Identify subjects at risk for falling using an acceleration based gait analysis system

Falls in the elderly are a major public health issue because of their
associated injuries, morbidity and mortality rates, social cost and financial
cost. Most falls in older adults are caused by impaired locomotion and balance,
reduced vision and other age related body adaptations like reduced muscle
strength. Falling because of tripping or slipping while walking is also a
commonly reported cause of falls, responsible for 17 to 60% of falls in healthy
older adults. With the increased life expectancy of the elderly and their more
active lifestyle there is now an emphasis on identifying subjects at risk for
falling.
Currently fall risk assessment is performed using questionnaires and scales
which lack diagnostic accuracy because of their subjectivity. Other methods are
expensive, complicated and laboratory based. Previous studies used
accelerometers for several targets, like gait analysis in which movement
characteristics were derived from the acceleration signals using specific
algorithms. These studies showed promising results.

Because the increasing fall problem, mainly due to an impaired gait and balance
ability and partly caused by trips, this study will investigate fall risk by
detecting fall related movement characteristics and by detecting stumbles
inclusive the compensation mechanism to recover from the trip. Based on the
promising results using accelerometry for accurate and objective gait analysis,
fall risk will be measured in younger and older (>60y) subjects using a
triaxial accelerometer.

To investigate fall risk objectively, fall related movement characteristics
(based on gait and balance) and the ability to compensate for near falls are
analyzed in younger and older (>60y) subjects under standardized laboratory
conditions.

Four tests will be performed to link specific movement parameters, balance
performance and the ability to recover from a trip with fall risk:

a. Fall risk will be assessed using the Tinetti Scale, the gold standard for
fall risk assessment . This scale consist of a gait and balance score. Based on
this scale, subjects are classified as being at risk/ not at risk for falling.

b. A gait test will be performed to analyze movement parameters. Subjects have
to walk 6 times a 20 meter distance at preferred speed while a small
(56mmx61mmx15mm), light weight (5g) and ambulant accelerometer is attached on
the sacrum with an elastic belt. The accelerometer measures accelerations of
the body in three directions (antero-posterior, media-lateral and
cranial-caudal) with a sample frequency of 100Hz.

c. The balance ability will be tested by performing 4 balance tasks while the
same accelerometer measures the movements of the body. Subjects have to stand
with feet closed on a normal or foam surface while having the eyes open and
closed.

d. Finally a stumble experiment is done to assess the ability to compensate for
a trip. Subjects are asked to walk at their preferred speed on a treadmill,
while wearing the accelerometer attached at the sacrum. After 2 minutes of
normal walking, stumbles are simulated unexpectedly using an extending tripping
leash attached to both legs. The subjects are able to move freely due to
unwinding and winding of the cord on a spill on a fixed frame behind the
treadmill. This spill has a blocking device capable of blocking the leash very
shortly. During the mid or initial swing of one of the legs, the examiner
blocks the leash causing the subject to trip. Falling, however, is not possible
due to a safety harness attached to the ceiling. When this harness starts
bearing weight an emergency switch is engaged stopping the treadmill
immediately. After a perturbation, subjects get several seconds to recover
until a new perturbation is applied. This measurement is repeated while walking
at a slow (40% Fr preferred speed) and fast (20% Fr preferred speed) speed. The
whole experiment will be recorded on video to validate the acceleration based
stumble detection.

To investigate the underlying mechanism of age related mobility changes, two
other tests are incorporated:
* The cognition of hte subjects will be tested using a reaction time test.
Subjects sit behind the computer with their middle and forefinger on two keys
of the typewriter. Four squares will appear at the computer screen. Or all
sqares are red and one of them will change in green (=uncued condition) or two
red squares appear at the left (right) side while one square at the rigth
(left) side will turn green. Subjects have to push the corresponding key on the
typewriter as fast but also as accurate as possible. With this test the ability
to change an automatich repons (in stead of responding left when stimulus
appears left) can be investigated using reaction times. First the subjects get
an instruction, than they get the possibility to try and esxercise 10 times,
while afterwards the real test will start. 120 attempts are performed in which
the cued and uncued stimuli appear with different time intervals (100ms, 250ms,
500ms, 750ms and 12000 ms).

* Muscle strenght of the right lower leg is tested using the Cybex. Subjects
have to produce maximal force during an extension and flexion test.

All acceleration data will be analyzed using specific algorithms programmed in
Matlab(c). Statistical analysis will be performed in SPSS using pearson
correlation to investigate correlations between gait parameters, balance
characteristics and the ability to recover from a trip. Moerover Pearson
correlations and regression analysis will be done to investigate the relation
of cognition and muscle strength with fall risk (gait, stumble recovery and
balance). Pearson correlation will also be used to validate the objective gait
and balance test with the Berg Balance Scale. Differences in muscle strengt,
cognition gait, balance and compensation ability between younger and older
subjects will be investigated using ANOVA (p< 0.005).

The total study takes maximal one hour and is performed once.
Because balancing and walking are activities of daily life, no additional
burden or risk is associated with participation.

Only the stumble experiment can be fatique for the participants, but provide no
risk because the subjects are protected by a safety harness.