Research Reporf

Balance Performance on the Postural Stress Test:

Comparison of Young Adults, Healthy Elderly, and Fallers

The putpose of thI3 study was to compare the postural responses of thee groups of Julie M Chandler
indiuid~ak--healthyyoung adults (n = 42; age, 20-40 yean); healtby, Pamela W Duncan
community-dwelling elderly indiuiduak (n = 66;age, 60-102 years), and elderly Stephanie A Studenski
indiz/idLak with a history offrequent falls ( n = 10; age, 6 6 9 5 yean)--using the
postural stress test (PST). i%e PST is a simple, clinically appliulble, quuntitative
measure of a n individual's ability to withstand a series of graded destabilizing
fwces applied at the level of the subject5 waist. Elderly fallers tend to score lower
on the P,Tthan elderly nonfallen, but age-related d@wnces in pastural
responses during the PST bate not been established. Each subject uundenvent the
PST using a method and scoring procedure desm'bedpreuiously. Results of this
study con$m previous findings that elderlyfallers score sign@cantly lolver on the
PST than either young adult or nonfalling elderly individuals. i%I3study also
showed that there was no dtfeyence in balance strategy scores between the young
adults and the healthy elderly subjects. W e f o r e , it appears that poor
p e r j i i n c e on the PST cannot be anributed to age alone, but may be predictive
of pathological processes that predkpase an individual to j i q w t falls. [Chandler
JM, Duncan P K St-i SA. Balance p e r f o m m e on the postural stress test:
comparison of young adul&,haltby elderly, and fallen Pbys 7ber.1990;
70:410-415.]

Key Words: Equilibrium; Geriatl^ltl^la;

Tests and measurements,functional.

The postural adjustments underlying of support. Assessment of these com- that have evolved include 1) timed
good standing balance are the result ponents of standing balance is a basic balance tests,',2 2) measures of static
of integration of afferent input-pro- pan of the evaluation of instability and dynamic postural sway that use
prioceptive, vestibular, and visual- and falling in a variety of patients. force platforms or other instruments
into effective motor responses that Many quantitative methods for testing to measure body sway?.4 and 3) bal-
minimize body sway and maintain the standing balance have been devel- ance tests that challenge the subject's
body's center of mass within its base oped. The major quantitative methods postural control system by perturbing
the base of support and that analyze
the subject's motor responses by inte-
-- - -
grated electromyography.5,6The cost
J Chandler, MS, PT, is Clinical Associate, Graduate Program in Physical Therapy, Duke University, and complexity of some of these tests
PO Box 3965, Durham, KC 27710 (USA). Address all correspondence to Ms Chandler.
make them impractical for clinical
P Duncan, MA, F'T, is Associate Professor, Graduate Program in Physical Therapy, Duke Ilniversity. application.
S Studenski, MD, is Assistant Professor, Department of Medicine, Duke University, and Chief, Reha.
bilitation Medicine Service, Veteran's Administration Hospital, Durham, UC 27705. One safe, semi-quantitative, and inex-
pensive measure of balance perfor-
This research was conducted in the Department of Physical Therapy at Duke University and in the
Postural Control Laboratory at the Durham Veteran's Administration Hospital and was supported mance introduced by Wolfson and
with Funding from the Charles A Dana Foundation Inc. colleagues7 is the postural stress test
This study was approved by the Duke University Institutional Review Board. (PST). In this test, motor responses to
postural perturbations of varying
7bb article ulas submitted Janzuuy 18, 1983,and ulas accepted March 12, 1990.

Physical TherapyNolume 70, Number 7/July 1990 410111

Fig. 1. Scoring of the postural shes test. Ratings are based on the adaptit~enessof balance strategies used @ subjectsfollowing
each po.~teriorperturbation(I%%, 3%,and 4% of body weighlj. Reponses are ranked in qualitatiz~elydecreasing order of adaptive-
nes. Balance strate@,scores of 2 through 0 indicate ineffective balance responses tbat u~ouldremit in a fall. (Reprinted with permis-
sion @om the Ammencan Geriabia SocieQ and Wolfson LI, Whipple R, Amaman P, et al. Shesing the postural response: a quantitative
method for testing balance.J Am Geriatr Soc. 1786;34:845-850.7)

degrees are measured during normal
standing by using a simple pulley-
weight system that displaces the center
of gravity behnd the base of suppon.7
Specifically, the PST measures an indi-
vidual's ability to withstand a series of
destabilizing forces applied at the level
of the subject's waist. Scoring of the
postural responses is based on a nine-
point ordinal scale (Fig. I), where a
score of 9 represents the most e5cient
postural response and a score of 0
represents a complete failure to
remain upright.
Wolfson et al7 have used the PST pri-
marily with elderly individuals and
have determined 1) that elderly
nursing-home residents who fall score
signif cantly lower than elderly, non-
falling nursing-home residents or
young controls and 2) that elderly,
nonfalling nursing-home residents
score significantly lower than young
controls. They conclude that the PST
can be used to effectively predict
those at high risk for falling. Further-
more, because older subjects tended
to have lower balance scores, Wolfson
et a1 suggest that the PST can be used
longitudinally to follow balance
responses in an individual and that
the PST can be used as a tool for fur-
ther clarifying the nature of balance
responses. Once individuals are iden-
tified to be at risk for falling, they may
be aided by conditioning of balance
responses or other interventions to
counteract balance deficits. The
authors' sample of nonfalling elderly
individuals, however, consisted pri-
marily of nursing-home residents, a
group not representative of the
healthy, community-dwelling, elderly
population. It remains unclear, there-
fore, whether truly age-related differ-
ences in balance performance are
measured by the PST.
Age-related changes in postural con-
trol are well documented in the litera-
ture. Woollacott and colleagues6 cite
evidence for changes at all levels of
the postural control hierarchy in the
aging motor system. Such changes
appear to be greatest at the higher
level of vestibular control, moderate
at the level of automatic postural
responses, and minimal at the mono-
synaptic level. Specifically, the authors
report results of their own work
showing that automatic postural
responses were delayed and that syn-
ergistic organization of postural
responses was altered in a group of
elderly individuals (aged 61-75 years)
who underwent sudden movement of
the support surface. Furthermore,
Overstall et a18 have reported that
sway while standing on a nonmoving
surface (static sway) increases with
age, especially in women. In a group
of individuals aged 75 to 84 years,

Physical TherapyNolume 70, Number 7/July 1990

Rrocklehurst et a19 found that static
sway was also significantly increased
in individuals who experienced
chronic falls.

Although such measures of static and

dynamic postural control are sensitive
-
Table 1 . Subject Groups IdentiJied @ Age and Gender

Group
Age ( ~ r )
-
X s Range
Gender
M F

to the effects of age, they require

expensive equipment and are there- Young adult controls 30.6 6.3 (2wo) 14 28
fore not widely available for clinical Healthy elderly nonfallers 70.7 7.4 (66102) 29 37
use. The PST differs from these mea- Elderly fallers 78.2 8.0 (6695) 7 3
sures in that it is an inexpensive test
that can he easily administered in a
clinical setting. Wolfson et a17 demon-
who has had two or more un-
strated that the PST is sensitive to
explained falls within the 6-month
changes in postural control in both
period prior to the study in the
elderly nursing-home nonfallers and
absence of syncope, acute illness, or
in elderly nursing-home fallers. It
an unusual environmental event or
remains unclear, however, whether
activity. A fall was defined as any dis-
the PST is sensitive to alterations in
turbance of balance that results in a
postural control in healthy,
failure to maintain upright posture
community-dwelling, elderly
during routine activities.
individuals.
All healthy, community-dwelling,
The purpose of this study was to com-
elderly volunteers were screened for
pare performance on the PST among
a history of lower-quarter orthopedic
three groups: 1) healthy young adults;
problems, neurologic disease, dizzi-
2) healthy, community-dwelling,non-
ness, and visual deficits. Any volun-
falling, elderly individuals; and
teers who had a history of major
3) elderly individuals with a history of
orthopedic (eg, hip replacement,
frequent falls. The research question of
fused joint, or amputation), visual,
interest was whether age-related
neurologic, vestibular, or other bal-
changes in balance performance are
ance disorder were excluded from
measureti by the PST. We hypothe-
the study.
sized that there would be no differ-
ence in balance performance between Fig. 2. Subject positioning,for the
Informed consent was obtained from
the young adults and the healthy postural stress test. Subject stands with
each individual prior to participation normal posture as weights are dropped
elderly subjects, but that there would
in the study. Subjects were recruited along pullq track, providing a destabi-
be a significant ddference in balance
from the following sources: Duke lizing ,force posteriorly.
performance between the elderly
University Medical Center (Durham,
fallers and both young and elderly
NC), the Durham Veteran's Adminis- the waist level of each subject. The
nonfallers.
tration Medical Center, Duke Univer- subjects faced away from the pulley
sity medical and graduate schools, system and stood with their arms at
Method
and the Duke University Aging Cen- their sides, their eyes open, and their
ter's registry of healthy, community- feet in a normal, comfortable stance
Subjects
dwelling, elderly individuals. (Fig. 2). A weight belt was fastened
One hundred eighteen male and around each subject's waist, and the
female volunteers participated in this Procedure pulley system was then attached to
study. The subjects were divided into the belt at the subject's back. Each of
Each subject underwent the postural three specified weights (llh%, 3%,
three groups: 1) healthy young adult
stress test as described by Wolfson et and 4V"% of body weight, with a max-
controls; 2) healthy, community-
al.7 We designed a pulley-weight sys- imum weight of 10 Ib*) was used to
dwelling, nonfalling, elderly individu-
tem and followed their testing proce- produce a destabilizing force.
als; and 3) elderly fallers (Table 1). A
dure to deliver a destabilizing force at
faller was defined as an individual
For each trial, one investigator stood
behind the subject, supported the
weight, and then dropped it approxi-
'1 Ib = 0.4536 kg.

Physical TherapyNolume 70, Number 7/July 1990

mately 2 ftt along the pulley track,
creating a posterior force at the sub-
ject's waist. A second investigator
guarded the subject to ensure that he
or she would not fall. A third investi-
gator stood about 12 ft from the sub-
ject and videotaped the subject's per-
formance on the PST. No visible cues
were given to the subject by any of
the investigators to indicate when the
weights were about to be dropped.
Prior to the start of the test, the sub-
jects were informed that their balance
would be disturbed and that their
goal was to maintain their balance in
whatever way was necessary during
the test. No practice trials were per-
formed. All subjects were guarded
with similar care, regardless of
whether they were fallers or nonfall-
ers. Weights were dropped consis-
tently in order of increasing magni-
tude; the subjects, however, were not
informed about either the relative
magnitude of the perturbing force or
the exact moment that the perturba-
tion would occur. Each subject per-
formed a total of three trials.
Scoring
Balance strategy scores. The
motor responses that the subject used
to recover balance after each pertur-
bation were videotaped. An RCA high-
quality camera-recorder* was located
approximately 12 ft from the subject
and about 45 degrees to the right of
the frontal plane. Two of the three
investigators independently viewed
the videotape in order to score the
subject's balance responses.
Scoring was based on the nine-point
scale described by Wolfson et a17
(Fig. 1). A balance strategy score
(HSS) of 9 represents the most efi-
cient level of response in which only
minor postural adjustments are used
to recover balance. Balance strategy
scores of 8 and 7 indicate that addi-
tional ankle, shoulder, or trunk
motion is needed to maintain balance.
* ~ o d eCPR
l 250, RCA Corp, Consumer Electronics Div, 600 N Sherman Dr, PO Box 1976, Indianap-
olis. IN 46206.
A BSS of 6 or 5 is assigned if the sub-
ject takes one or two steps (balance
strategy 6) or more than two steps
(balance strategy 5) backward in addi-
tion to flexing the trunk and shoulder
to recover balance. A BSS of 4 or 3 is
assigned when a subject shows no
evidence of trunk and shoulder syner-
gies and compensates for posterior
displacement solely by taking one or
two steps (balance strategy 4) or
more than two steps (balance strategy
3) backward. A BSS of 2 is assigned
when the subject shows inadequate
shoulder and trunk synergies and
stepping reactions to recover balance.
A BSS of 1 is assigned when the sub-
ject shows no synergies or stepping
reactions but demonstrates a protec-
tive landing response in anticipation
of a fall. A BSS of 0 is assigned when
no corrective or protective landing
responses are demonstrated.
Each of three trials was scored for
each subject. The total score for the
three trials was calculated and
referred to as the total BSS.7 A maxi-
mum total BSS of 27 (9 X 3 trials)
was possible. All subjects tested
(N = 118) were videotaped during
each of the three PST trials. A sample
(n = 88) of those tested was used to
determine the interobserver agree-
ment rate. Two observers indepen-
dently viewed the videotapes and
scored each trial. The number of trials
in which there was total agreement
between the observers was divided by
the total number of trials (3 X 88)
and multiplied by 100. The inter-
observer agreement rate was 89.2%.
Being satisfied that our interobserver
agreement for PST scoring was sufi-
ciently high, we proceeded to analyze
the PST data for the three groups of
subjects. In cases of disagreement, the
BSS assigned by the third examiner
was used.
The investigators who scored balance
responses from the videotapes may
have been aware of the falling status
Physical TherapyNolume 70, Number 7/July 1990
of the subject. Because scoring was
based on the presence or absence of
specific balance responses, however,
they believed that their scoring was
not influenced by knowledge about
the subject.
Number of trials with effective
balance. In addition to the total BSS,
the number of trials with effective
balance was recorded for each sub-
ject. E$ectiw balance was defined as
the subject's ability to maintain
upright posture without intervention
of another person or object. A trial
with effective balance, therefore, was
associated with a BSS of 3 or above
on the rating scale (Fig. 1). A maxi-
mum of three trials with effective bal-
ance was obtainable.
Data Analysis
The total BSS for each subject was
calculated by summing the BSSs
across the three individual trials.
Because of the ordinal nature of these
data, the Kruskal-Wallis rank-order
nonparametric statistical test was used
to assess differences in the total BSSs
among the three groups of subjects.
Further pair-wise comparisons
between groups were then made
using Ryan's test for ordered data.
The percentage of subjects in each
group maintaining effective balance
on each trial was calculated. A com-
parison of the number of effective
balance trials among the three groups
is presented graphically in Figure 3.
To control for a potential age effect
on total BSS between the healthy
elderly subjects (mean age = 70.7
years) and the elderly fallers (mean
age = 78.2 years), we reanalyzed the
data, eliminating the data on all
healthy elderly subjects aged 60 to 69
years from the analysis. The median
total BSS of the remaining healthy
elderly subjects (n = 37, mean age =
75.5 years) was 21, the same as that of
the entire group of healthy elderly
subjects, thereby yielding the same
statistically significant result on the
Kruskal-Wallis test. Subsequent discus-
sion will therefore refer to results
obtained on the entire group of
healthy, community-dwelling, elderly
subiects.

Fallers Healthy Elderly Young Adults

(n=10) (n=66) (n=42)
There was a significant difference in 100- :.:.:. ......
the ranks of the BSSs across the three ::::::
~iiiii g.;
::::::
groups (H = 21.94, df = 2, p < .01). 333 ,#$
Post hoc analysis (Ryan's test for :.:.:.
...... ......
ordered data) demonstrated a signifi- m.
ii#
::::::
#$
::::::
......
cant difference in the ranks of the ....
:::::: g.'
BSSs between the young adult con- :::::::
....... 32;
.....
trols and the fallers and between the % of iiiiiii 5::
gj
healthy elderly subjects and the fall- Group 50- :::::::
::::::: I:;:;:
:<:::
:......
:::::
ers, but not between the healthy ......
::::::::::::.
elderly subjects and the young adult :::::: iiiiii
......
iiiiii
...... ::::::
controls (Table 2). :::::::
iiiiii
......
::::::
3;;;
......
All 118 subjects demonstrated effec- ......
......
:::::: iiiiii
......
tive balance on trial 1, the mildest :::::: g.:
:>>
perturbation (I%% of body weight). --- I rn
I - .t:. I l l .:t:

i h i
1 1
1 1 1 1 1 1

On trial 2 (3% of body weight), all i, 0 1 2 3 0 1 2 3

young adult controls and healthy Number of Trials with Effective Balance
elderly subjects maintained effective
balance, whereas only 60% of the fall- Fig. 3. Companion of the number of effective balance trials among the three
ers were able to d o so. On the third groups. Fipre shows the percentage ofsu&ects in each group scoring 3 or higher (effec-
and most forceful perturbation (4%% tive balance) on each of the three postural stress test trials. At least 98% of all su&ects
of body weight), all of the young in the young adult and healtby elderly groups scored 3 or higher on all three trials,
adult controls and all but one of the whereas only 50% of the fallers were able to do so. Forty percent of the fallers had only
one trial with effective balance.
healthy elderly subjects were able to
maintain effective balance. By contrast,
only 50% of the fallers were able to three-step strategies, along with the This latter finding is in contrast to that
maintain effective balance (Fig. 3). shoulder and trunk synergies, to of Wolfson et a1,7 who found that
realign their center of gravity over elderly nonfalling nursing-home resi-
their base of support. Balance strategy dents used less shoulder and trunk
scores of 6 o r 5 were common for flexion during the more forceful per-
Our findings suggest that healthy, both groups. turbations and therefore received
community-dwelling, elderly individu- lower BSSs than the younger controls.
als demonstrate balance strategies They further suggest that the loss of
similar to those of young adults as
measured by the PST. Elderly fallers,
however, showed significantly less
effective balance strategies and were Table 2. Comparison of Balance Strategy Scores (BSSs) Among the Three Su&ect
therefore more likely to fail portions Groups
of the test, especially as the backward-
perturbation force increased. Total BSS
Young Adults Healthy Elderly Fallers
The balance strategies demonstrated (n = 10)
(n = 42) (n = 66)
by the healthy elderly subjects and the
young adult controls follow the same
pattern described by Wolfson et al.' Median BSS 21 21 12
In both studies, for example, elderly Median rank 65 65 4.5
subjects and young adults primarily Medlan BSS on each Trlal
used an ankle dorsiflexion strategy at
Trial 1 (1l/2% BWB) 9
the lowest perturbation in order to
effectively recover balance. A BSS of 9 Trial 2 (3% BW) 6
was common in both groups. With Trial 3 (41/2% BW) 6
increasing perturbations, both groups
tended to use wellcontrolled one- to "BW = body weight.

Physical TherapyNolume 70, Number 7 m y 1990

shoulder and trunk flexion synergies
in older nonfalling individuals repre-
sents a mild deterioration in postural
response that may be associated with
age alone. Because our study did not
show a significant difference between
BSSs on the PST in the young adults
and the healthy elderly subjects, we
hypothesize that the postural
responses to backward perturbations,
as meawred by the PST, d o not nec-
essarily deteriorate with age. Wolfson
and colleague^'^ observation is more
likely related to the fact their their
sample of nonfalling nursing-home
residents is not representative of the
healthy, community-dwelling, elderly
population.
Our finding that elderly fallers
showed significantly less effective bal-
ance strategies on the PST than either
nonfalling elderly subject5 or young
adult controls is similar to that
reported by Wolfson et al.' At the
lowest perturbation, RSSs ranged
from 7 to 3, indicating that subtle,
covert postural adjustments (as
described by a BSS of 9) were not
sufficient to recover balance. At the
more forceful perturbations, fallers
typically took multiple small steps
backward, with or without associated
trunk and shoulder synergies, but
often failed to recover their balance
without intervention from one of the
examiners.
Although there were gender differ-
ences between our group of healthy,
community-dwelling, elderly subjects
(56% women) and our group of
elderly fallers (30% women), we d o
not believe that gender account? for
the differences in RSSs on the PST
between the two groups. Insubstantial
evidence exists in the literature to
suggest that gender significantly influ-
ences age-related changes in postural
control. Overstall et aln found that
static sway increased with age, espe-
cially in women. In a more recent
study, Rrocklehurst et a19 found that
sway wa5 influenced by age only, not
by gender. To date, no other investi-
gator has shown that there are signifi-
cant differences between men and
women in postural control mea5ures.
Overstall and a5sociates'n finding that
women may show increased sway
should not be overlooked. However,
the fact that our sample of fallers con-
sisted primarily of men and our sam-
ple of healthy, nonfalling, elderly sub-
jects consisted primarily of women
strengthens our argument that gender
did not contribute significantly to our
findings.
Summary
Based on data from our large sample
of healthy, community-dwelling, non-
falling, elderly individuals, it appears
that the PST may not be sensitive
enough to detect subtle age-related
deteriorations in postural control. In
addition, the fact that our sample of
elderly fallers was limited to only 10
subjects may restrict the strength of
Physical TherapyNolume 70, Number 7(July 1990
our conclusions. Yet, the relatively
high incidence of ineffective balance
responses (BSSs of 2 or less), coupled
with relatively low RSSs on the PST in
the elderly fallers as compared with
the healthy elderly subjects and the
young adult controls, suggest5 that the
PST may be a sensitive, easily adminis-
tered clinical tool for identifying and
monitoring individuals who have seri-
ous balance deficits. Further testing of
the PST in a larger sample of elderly
fallers is warranted.
References
1 Potvin AR, Syndulko K, Tourtellotte WW, et
al. Human neurological function and the aging
process. J Am Gerian Soc. 1980;28:1-9.
2 Bohannon RW, Larkin PA, Cook AC, et al.
Decrease in timed balance test scores with
aging. P@s 7bm. 1984;64:1067-1070.
3 Fernie GR, Gryfe CI, Halliday PJ, et a]. The
relationship of postural sway in standing to the
incidence of falls in geriatric subjects. Age Age-
ing 1982;l:ll-16.
4 Shimba T. An estimation of center of gravity
from force platform data. J Biomech.
1984;17:5>60.
5 Nashner LM. Fixed patterns of rapid posture
responses among leg ~nusclesduring stance
Exp Brairz Kes 1977;30:1>24.
6 Woollacott MH, Shumway-Cook A, Nashner
I.M. Postural reflexes and aging. In: Mortimer
J, Pirozzolo F, Malletta G, eds. 7be Aging Ner-
11ou<System. New York, NY: Praeger Publish-
ers; 1982:98.
7 Wolfson LI, Whipple R, k n e r m a n P, et al.
Stressing the postural response: a quantitative
method for testing balance. J Am Gerian Soc.
1986;34:84>850.
8 Overstall PW, Exton-Smith AN, Imms FJ, et
al. Falls in the elderly related to postural
imbalance. Br Med J. 1977;1:261-264.
9 Brocklehurst JC, Robenson D, James-Groom
P: Clinical correlates of sway in old age: sen-
sory modalities. Age Ageing. 1982;ll :I-10.