Validity of six balance disorders scales in persons with multiple sclerosis
DAVIDE CATTANEO, ALBERTO REGOLA, & MATTEO MEOTTI
LaRiCE: Gait and Balance Disorders Laboratory, Servizio riabilitazione neurologica adulti (Int. 282); Don Gnocchi
Foundation, Milano, Italy
Accepted October 2005
in a sample of subjects suffering from MS.
Method.A group of 51 patients were enrolled in the study. The following tests were administered: Berg Balance Scale
(BBS), Timed Up and Go Test (TUG), Dynamic Gait Index (DGI), Hauser Deambulation Index (DI), Dizziness Handicap
Inventory (DHI), and Activities-specific Balance Confidence (ABC). The scales used in this study were initially translated
Results. The sample of subjects reported a mean of 0.98 (1.8SD) falls in the month prior to evaluation. The tests
demonstrated good concurrent validity: Higher correlation coefficients among tests tapping the same aspect and lower
correlation coefficients among tests tapping slightly different aspects. ABC and DHI tests discriminated better than the
others between fallers and non-fallers and appeared the best predictors of fall status. BBS and DGI were not as efficient in
discriminating between groups. Conversely all tests showed good difference validity in the prediction of patients who used an
Conclusion.BBS, TUG, DI, DGI, ABC, DHI have acceptable concurrent validity. The scales have poor performance in
discriminating between faller and non-faller.
The aim of this study was to test concurrent and discriminant validity of several tests of static and dynamic balance
Keywords: Balance, validity, scales, multiple sclerosis
Multiple sclerosis (MS) is the commonest progres-
sive neurologic disease in young adults . In Italy
MS has a prevalence of 30 cases per 100,000
inhabitants and an incidence of 2.2–3.7 new cases
per year .
Abnormalities in balance control are common
findings in subjects suffering from multiple sclerosis
[3–6]. This abnormality, along with other risk
factors, increases risk of falls  preventing the
patients from performing their daily life activities. In
the past decade much attention has been directed
toward the rehabilitation of balance in elderly
subjects. However, the frequency of falls in this
population was found to be 1.5 falls per hospital bed
per year , while another study looking at a mixed
population of MS inpatients and outpatients showed
a mean of 9 falls per year . Assessment of different
aspects of motor impairment and accurate determi-
nation of factors contributing to falls are necessary
for the development of a program for falls prevention
[9–11]. Since falls among patients with MS have a
multifactorial etiology, multiple assessments are
necessary to explore the balance system in order to
classify patients according to their degree of impair-
ment, disability, handicap or quality of life.
Although several scales have been developed to
assess functional level of patients with MS [12,13]
the possibility of using frequently administered scales
also in this group of patients allows comparison of
the results with other studies and with other
pathologies. Moreover, the use of frequently admi-
nistered scales does not require specific training of
raters. Over the past decade balance related instru-
ments such as Berg Balance Scale, Dynamic Gait
Index and Timed Up & Go Test have become
Correspondence: Davide Cattaneo, LaRiCE: Gait and Balance Disorders Laboratory Servizio riabilitazione neurologica adulti (Int. 282); Don Gnocchi
Foundation I.R.C.C.S. V. Capecelatro 66–20148 Milano, Italy. Tel: þ39 (0) 240308282. Fax: þ39 (0) 240308498. E-mail: email@example.com
ISSN 0963-8288 print/ISSN 1464-5165 online ª 2006 Taylor & Francis
Disability and Rehabilitation, June 2006; 28(12): 789–795
increasingly popular in clinical studies. Despite of
the high incidence of falls in the MS population no
study reports the validity of those instruments for this
population of patients.
The aim of this study was to test concurrent and
discriminant validity of several tests on static and
dynamic balance in a sample of subjects suffering
A total of 63 patients suffering from MS attending
the Department of Multiple Sclerosis of Don
Gnocchi Foundation were assessed. To be eligible
for the study, the patients had to meet the following
inclusion criteria: Clinically or laboratory definite
relapsing-remitting or secondary progressive multi-
ple sclerosis, ability to stand independently in
upright position more than 3 seconds, ability to walk
for 6m even with an assistive device.
We excluded patients with cognitive impairments
that might hinder understanding of the tasks to be
accomplished. Patients with visual problem and
impairment of VIII cranial nerve were not excluded.
A sub-group of 51 patients met the inclusion-
exclusion criteria and were enrolled in the study. The
group consisted of 16 males and 35 females, mean
age 45.3 years (standard deviation, SD, 18.1 years).
The onset of pathology was 15.6 years (7.6 years SD)
before the beginning of the study. Fifteen patients
used a walking aid in their daily activities. After
informed consent was obtained, patients completed a
questionnaire providing information about their age,
the onset of pathology, and the number of falls one
month before the assessment procedure. A fall was
defined as any event that led to an unexpected
contact with a support surface. Due to the natural
course of the pathology we restricted the falls
reported by patient to one month prior to evaluation
in order to obtain a reliable picture of patients’
characteristics. Patients were tested wearing their
normal shoes. The assessment was carried out in one
session, the testing protocol took 30–40min to be
administered, patients were allowed to rest during
testing if necessary. The assessment protocol con-
sisted of 1 test on static balance, 3 tests on dynamic
balance and 2 self-reported scales on balance
confidence and level of handicap.
Berg Balance Scale (BBS). The scale rates perfor-
mance from 0 (cannot perform) to 4 (normal
performance) on 14 items. The items explore the
ability to sit, stand, lean, turn and maintain the
upright position on one leg . The psychometric
properties of the scale have been assessed on
populations of elderly subjects. In those groups of
subjects the scale proved to be a valid and reliable
instrument [15–17]. The intrarater and interrater
reliability of BBS were very high, the ICC ranged
from 0.98–0.99 for intrarater reliability and 0.98 for
interrater reliability [16,18]. Moreover the BBS was
found to be a good predictor of falls in a cohort of
elderly subjects [19,20]. A cut-off score of 45 is an
established criterion to identify elderly subjects with
high risk of fall .
Timed Up and Go Test (TUG). The test is a measure
of dynamic balance. It requires the subjects to stand
up from a chair, walk 3m, turn around and be seated.
The subject is timed from the moment he lifts the
pelvis from the chair until he returns with the pelvis in
the chair . The validity and reliability of the
instrument was tested on a cohort of elderly subjects
and patients with unilateral lower limb amputation.
The results showed good correlation between TUG
and BBS (r¼0.81) and moderate correlation be-
tween TUG and Groningen Activity Restriction
Scale (r¼0.39GARS) . The tests showed also
good interrater (0.96) and intrarater (0.93) reliability.
Times of 13.5 seconds or greater have been related to
increased risk of falling in older adults .
Dynamic Gait Index (DGI). The scale measures the
mobility function and the dynamic balance. The
eight tasks of this scale include walking, walking with
head turns, pivoting, walking over objects, walking
around objects and going up stairs. The performance
is rated on a 4-point scale. Whitney found a
correlation between the scores of the Dynamic Gait
Index and the Berg Balance Scale of 0.71 in persons
with vestibular disorders . In a similar group of
patient a score of 19 or less has been shown to be
related to self-reported number of falls in persons
with vestibular disorders . Data on reliability
have already been published in a population of MS
patients and showed both good inter and intrarater
Hauser Deambulation Index (DI). The index rates gait
performance. It has been used in several studies
focusing on patients suffering from MS [26,27]. It
has 10 grades ranging from 0–9; a grade of 0 means
‘‘no gait impairment’’, a grade of 10 means
‘‘Restricted to wheelchair’’.
Dizziness Handicap Inventory (DHI). It is a multi-
dimensional self-assessment scale that quantifies the
level of disability and handicap in three subscales:
Physical, emotional and functional. It is possible to
use both the sum score and the scores of the three
D. Cattaneo et al.
subscales separately. Scores range from 0–100;
where 100 means high level of disability and handi-
cap. DHI demonstrated good internal consistency
(0.91) and test retest reliability (0.97) .
Activities-specific Balance Confidence (ABC). It is a
scale in which the patient rates his perceived level of
confidence while performing 16 daily living activities
. The Test-retest reliability of the ABC Scale
among people who have a lower-limb amputation
was 0.91; moreover, the associations with the 2
Meters Walking Test and TUG test were observed
with correlations of 0.72 and 70.70 . A study
aimed to assess convergent validity between ABC
Scale and Falls Efficacy Scale (FES) in a population
of elderly people indicated the ABC and FES were
highly correlated .
The scales used in this study were initially
translated by two physical therapists proficient in
English whose native language was Italian. The two
translations were compared and when differences
were identified, the texts were modified to obtain
consensus between the two translations. Proble-
matic items and response choices were clarified
contacting the authors of the scales. The final
version was also given to four physical therapists, so
that they could apply the scale to patients in order
to test the intelligibility of the items. The Italian
versions of the scales are available from the authors
Descriptive statistics were used to describe the study
population in terms of demography and disease
characteristics and number of falls.
To study the ceiling effect, the percentages of the
sample scoring the maximum possible scores were
computed. Those percentages reflect the extent that
scores cluster at the top of the scale range. Ceiling
effects 420% was considered to be significant.
Group differences validity was assessed by testing
the extent to which the scores of these scales
differentiated between patients using or not an
assistive device to walk and between fallers and
non-fallers. Mann-Whitney U Test was used to
determine group differences. Logistic regression
was carried out to determine the cut-off value that
predicted a probability to be classified as a faller of
0.5 or greater.
Sensitivity and specificity in predicting falls status
were calculated. The sensitivity was defined as the
percentage of the fallers who were correctly identified.
Alternately, specificity was defined as the percentage
of non-fallers who were correctly identified.
Concurrent validity was tested by assessing the
degree to which each scale correlated with the other
scales, by Spearman correlation coefficient. Correla-
tion coefficients of 0.35–0.49 were interpreted as
weak, those of 0.5–0.79 as moderate and those of
0.8 or greater as strong.
The whole sample of subjects reported a mean of
0.98 (1.8SD) falls in the month prior to evaluation.
The faller group consisted of 20 subjects that
accounted for 39% of the sample with a mean of
2.34 (2.1SD) falls/month.
Ceiling effects occurred respectively 4 times in the
DI and DGI tests (7.8%), 3 times in the BBS
(5.8%), 1 time for the DHI (1.9%), no ceiling effect
was observed for the ABC test.
Table I shows the test scores for the whole group
and those for the faller and non-faller groups.
The whole group of subjects reported moderate
impairment in static balance (BBS scale) and more
pronounced impairment in dynamic balance (DGI
scale). The self-administered tests showed marked
low confidence of patients in their balance skills.
With respect to the three domains of DHI the
emotional domain was the less impaired, with a score
reaching 41% of the maximum possible score.
Conversely, physical and functional domains were
more impaired with scores reaching respectively
50.8% and 51.6% of the maximum possible score.
Table I. Mean and SD scores of the tests for the whole group of patients and the subgroup of fallers and non-fallers.
Whole groupNon-fallers Fallers
Fall vs. non-fall
MeanSD MeanSD MeanSD
BBS, Berg Balance Scale; DGI, Dynamic Gait Index; TUG, Timed Up & GO Test (s, seconds); DI, Deambulation Index; ABC, Activities-
specific Balance Confidence; DHI, Dizziness Handicap Inventory; p column reports the p level of the differences between fallers and
non-fallers (U test).
Validity of balance scales in the MS population
A statistically significant difference of three points
in BBS and DGI were observed between fallers and
non-fallers. DI did not show a statistically significant
difference between groups. TUG showed no statis-
tical or clinical significant difference between groups.
Marked statistically significant differences in ABC
and DHI tests were observed between fallers and
Figure 1 shows the means for each item of the BBS
and the DGI tests. Since the inclusion-exclusion
criteria Item 3 of BBS (Sitting unsupported) was
always scored as 4, also Item 1 (sitting to standing), 2
(standing unsupported) and 5 (transfer) showed
poor variability among subjects and were almost
always scored as 4. Items 6–10 showed more
variability among subjects. The scores were lowest
for the last 4 items of the scale. Those items tap the
ability to keep the balance on a narrower base of
support and during weight shifting.
DGI showed greater impairment in dynamic
balance with respect to static balance. None of the
items of the scale reached the maximum score;
moreover the variability of the items among patients
was higher with respect to BBS. The scores of the
items varied around a mean value of 1.9 points with
the exception of item number 7 (walking around
objects) and 3 (walking with head turning). The
former was the easiest while the latter was the most
difficult (1.5 points).
Group differences between fallers and non-fallers
are shown in Table I. ABC and DHI tests
discriminated better than BBS and DGI tests
between groups. Hauser Ambulation Index (DI)
and Timed Up and Go Test did not discriminate
between the two groups.
Logistic regression was used to determine the cut-
off value sensitivity and specificity for the tests
discriminating between groups (Table II). Generally
sensitivity was poor, while the ABC test showed the
best psychometric characteristics. Specificity was
good especially for the Berg Balance Scale.
All tests were able to discriminate between subjects
who used an assistive device and subjects who did not
use it (p values were always less than 0.001).
Concurrent validity among tests is depicted in
Table III. Berg Balance Scale showed moderate
Figure 1. Mean scores of the Items of the Berg Balance Scale (BBS) and the Dynamic Gait Index (DGI). Means and standard deviations
(T bars) of the Items of Balance Scale (BBS) and the Dynamic Gait Intex (DGI) ordered by Item score.
D. Cattaneo et al.
correlation with dynamic tests especially with the
Dynamic Gait Index. Low correlation between Berg
Balance Scale and self-assessment scales was ob-
served. The dynamic balance tests showed moderate
to strong correlations among them. ABC and DHI
test correlated moderately (70.7).
The aim of this study was to assess the psychometric
properties of well-known scales on balance to
promote functional evaluation of this construct also
in patients suffering from MS.
Our results highlight the incidence of balance
disorders in patients suffering from MS. Our data are
in agreement with those of other studies that
emphasize balance disorder in this pathology [3,5].
Those disorders in the MS population lead to an
increase in the number of falls far beyond those
which are reported for elderly patients .
In this study, the frequency of falls was slightly
higher with respect to our previous findings ; this
difference could be due to differences in the sample
Falls are due to the sum of multiple impairment;
this must be appreciated when evaluating a patient
with balance disorder. With respect to static balance
patients showed poorer performance in the items
requiring a small base of support and weight shifting
and in tasks involving head or whole body rotation.
Some items seem to have poor discriminating
properties because too easy for this population of
Similarly to static balance, dynamic balance was
impaired especially during head rotation in the
sagittal and horizontal planes. Those findings,
combined with the ones obtained with measurement
of head motion during walking , suggest impair-
ment in trunk-head control and/or difficulty to
maintain the balance when visuo-vestibular informa-
tion is challenged by head movements.
Although the four tests discriminate between
fallers and non-fallers, the ability to categorize the
subjects in two groups was poor. The tests showed
low sensitivity. ABC and DHI discriminated better
than the other tests between fallers and non-fallers
and appear the best predictors of fall status. The
scores of those tests were substantially worse for the
faller group than for the non-faller group. ABC
scored 65% higher and DHI 31% lower in the faller
group than in the non-faller group.
Surprisingly, BBS and DGI were not very efficient
in discriminating between groups compared with
other pathologies. Even taking into account the
difference in the population, the Equiscale Test
showed better performance in discriminating be-
tween fallers and non-fallers. BBS showed the
highest specificity but the lowest sensitivity. This
finding correlated with that of Thorbahn and
Newton  which found a sensitivity of 53%
and a specificity of 96%. Conversely, Shumway-
Cook et al.  in a prospective study found values
of 77% for sensitivity and 86% for specificity.
The cut-off score was lower (44 points) than that
reported in the literature (45 points), however cut-off
points varied on the basis of different parameters.
The cut-off point for BBS in an elderly population
varied from 47–38 depending on whether the
patients were categorized as fallers with 1 fall or
more than one falls within 6 months prior to the
study . DGI showed performance similar to
In contrast with other studies on elderly subjects
[33,34] and patients with unilateral lower limb
amputation , the TUG test was not able to
discriminate between fallers and non-fallers in this
population of patients. The mean difference between
the two groups of patients was just 0.2s. The
difference in the discriminating power between
DGI and TUG may be due to the DGI having
multiple items on a wider spectrum of activities;
moreover the DGI also rates the performance of the
No total score ceiling effect was observed on any of
the scales used in the study.
Low sensitivity of tests was generally due to high
scores of patients in the faller group. The falls in this
Table III. Spearman correlation coefficients among tests.
BBSDGI TUGDI ABCDHI
BBS, Berg Balance Scale; DGI, Dynamic Gait Index; TUG,
Timed Up & Go Test (s, seconds); DI, Deambulation Index;
ABC, Activities-specific Balance Confidence; DHI, Dizziness
Handicap Inventory; All correlation coefficients were statistically
Table II. Cut-off scores, sensitivity and specificity of tests.
BBS, Berg Balance Scale; DGI, Dynamic Gait Index; ABC,
Activities-specific Balance Confidence; DHI, Dizziness Handicap
Validity of balance scales in the MS population