubMed] [Google Scholar]
8. Crenshaw JR, Kaufman KR. The intra-rater reliability and agreement of compensatory stepping thresholds of healthy subjects. Gait Posture. 2014;39(2):810–5. [PMC free article] [PubMed] [Google Scholar]
9. Batcir S, Sharon H, Shani G, Levitsky N, Gimmon Y, Kurz I, et al.. The inter-observer reliability and agreement of lateral balance recovery responses in older and younger adults. J Electromyogr Kinesiol. 2018;40:39–47. [PubMed] [Google Scholar]
10. Batcir S, Shani G, Shapiro A, Alexander N, Melzer I. The kinematics and strategies of recovery steps during lateral losses of balance in standing at different perturbation magnitudes in older adults with varying history of falls. BMC Geriatr. 2020;20(1):249. [PMC free article] [PubMed] [Google Scholar]
11. Holliday PJ, Fernie GR, Gryfe CI, Griggs GT. Video recording of spontaneous falls of the elderly. ASTM International; 1990. [Google Scholar]
12. Connell BR, Wolf SL. Environmental and behavioral circ*mstances associated with falls at home among healthy elderly individuals. Atlanta FICSIT Group. Arch Phys Med Rehabil. 1997;78(2):179–86. [PubMed] [Google Scholar]
13. Robinovitch SN, Feldman F, Yang Y, Schonnop R, Leung PM, Sarraf T, et al.. Video capture of the circ*mstances of falls in elderly people residing in long-term care: an observational study. Lancet. 2013;381(9860):47–54. [PMC free article] [PubMed] [Google Scholar]
14. Yang Y, Feldman F, Leung PM, Scott V, Robinovitch SN. Agreement between video footage and fall incident reports on the circ*mstances of falls in long-term care. J Am Med Dir Assoc. 2015;16(5):388–94. [PubMed] [Google Scholar]
15. Woolrych R, Zecevic A, Sixsmith A, Sims-Gould J, Feldman F, Chaudhury H, et al.. Using video capture to investigate the causes of falls in long-term care. Gerontologist. 2015;55(3):483–94. [PubMed] [Google Scholar]
16. Yang Y, Komisar V, Shishov N, Lo B, Korall AM, Feldman F, et al.. The effect of fall biomechanics on risk for hip fracture in older adults: a cohort study of video-captured falls in long-term care. J Bone Miner Res. 2020;35(10):1914–22. [PMC free article] [PubMed] [Google Scholar]
17. Shapiro A, Melzer I. Balance perturbation system to improve balance compensatory responses during walking in old persons. J NeuroEng Rehabil. 2010;7:32. [PMC free article] [PubMed] [Google Scholar]
18. Folstein MF, Folstein SE, McHugh PR. “Mini-Mental State”. A practical method for grading the cognitive state of patients for the clinician. J Psychiatr Res. 1975;12(3):189–98. [PubMed] [Google Scholar]
19. Yardley L, Beyer N, Hauer K, Kempen G, Piot-Ziegler C, Todd C. Development and initial validation of the falls efficacy scale-international (FES-I). Age Ageing. 2005;34(6):614–9. [PubMed] [Google Scholar]
20. Haley SM, Jette AM, Coster WJ, Kooyoomjian JT, Levenson S, Heeren T, et al.. Late life function and disability instrument: II. Development and evaluation of the function component. J Gerontol A Biol Sci Med Sci. 2002;57(4):M217–22. [PubMed] [Google Scholar]
21. Batcir S, Shani G, Shapiro A, Melzer I. Characteristics of step responses following varying magnitudes of unexpected lateral perturbations during standing among older people – a cross-sectional laboratory-based study. BMC Geriatr. 2022;22(1):400. [PMC free article] [PubMed] [Google Scholar]
22. Yang F. Identification of optimal foot tactile sensation threshold for detecting fall risk among community-dwelling older adults. Phys Ther. 2021;101(8):pzab119. [PubMed] [Google Scholar]
23. Tuerk C, Zhang H, Sachdev P, Lord SR, Brodaty H, Wen W, et al.. Regional gray matter volumes are related to concern about falling in older people: a voxel-based morphometric study. J Gerontol A Biol Sci Med Sci. 2016;71(1):138–44. [PubMed] [Google Scholar]
24. Caserotti P. Strength training in older adults: changes in mechanical muscle function and functional performance. Open Sports Sci J. 2014;3(1):62–6. [Google Scholar]
25. Simpkins C, Yang F. Muscle power is more important than strength in preventing falls in community-dwelling older adults. J Biomech. 2022;134:111018. [PubMed] [Google Scholar]
26. Campbell AJ, Borrie MJ, Spears GF. Risk factors for falls in a community-based prospective study of people 70 years and older. J Gerontol. 1989;44(4):M112–7. [PubMed] [Google Scholar]
27. Guralnik JM, Ferrucci L, Simonsick EM, Salive ME, Wallace RB. Lower-extremity function in persons over the age of 70 years as a predictor of subsequent disability. N Engl J Med. 1995;332(9):556–61. [PMC free article] [PubMed] [Google Scholar]
28. Lipsitz LA, Jonsson PV, Kelley MM, Koestner JS. Causes and correlates of recurrent falls in ambulatory frail elderly. J Gerontol. 1991;46(4):M114–22. [PubMed] [Google Scholar]
29. Lord SR, Murray SM, Chapman K, Munro B, Tiedemann A. Sit-to-stand performance depends on sensation, speed, balance, and psychological status in addition to strength in older people. J Gerontol A Biol Sci Med Sci. 2002;57(8):M539–43. [PubMed] [Google Scholar]
30. Nevitt MC, Cummings SR, Kidd S, Black D. Risk factors for recurrent nonsyncopal falls. A prospective study. JAMA. 1989;261(18):2663–8. [PubMed] [Google Scholar]
31. Han L, Yang F. Strength or power, which is more important to prevent slip-related falls? Hum Mov Sci. 2015;44:192–200. [PubMed] [Google Scholar]
32. Alcazar J, Losa-Reyna J, Rodriguez-Lopez C, Alfaro-Acha A, Rodriguez-Mañas L, Ara I, et al.. The sit-to-stand muscle power test: an easy, inexpensive and portable procedure to assess muscle power in older people. Exp Gerontol. 2018;112:38–43. [PubMed] [Google Scholar]
33. Bonnefoy M, Jauffret M, Jusot JF. Muscle power of lower extremities in relation to functional ability and nutritional status in very elderly people. J Nutr Health Aging. 2007;11(3):223–8. [PubMed] [Google Scholar]
34. Buatois S, Miljkovic D, Manckoundia P, Gueguen R, Miget P, Vançon G, et al.. Five times sit to stand test is a predictor of recurrent falls in healthy community-living subjects aged 65 and older. J Am Geriatr Soc. 2008;56(8):1575–7. [PubMed] [Google Scholar]
35. Tiedemann A, Shimada H, Sherrington C, Murray S, Lord S. The comparative ability of eight functional mobility tests for predicting falls in community-dwelling older people. Age Ageing. 2008;37(4):430–5. [PubMed] [Google Scholar]
36. Bento PC, Pereira G, Ugrinowitsch C, Rodacki AL. Peak torque and rate of torque development in elderly with and without fall history. Clin Biomech. 2010;25(5):450–4. [PubMed] [Google Scholar]
37. LaRoche DP, Cremin KA, Greenleaf B, Croce RV. Rapid torque development in older female fallers and nonfallers: a comparison across lower-extremity muscles. J Electromyogr Kinesiol. 2010;20(3):482–8. [PubMed] [Google Scholar]
38. Pai Y, Patton J. Center of mass velocity-position predictions for balance control. J Biomech. 1997;30(4):347–54. [PubMed] [Google Scholar]
39. Pai Y, Rogers MW, Patton J, Cain TD, Hanke TA. Static versus dynamic predictions of protective stepping following waist–pull perturbations in young and older adults. J Biomech. 1998;31(12):1111–8. [PubMed] [Google Scholar]
40. Meyer P, Oddsson LIE, De Luca CJ. Reduced plantar sensitivity alters postural responses to lateral perturbations of balance. Exp Brain Res. 2004;157(4):526–36. [PubMed] [Google Scholar]
Articles from Gerontology are provided here courtesy of Karger Publishers
