healthmedicinet

Participants

Community dwelling older adults were recruited from two protected housing institutes.
Eligibility criteria were: 70 years or older; walking independently; Mini-Mental Score
higher than 24; no severe focal muscle weakness or blindness; no known neurological
disorders; no metastatic cancer. Out of 72 seniors who were assessed for eligibility,
19 were excluded (see Fig. 1). All subjects provided a medical waiver signed by their primary care physician clearing
them to participate in moderate physical exercise. The study was approved by the Helsinki
committee of Barzilai University medical center, Ashkelon, Israel (ClinicalTrials.gov
Registration number #NCT01439451). All subjects signed an informed consent statement.

Fig. 1. Study flow chart

Study design

After eligibility and baseline assessments subjects were randomized to two blocks
(27 and 26 subjects, respectively). In the first site, 28 subjects were randomly allocated
to 2 intervention groups and in the second site 25 subjects were randomly allocated
to 2 intervention groups. The subjects random allocation was made by an investigator
not involved in the assessments using computer random allocation software (Random
allocation software version 1.1, Isfahan Iran). Performance-based and laboratory balance
functions were tested before and after the training period by a blinded investigator.
All assessment sessions were performed at the same time of day, and in the same order.

Training programs

We used a mechatronic device that provides controlled and unexpected anterior-posterior
and Medio-lateral platform translations during a single belt, treadmill walking (details
in reference 21] and Fig. 2a-c). The intervention group received 24 training sessions, twice a week for 12 weeks.
Each session lasted for about 20 min and included 3 min warm-up walking in subject
own preferred pace, 14 min of unannounced perturbations exercises, given in random
direction order, during walking (every 20–40 s) and 3 min of cool down walking. During
the training sessions the subjects were instructed to walk on a treadmill, wearing
their own walking shoes, with their hands free to swing; there were no handrails on
the treadmill. To prevent injury if loss of balance occurred during the treadmill
walking, the subject wore a loose safety harness that could arrest the fall, but that
allowed the subject to walk comfortably as well as freedom to execute recovery reactions
without suspension (Fig. 2a). The instructions given to the subjects were: “Walk as naturally as possible at
your preferred stride frequency”. The treadmill’s walking speed was adjusted to the
subjects own preferred speed.

Fig. 2. The perturbation treadmill system used a Photo of the perturbation system during balance training. The system is compose of
a motor-driven treadmill, mounted on a moving platform, motion controller, safety
harness and an operator station; b the perturbations velocity control diagram during training delivered unpredictably
in forward, backward, left, and right directions. Note those are actual measurements
taken during perturbation training. c Example of the perturbation applied during the treadmill walking training (c1–c12). The perturbation applied unpredictably (c5) by horizontal movement of the platform towards the left side during the right foot
initial contact-loading response phases of gait cycle. The participant right foot
was slipped unpredictably to the left while walking in the center of the platform.
The participants performed a cross over stepping response by his left foot (c6–c9), than additional side step was performed by the right foot stepping outside the
treadmill (c10–c12)

The perturbations were in anterior-posterior direction (i.e., sudden acceleration
or stop of treadmills belt) and sudden medio-lateral horizontal translation of the
treadmill that challenges the medio-lateral dynamic control. During all sessions,
400 ms horizontal surface translations were delivered as the subject walked on the
treadmill. The velocity profiles were triangular waveforms with peak velocities of
0.1–3.2 m/s, resulting in displacements of 1–18 cm and peak accelerations of 0.5–16.0 m/s
²
. Perturbation timing was preset and therefore was not given in a specific phase of
the gait cycle or to a specific leg. The perturbation training program had 24 levels
of difficulty with increasing levels of perturbations (i.e., increased displacement,
velocity and accelerations of the horizontal translations, see Table 1). The difficulty level was adjusted according to the subject abilities, starting
from the lowest level of 1 cm displacement at 0.1 m/s velocity and 0.5 m/s
²
acceleration at the first training session. If the subject was able to recover from
all perturbations during the session (i.e., did not fell during the session) and felt
that he can be further challenged, a higher level of perturbation was introduced in
the next session. If not, the same level of perturbation was introduced again until
successfully dealt with. Fall during the training session was defined as load cell
sensors detected 30 % or more body weight suspended by the safety harness.

Table 1. Details of Protocol Used in the perturbation intervention training program

The control group received 24 sessions, twice a week for 12 weeks, 20 min treadmill
walking on the same treadmill but without unexpected perturbations. similar to the
intervention group the control group subjects walked at their own preferred speed
and in their own walking shoes, with their hands free to swing; there were no handrails
on the treadmill, thus they wore a loose safety harnesses that allowed comfortable
walking. Since both group trained on the same system, they were blinded to the allocation
to intervention or control group.

Assessments

For the balance control testing the subjects were instructed to stand barefoot as
still as possible and on a force platform in a standardized stance, their feet close
together. Ten 30-s quiet-standing trials with eyes blindfolded. Center of pressure
and ground reaction force data were collected with a Kistler 9287 force platform (Kistler
Instrument Corp, Amherst, NY, USA), sampled at a frequency of 100 Hz. Evaluation of
balance control was made using both traditional measure of postural sway in eyes closed
condition (e.g. ML-sway, AP-sway, mean sway velocity, and Mean sway area), we also
calculated the Stabilogram-Diffusion Analysis parameters from Center of pressure data.
The Stabilogram-Diffusion Analysis plots (SDA) of the mean square center of pressure
displacement (Critical Displacement, Cd) vs. time interval (Critical Time, Ct) parameters
were extracted from the center of pressure trajectories. The SDA plots derived from
COP trajectories during standing indicate the presence of two different behaviors
depending on the time interval of interest. For shorter time intervals (less than
1 s) the COP tend to drift away from a relative equilibrium point while longer time
intervals (more than 1 s) the COP tends to return to a relative equilibrium point
29]. It has been suggested that long-term region is governed by closed-loop control mechanisms
whereas the postural control systems operate with sensory feedback, while during the
short-term region the postural control system is governed by open-loop control mechanisms
whereas the postural control systems operate without sensory feedback. The transition
point between the short-term and long-term behavior has been termed the Critical Time
(Ct) and sway displacement has been termed the Critical Displacement (Cd) at which
closed-loop control begins to dominate sway behavior. It was described in detail by
Collins and De Luca 29], 30]. The SDA method has been adopted by our research group, we found that SDA parameters
(e.g., Critical Displacement (Cd), and Short-term Effective diffusion coefficients (Ds) were able to predict falls 26] and injury from fall 27].

For the Voluntary Step Execution Test participants were instructed to stand with both
feet on a single force platform, they were instructed to voluntary step as quickly
as possible following a somatosensory cue, given randomly on one of their feet 24], 31], 32]. Center of pressure movement and ground reaction force data were collected from the
force platform, sampled at a frequency of 100 Hz. A total of 8 trials were conducted
in single task condition, 4 forward and 4 backward as well as in dual task conditions.
The average result across task condition was used for statistical analysis. For the
single task, subjects viewed an “X” displayed on a screen in front of them. During the dual task they conducted the
same test while performing the modified Stroop task 33], 34]. Specific temporal events were extracted from the step execution data: (a) Reaction
Time; (b) Foot Contact Time; (c) Preparation Time; (d) Swing Time; as previously described
in detail 24], 31], 32]. The foot contact times (i.e., stepping time) and reaction time duration especially
in dual task condition were able to predict future fall 23] and injury from fall 22], thus both were selected as the primary outcome measure in the present study.

The secondary outcome measures were the self-reported function (Late Life Function
and Disability Instrument (LLFDI)) 35], Fall Efficacy Scale (FES) 36] and performed the Performance-Oriented Mobility Assessment (POMA) 37].

Sample size

Sample size requirements were calculated based on AP postural sway in eyes closed
condition and voluntary step execution times, both were found to predict injury from
fall in older adults (22], 27] respectively). For both calculations, the probability of type I error was 0.05, and
probability of type II error was 0.2. Based on data presented by Kurz et al. 27] that found that the traditional AP postural sway in eyes closed condition was 42.3 mm
in older adults who were injured as a result of falling compared with 32.6 mm in non-fallers
older adults; and Melzer et al. 22] found that the step execution times (i.e., foot contact time) in dual task condition
of older adults who fell and as a result injured were 217 ms longer than those of
non-fallers (1,394 ms vs. 1,177 ms). Using net reduction values (9.7 mm and 217 ms,
respectively) in combination with the initial variance estimates (standard deviations
of 11 mm and 250 ms, respectively), it was determined that 21 and 22 participants
per group would be required, respectively. To account for reported attrition rates
of about 25 % in studies involving older adults 38], we decided to include about 27 participants in each group for a total of 54 (22?×?1.25?=?27).

Data and statistical analysis

PASW Statistics version 18.0 was used for statistical calculations (Somers, NY, USA,
version 18). Baseline characteristics were compared using Independent t-test and Mann–Whitney U-tests for continuous and ordinal variables, respectively.
To analyze the effect of the intervention program a two-way repeated-measures ANOVA
for within subjects (pre vs. post-test) and between group (perturbation intervention
vs. control group) was performed. Since age was significantly different between intervention
and control group subjects, we included age as a covariate in the analyses. The primary
outcome variables were the parameters that we previously found to be related to falls
and injury from fall: the step execution times in single and dual task conditions
(i.e., foot contact-time) and step reaction time, traditional postural sway in eyes
closed condition (ML- and AP- sway, sway velocity and mean sway area) as well as Stabilogram
diffusion Analysis parameters (Critical Displacement (Cd), and Short-term Effective diffusion coefficients (Ds) in eyes closed condition. The secondary outcome measures were LLFDI, FES and POMA.
An intention to treat analysis was conducted by carrying the last obtained measurements
forward for those subjects who did not complete all aspects of the study. Adjustment
of level of significance for multiple comparisons were made. For each testing procedure
(e.g., single task voluntary stepping, dual task voluntary stepping, postural sway
and SDA), a full Bonferroni correction was used to achieve an overall significance
level of 0.05.

For the significant improvement the Effect Size (ES) of Hedge’s g was calculated. The ES of g was calculated by taking the difference between the means of both groups divided
by the average population standard deviation (SD). To estimate the SD for g, baseline estimated SDs of both groups was pooled. When interpreting correlation
magnitudes: 0.0–0.2 is considered small, 0.2–0.5 is considered moderate and 0.5–0.8
is considered large 39].