Assessment of Mood States in
Psychiatrically Disturbed
Patients With Dementia
Ralph H. B. Benedict, Ph.D.
Marion Zucker Goldstein, M.D.
Leonard R. Derogatis, Ph.D.
Labile affect is a conlmon characteristic of elderly psychiatry patients, especially among patients in the early
stages of prinzary progressive delnentia. The autbors
sought to develop a method of assessing tbe full range of
positive and negative moods in pSJJchiatl"ic patients with
early denzentia and healthy elderly subjects. They modified the Derogatis Affects Balance Scale (DABS) to ease its
adl1linistration in cognitively cOl1lpronlisedpatients and
presented it to 51 geropsychiatry inpatients witb demerltia of either Alz!Jeil1ler's or vascular etiologj'. The modified DABS was found to discriminate depressed and
nondepressed denzentia patiel1,ts and have acceptable
test-retest reliability coefficients, despite extensive variation in scores. Analyses of construct and criterion-related
validity were also consistent with previous DABS norl11ative studies. (American Journal of Geriatric Psychiatry
1996; 4:298-310)

L

ability of affect and mood swings are
common clinical features of geriatric
psychiatry patients suffering from progressive dementia. The adverse impact of
these emotional phenomena has been
documented by several authors. l -4 Included among Roth's5 early case descriptions were two geriatric psychiatry
inpatients diagnosed with "arteriosclerotic psychosis," each ofwhom presented

with labile affect and euphoria soon after
admission to the hospital for a suicide attempt. It is now widely recognized that
such episodes of uncharacteristic, fluctu..
ating mood can increase the burden of
caregivers and lead to psychiatric hospitalization. The tendency for psychiatric
patients with dementia to experience dramatic fluctuation in mood can also influence decisions made on behalf of patients

Received August 8, 1995; revised January 3, 1996; accepted January 31, 1996. From the State University of
Ne\vYork (SUNY) at Buffalo. Address correspondence to: Ralph H. B. Benedict, Ph.D., State University of Nc\v
York (SUNY) at Buffalo, Dept. of Neurology, Erie County Medical Center, Suite 164, Buffulo, NY 14215.
Copyright © 1996 American Association for Geriatric Psychiatry
298

VOLUME 4 • NUMBER 4 • FALL 1996

Benedict et at.
with regard to postdischarge living arrangements and expenditures for care.6-8
Although dysphoria is the most frequently encountered affective symptom
in psychiatry patients with dementia,
some recent data suggest that positive
mood states may be' equally important
where the patient's psychosocial adjustment is concerned. For instance, in a
pharmacologic trial designed to treat depression in dementia, Reifler et al. 9 observed that Alzheimer's disease patients
responding to treatment were more
cheerful, active, engaging, and in turn
more cooperative with caretakers and better able to participate in family activities.
Also, research with healthy adults has revealed significant correlations between a
high frequency of positive mood states
and increased socialization. 1o The implication of these findings is that it might be
useful to monitor both positive and
negative mood states in cognitively compromised psychiatric patients during hospitalization and to consider these data
when making discharge plans.
Intuitivel~ clinicians are apt to conceptualize negative and positive mood as
opposite poles of a single psychological
construct. It is logical to assume that if a
patient evidences fewer symptoms of depression, he or she will also experience a
higher frequency of positive moods, such
as jo)', affection, and contentment. However, most factor analytic studies have supported a model that portrays positive and
negative moods as separate, orthogonal
constructs. II According to the structural,
or "two-factor" model, positive moods reflect the subject'S general interest in or
motivation for activi~ as might be illustrated by the following adjectives: active,
excited, joyful, and warm. Negative moods
(or affects) include dysphoria and its related mood adjectives, as well as feelings
regarding general emotional distress,
anxie~ fear, and hostilit}'. Although psychometric tests of emotional function are
instrumental in the assessment of psych iTHE AMERICAN JOURNAL OF GEIUATIUC PSYCHIATRY

atric outcomes, the commonly used depression self-rating scales 12- 14 neglect
many of these positive and negative mood
states. The Geriatric Depression Scale
(GDS), for example t a reliable and valid
instrument for detecting symptoms ofmajor affective disorder in geriatric patients
with possible dementia,13,15,16 does not
measure positive mood states, or negative
moods other than guilt and dysphoria. Research related to questions concerning
the influence ofpositive moods in dementia requires a reliable and valid instrument
designed to assess the full spectrum of
affect.
To the best of our knowledge, there
are no psychometric instruments designed to assess positive mood states that
have been adequately tested in a sample
of dementia patients. To overcome the
limitations of depression self-rating scales,
we decided to examine the potential utility of the Derogatis Mfects Balance Scale
(DABS), 17,18 which was designed to measure the frequency of mood states in accordance with the two-factor model.
The DABS measures eight aspects of
mood (jo)', contentment, vigor, affection,
anxiet)r, depression, guilt, hostility), which
are grouped into the general constructs of
positive and negative affects. The test is
easy to administer and can be completed
by most healthy adults in 5 to 10 minutes.
Psychometric studies of the DABS have
demonstrated that the test is reliable and
valid in normal and general psychiatry
populations. 17, 18 Also, the practical utility
ofthe DABS has been demonstrated in several studies with a wide range of healthy
volunteers, medically ill, and psychiatry
patients. For instance, Rabins and colleagues 19 used the DABS to examine coping responses in a sample of caregivers of
patients with Alzheimer's disease or cancer. Positive mood frequencies correlated
significantly with number of social contacts, as well as measures of religious faith
and family cohesiveness.
In a study of psychosocial adjustment
299

Mood Assessment in Dementia
in medical students,20 there were significant associations found bernreen adverse
life events and lowered positive mood
scores and increased negative mood
scores. Hoehn-Saric and colleagues21
showed that patients with mixed anxiety
disorder and depression had higher negative mood state scores than patients with
anxiety disorder alone.
Therefore, given the wealth of information it provides about a patient's affective experience, as well as its brevity and
minimal respondent burden, the DABS
could be a valuable tool for studying positive and negative mood states in patients
suspected ofhaving progressive dementia.
Unfortunatel~we have found that the
DABS is difficult to use in psychological
evaluations of even mildly cognitively
compromised patients. Some patients become tangential when asked to indicate
how often they experience a particular
mood, and others tend to have difficulty
interpreting the mood adjectives correctl~

In this stud~ we endeavored to remedy the procedural difficulties with the
DABS so that the test could be used with
geriatric, psychiatrically disturbed patients with mild-ta-moderate dementia.
Toward this end, we developed a modified
administration format for the DABS and
presented it to geriatric psychiatry inpatients with dementia and age-matched
normal control subjects. We now report
our findings regarding the modified
DABS's test-cetest reliability and validi~

METHODS
Subjects
Fifty-one patients, age 55 and over,
with a diagnosis ofprimary progressive dementia, were studied. All were inpatients
on a geriatric psychiatry unit of a university-affiliated hospital. In each case, the
hospital admission had been precipitated
300

by an acute episode of altered mood and
disruptive or dangerous behavior. Specifically; the patients manifested one or more
of the following symptoms or behaviors:
night wandering, wandering in traffic, violence, threat ofviolence, threat ofsuicide,
or severe depression. Each patient underwent a comprehensive psychiatric examination performed by the second author, a
boa,rd-certified geriatric psychiatrist. The
first author, a clinical neuropsychologist,
also examined each patient in order to
establish the presence and quality of dementia. Together, the psychiatrist and
neuropsychologist arrived at a consensus
diagnosis based on the psychiatric. and
neuropsychological examinations, the patient's medical and personal history; cranial neuroimaging, and laboratory
studies. Patients meeting DSM..lV22 diagnostic criteria for dementia of the Alzheimer type (DAT) or vascular dementia
(VaD) were included in the stud~ In most
cases, the patients also met research criteria for either probable Alzheimer's disease 23 or Van,24 although there were
insufficient data to establish a clear course
of the dementia that would help in differentiating OAT and VaD in some cases.
Therefore, several of the VaD patients
were suspected ofhaving both cerebrovascular disease and Alzheimer s disease (Le.,
mixed disease). In all, there were 15 male
and 36 female psychiatry patients with dementia, 42 of whom were white, with the
remaining patients being Mrican American. A total of25 patients were diagnosed
with DAT and 26 with VaD or mixed disease. The average age of the dementia
sample was 74.8 ± 6.1 years, and the average educational level was 10.6 ± 2.9
years.
Seventy-four healthy volunteers from
the same metropolitan area as the patient
sample were also examined. Some of the
normal subjects were recruited through a
university-based volunteer organization,
and others responded to a newspaper advertisement and were paid for their part

VOLUME 4 • NUMBER 4 • FALL 1996

Benedict et all
ticipation. All of the normal control subjects were screened for the presence of
neurological and psychiatric illness by
telephone and face-ta-face intervie~ A1so t
each control subject tested in the normal
range on a short battery of neuropsychological tests (described below). All
were over the age of 60 years (mean =
72.2 ± 5.9) and had completed an average of 14.3 ± 2.6 years of education. The
racial composition of the control group
was 73 white subjects and 1 black subject;
there were 25 men and 49 women.
Procedures
The neuropsychology battel)) administered to both patients and control subjects, included the following tests: the
Mini-Mental State Exam (MMSE)t 25 the
Controlled Oral Word Association Test,
with letter cues, 26 and category cues, 27 the
30..item short form of the Boston Naming
Test,28 the Developmental Test of VisualMotor Integration,29 and the Trail Making
Test. 30 Recently revised versions of the
Hopkins Verbal Learning Test (HVLT_R)31
and the Brief Visuospatial Memory Test
(BVMT-R)32 were also administered.
Based on DSM..IV criteria, dementia was
defined as impairment in memory (as in..
dicated by the HVLT-R and/or BVMT-R)
plus impairment in at least one other domain of cognitive function. To screen for
depression, we employed a IS-item short
form of the GDS. Following the method
of Parmalee et aI., 15,33 the GOS items were
read to patients and volunteer subjects to
ensure maximum compliance and understanding. When administered in this for..
mat, the GDS was found to be reliable and
valid in a large sample of nursing home
residents. Each patient was also rated by
nursing staffand the first author on a short
form of the Montgomery-Asberg Depression scale (MADS). 34
The standard administration of the
DABS has been described elsewhere. 17,18
Subjects are asked to rank 40 mood adjec-

tives (e.g., happ)) irritable, loving) in terms
of the frequency of occurrence on a fivepoint scale ranging from 0 (not at all) to
4 (always). The test stimuli are presented
visually in the format of a questionnaire.
The frequency rankings are summed
within each of the following eight general
mood subscales: jo~ contentment t vigor,
affection, anxie~ depression, guilt, and
hostilit}'. Five summary scores are also calculated. The total positive affects (POS)
and negative affects (NEG) scores are the
respective sums of rankings for descriptors categorized under each general construct. The POS includes the moods jo~
contentment, vigort and affection. The
NEG includes anxiety; depression, guilt,
and hostili~ The range for each of these
measures is 0 to 80. The affects..balance
index (ABI) is calculated as POS - NEG /
20, representing the frequency ofpositive
moods over that of negative moods. The
ABI ranges from -4, which is the maximum
negative valence, to +4, which is the maximum positive valence. The affects..expressiveness index (AEI) is the sum total of all
mood rankings, regardless of valence,
ranging from 0 to 160. Finall~ the posi..
tive-affects ratio (PAR), calculated as POS
/ AEI, is the proportion of total positive
affective expression.
For this study; the standard administration of the DABS was modified for use
in a cognitively impaired subject sample.
First t subjects were provided with a visual
representation of the scale that was to be
used for each adjective. The scale depicted
each possible ranking, equally spaced
from left to right across a 5 X SI/2-inch paper, as follows: 0 - never, 1 - rarel~ 2 sometimes, 3 - frequently, 4 - always.
Once the subject understood the scale,
each mood adjective was read by the examiner with the question: "Over the past
several days, how often have you been
feeling __?U It was common for patients
to lapse into conversation, and when this
occurred, the examiner repeated the
question and directed the patient's atten-

THE AMEIUCAN JOURNAL OF GERIATRIC PSYCHIATRY

301

Mood Assessment in Dementia
tion to the ranking scale. The frequency,
rather than the quali~ of each mood descriptor was emphasized. Subjects were
frequently reminded that the test would
progress more quickly if they would concentrate on the scale and simply indicate
which descriptor best matched their experience over the past several days. For each
subject, the examiner was permitted to
pause and provide the meaning of any
troublesome mood adjective or explain
the meaning of the visual scale for frequency judgments.
The patients were all examined in the
neuropsychology clinic, in the same hospital as the geropsychiatry ward in which
they were being treated. The examinations were usually done within 1 week of
hospital admission. Normal control subjects \vere tested in the neuropsychology
clinic, at home, or in a testing room designated at a local senior citizens' recreation center. Regardless of the exam
location, neuropsychological testing was
always conducted in a room that was free
of distraction.
Thirteen patients and 21 control subjects were seen for a follow-up neuropsychological examination that included the
DABS and the neuropsychology test bat..
teJ.1r. The average test-retest interval was
27.6 ± 19.7 days for patients and 50.1 ±
34.4 days for control subjects. Participation in the follow-up examination was vol..
untary for both groups. Patients were seen
while in the hospital, during a 2-5·day period before discharge. Control subjects
were tested in the same environment both
times.
Statistical Analyses

The first goal of the study was to replicate previous findings of the DABS with
regard to its reliability and validity in normal geriatric subjects. Test-retest reliability was assessed by means of Pearson ,.
reliability coefficients. An extension of the
construct validity of the DABS was then
302

attempted by means of principal-components analysis with varimax rotation.
These same procedures were also performed in the patient sample.

RESULTS
Group Differences

The dementia and control groups
were matched on age, but disparate in
level ofeducation (t(I) = 7.54;P < 0.0001).
Therefore, we used one-way analyses of
covariance (ANCOVAs)t with education
entered as a covariate, to compare the
group means on the DABS. Table 1 presents mean scores on age, education,
MMSE, and DABS measures, with the dementia group separated according to
normal « 5) vs. abnormal ~ 5) GOS
score. I5 ,33 In Figure 1, the nlcan scores for
each subgroup are plotted in T-score
equivalents (mean = 50 ± 10) based on
preViously published normative tables for
adults (that are not balanced for age).17
Figure 1 demonstrates that the DABS
scores obtained in the normal elderly
group conform quite nicely to the normative database previously published by
Derogatis and colleagues. 17,18 As in previous normative studies with the DABS,
healthy control subjects reported positive
mood states much more frequently than
negative moods. As a concrete example,
the average ranking in the joy category
(e.g., happ~ glad) was 2.9, which corresponds roughly to the descriptor "frequcntl~" On the other hand, the average
ranking in the depression category (e.g.,
sad, miserable) was 0.7, or urarel~" Hence,
like younger subjects, elderly individuals
tend to experience positive moods at a
higher frequency than negative moods, re·
suIting in a positive ABI. Note that normal
elderly group means fall within 0.5 SO of
the norm for the test. This finding suggests
that scores from the modified DABS used
here can be interpreted using the existing
VOLUME 4 • NUMBER 4 • FALL 1996

Benedict et ale
TABLE 1.

Group means ± standard deviations (and ranges) for demographic variables and
measures derived from tbe Derogatis Affects Balance Scale (DABS)
Dementia Patients
Nondepressed
Depressed
(II = 22)
(II
29)

=

Age, years

Education, years
Mini-Mental Scate Exam

74.8 ± 5.7 (65-84)
10.7 ± 3.3 (6-20)
22.0 ± 4.5 (7-28)3

74.7 ± 6.5 (55-85)
10.5 ± 2.6 (6-18)
22.4 ± 3.8 (16-28)3

Control Subjects
(II = 74)

72.2 ± 5.9 (64-88)
14.3 ± 2.6 (8-20)
28.3 ± 1.4 (25-30)

(MMSE)

Derogatis Affects Bahlncc Scale (DABS)
joy
13.7 ±
Contentment
13.3 ±
Vigor
9.7 ±
Affection
13.0 ±
Anxiety
7.6 ±
Depression
5.4 ±
Guilt
3.6 ±
I-Iostility
6.3 ±
Positive Affects Total (POS) 49.7 ±
Negative Affccts Total (NEG) 22.8 ±
Affects Balance Index (ABI)
1.4 ±
Affects Expressiveness Index 72.6 ±
Positive Affects Ratio (PAR)
0.69 ±

4.9 (3-20)b
4.2 (6-20)Olb
4.3 (0-18)3
5.0 (4-20)
3.6 (0-14)b
4.1 (O_12)ab
2.9 (O-9)b
4.3 (O-l~i)
13.6 (27_73)ab
11.4 (1-41)b
1.1 (-0.3 - 3.5):'b
13.1 (56-98)
0.15 (OA6-0.98)b

8.1
7.9
6.8
10.0
11.9
12.6
10.1
9.3
32.8
43.9
-0.56
76.7
0,41

± 5.4 0-17 3
± 4.8 (0-18)3
± 4.3 (0-15)3
± 6.0 (0-20)3
± 4.8 (2-20)3
± 5.4 (3-20)3
± 5.0 (0-19)3
± 5.6 (1-20)3
± 18.2 (O~9)a
± 17.0 (16-76);1
± 1.4 (-3.4 - 2.0)a
± 20.5 (17-118)
± 0.21 (0-0.72);\

14.6
14.7
13.3
14.3

6.5
3.3
3.9
4.9
56.8
18.6
1.9
75.4
0.76

± 2,4 8-20
± 2.5 (8-20)
± 2.8 (5-19)
± 2.9 (7-20)
± 2.6 (0-12)

± 2.6 (0-9)
± 2.6 (0-9)
± 2.7 (0-10)
± 8.9 (28-79)
± 8.2 (0-31)
± 0.7 (-0.15 - 3.5)
± 10.7 (48-102)
± 0.10 (0.47-1.0)

Onc·\vay ANOVA comparing group means for age \vcre not significant. ANOVA comparing group
means for years of education \vcre significant (F(2 122) = 28.2; P < 0.0001). Nc\vman-Keuls comparisons
revealed significant differences bct\vcen control subjects and each dementia subsample.
uANCOVA, controlling for level of education t reveals a significant difference bct\vccn the patient
subsample and control subjects, using an alpha criterion of 0.01.
bANCOVA, controlling for level of education, reveals a significant difference bet\veen depressed and
nondcprcsscd dementia subsamplcs, using an alpha criterion of 0.01.

Note:

t

normative base, and that the influence of
age on DABS scores is minimaL
The ANCOVAs in Table 1 comparing
the subgroup means were significant for
each DABS measure except the Affects Expressiveness Index (AEI). Thus, although
large differences were apparent with respect to the valance of mood frequency
ratings, the subgroups tended to use the
descriptors "always" or "not at all" with
equal frequen~
Table 1 indicates that the dementiaonly subgroup obtained mostly normal
DABS scores, although there were statistically significant differences on the
subscales for contentment, vigor, and depression, as well as POS and ABI. However, as is readily apparent in Figure 1, the
effect sizes between the dementia-only patients and control subjects were quite
THE AMEIUCAN JOURNAL OF GERIATIUC PSYCHIATRY

small relative to the dementia-depressed
subgroup. The most noteworthy difference between dementia-only and control
subjects was found on the vigor subscale.
The vigor subscale includes items such as
"energetic," Uactive,U and "lively;U and it is
understandable that the frequency of
these mood states would decline in individuals with cognitive dysfunction as a result of limitations imposed on activities of
daily living.
In contrast, dementia patients with
abnormal GDS scores were found to have
abnormal scores on all DABS measures except AEI. As can be seen in Figure 1, lower
scores on positive affects and substantial
elevations on the negative subscales characterized this subgroup. Mean subscale
scores fell from 1.5 to 3.6 standard deviations (5Ds) outside of the average range.
303

Mood Assessment in Dementia
FIGURE 1. T scores of the Derogatis Affects Balance. Scale (DABS) for denlcntia patients and normal
elderly control subjccts.

Score
80 - - - - - - - - - - - - - - - - - - - - - ,

~-------.....,

~

::

Control subiects

::::::::::::::::::~::::::::........'- -:_g_:~_:_~_: ~_:_~ _~_~e _~_~ _s _..e.d.
~

50 .. ·._

.

40

30-.~••••••••••••••••••••• _ •••••••••
20

>0
..,

c

~

OJ

E
"E
QJ
'E
0

U

Note:

....
0

c

:>

~

m

0

£
-«

~
Q)

'x
c:

<

t:
0

•iii
VI

-g

·5
C)

.~

~

a.

eu

C

a..

.~ iii
<
"0

tn

CD

Z

The Tscore scale has a mean of 50 and a standard deviation of 10. ABI = Affects Balance Index.

These intergroup differences could be the
result of both dementia and depression.
However, on the majority of subscales t as
well as POS, NEG, ABI, and PARt the depressed subgroup means were also significantly different from dementia-only
patients. Therefore, even within a demen..
tia-depression sample, the DABS appears
to be sensitive to diminished positive affectivity and increased negative affectivi~
In brief, whereas the expected group
means on the DABS were found in healthy
elderly control subjects, the dementia patients, as a group, evidenced significantly
reduced positive mood rankings and elevated rankings on negative mood states.
The variances associated with these pa..
tieut sample means were also quite high
compared with control subjects, suggest..
ing a higher degree of variability in mood'
frequency ratings. When the demen304

0

J:

Q)

>

'.t=

'Vi
0

tia-depression sample was separated by
normal vs. abnormal GDS score, nondepressed subjects obtained scores approxi..
mating the normal range, whereas those
with depression evidenced abnormal frequencies of positive and negative mood
states.
Test-Retest Stability of the DABS in
Normal and Dementia Samples

In evaluating the psychometric aspects of the DABS, our first task was to
assess its temporal stability in both dementia patients and healthy control
subjects. This was accomplished via
test-retest Pearson r correlation coefficients calculated for each of the DABS
summary scores, as well as t-test comparisons of group means at Test 1 and Test 2.
Beginning with the control group
VOLUME 4 • NUMBER 4 • FALL 1996

Benedict et al.
TABLE 2.

Test-retest means ± standard deviations and reUability coefficients for the Derogatis
Affects Balance Scale (DABS)

Normal Control Subjects (II
Positive Mfects Total (POS)
Negative Mfects Total (NEG)
Mfects Balance Index (ABI)
Affects Expressiveness Index
Positive Affects Ratio (PAR)

Dementia Group (II

= 13)

Positive Affects Total (POS)
Negative Meets Total (NEG)
Affects Balance Index (ABI)
Meets Expressiveness Index
Positive Mfeets Ratio (PAR)

= 21)

Test 1

56.9 ±
18.4 ±
1.9 ±
74.3 ±
0.76 ±

8.6

7.3
0.6
10.8
0.09

Test 1
41.5 ± 21.2
38.8 ± 21.7

0.1 ± 2.1
± 10.8
0.52 ± 0.26

80.4

Test 2

53.6 ±
16.1 ±
1.9 ±
69.6 ±
0.77 ±

10.7
8.2
0.8

9.0
0.11

Test 2
45.1
28.1
0.8
73.2

± 16.6

±
±
±
0.63 ±

20.1
1.8

8.6
0.25

ReUabillty Coefficient
0.81***
0.77***
0.87***
0.70***
0.81***

Reliability Coefficient
0.62*
0.71·*
0.68**
0.40
0.71**

Note: All test-retest mean score comparisons by t..test, using an alpha criterion of 0.01, were not
statistically significant.
*p < 0.05; **p < 0.01; ***p < 0.001; all significance levels for Pearson r.

data shown in Table 2, it can be seen that
the follow-up examination means fell con·
sistently within 0.5 SD of the Test 1 score
on all DABS measures. Using an alpha criterion ofP < 0.01, none ofthe paired..sample
t..tests were statistically significant. The
test-retest reliability coefficients were also
well within acceptable psychometric limits.
The test-retest interval in the patient
group was smaller (28 vs. 50 days) than in
the control group, but the fact that the patients were heterogeneous with respect to
psychiatric symptoms and undergoing
change in psychiatric symptoms resulted
in a more conservative test of the DABS's
reliability. Again, t-test comparisons of
group means were not statistically significant on any of the five measures. The
test-retest correlations, however, were
considerably more modest. The reliability
coefficient was only 0.40 for AEI, indicating that this measure is not reliable in a
dementia sample. On the other hand, the
reliability coefficients were more acceptable on DABS measures reflecting mood
valence, ranging from 0.62 to 0.71. These
coefficients may be considered adequate,
given the wide range of medications that
were used for treatment of this sample,
the various psychiatric symptoms present
THE AMEIUCAN JOURNAL OF GERIATRIC PSYCHIATRY

on admission, and the fact that some individual patients were likely improving
psychiatrically; whereas others were not.
Validity

In order to further examine the construct validity ofthe DABS, the eight mood
subscale scores were submitted to a principal-components analysis with varimax
rotation. The two-factor solution obtained
from the control group, accounting for
69% of the variance in the matrix, is presented in Table 3. The two factors are easily
interpreted. Factor 1, which accounted for
43% of the variance, is the Positive Mfects
factor. Factor 2, the Negative Affects factor,
accounted for an additional 26% of the
variance. The factors are strikingly or..
thogonal, with a complete absence ofsplit
loadings (Le., very low correlations between mood states and factors on which
they do not load significantly). This is also
the same factor solution obtained in previous research with the DABS using a
younger normal sample. 17,18
A very similar two-factor solution was
obtained from the dementia group DABS
scores, this time accounting for 73% of the
variance in the matrix (Factor 1 accounted
305

Mood Assessment in Dementia
for 55% of the variance, and Factor 2 accounted for 18% of the variance). There
was one split loading, found on the depression scale. Thus, the mood state, depression, appears to share variance with
both the positive and negative factors. The
remaining seven mood variables, howTABLE 3.

ever, are uniquely positive or negative.
In the final analysis, Pearson r correlations were calculated to gauge the relationship between affective constructs
measured by the DABS and the depression
inventories. This analysis was limited to
the patient sample because of the re-

Factor analysis matrices for normal control subjects and dementia patients
Factor 1

Factor 2

0.91919
0.87525
0.76851
0.82644
0.13049
-0.24836
-0.07951
-0.18808

-0.19685
-0.15880
-0.13300
0.10673
0.73627
0.77891
0.80022
0.78794

0.84967
0.80593
0.78103
0.83249
-0.07155
-0.41615
-0.28400
-0.15298

-0.33429
-0.10472
-0.10749
0.86720
0.80049
0.84585
0.73905

Control Subjects

JoY

Contentment
Vigor
Mfection
Anxiety
Depression
Guilt
Hostility

Dementia Patients

Joy
Contentment
Vigor
Mfection
Anxiety
Depression
Guilt
Hostility

-0.36775

Note: Values are the resulting factor matrices after varimax rotation. Italicized coefficients are those that
are more than twice the value of the associated coefficient from the opposing factor.

TABLE 4.

Correlations (r) between the Derogatis Affects Balance Scale (DABS) and criterion
measures in the dementia sample
Geriatric Depression Scale (GDS)
Entire Sample
MMSE > 24
MMSE < 20
(N = 51)
(II = 16)
(II = 13)

DABS
Positive Total (POS)
Negative Total (NEG)
Affects Balance Index (ABI)
Mfccts Expressiveness Index
Positive Affects Ratio (PAR)
DABS
Positive Total (POS)
Negative Total (NEG)
Mfects Balance Index (ABI)
Affects Expressiveness Index
Positive Affects Ratio (PAR)

Note:

306

MMSE

-0.62
-0.80
-0.50
0.69
0.80
0.62
-0.75
-0.82
-0.63
0.07
0.28
0.06
-0.74
-0.82
-0.56
Montgomery-Asberg Depression Scale (MADS)
Entire Sample
MMSE> 24
MMSE < 20

-0.65
0.58
-0.70
0.08
-0.73

-0.82
0.85
-0.85
0.38
-0.85

-0.74
0.49
-0.70
0.41

-0.63

= Mini-Mental State Exam.
VOLUME 4 • NUMBER 4 • FAll 1996

Benedict et all
stricted range of GDS and MADS scores
among control subjects. Obviousl~ associations between tests such as the GDS
and MADS, which are designed to measure
signs and symptoms ofmajor affective disorder, and the DABS would be difficult to
uncover in a sample of subjects who are
free of depression.
Table 4 summarizes the results of the
correlational analysis. A5 was expected,
there was no relationship between the AEI
and either of the depression scales. Correlations between the POS, NEG, ABI, and
PAR measures from the DABS and the depression scale scores were in the expected
direction and consistent across depression scales. Overall, magnitude of correlations was moderate, ranging from 0.58
(NEG and MADS) to-0.75 (ABlandGDS).
To investigate the role of dementia in
these correlations, we compared the validity coefficients of 13 patients with
MMSE scores under 20 to validity coefficients obtained from 16 patients with
MMSE scores above 24. The correlations
for these mild- and severe-dementia subgroups are also presented in Table 4. It can
be seen that for POS, NEG, ABI, and PAR,
the validity coefficients are smaller for the
subgroup with severe dementia. For the
GDS, the largest mild- vs. severe-dementia
discrepancy in correlations was -0.80 vs.
-0.50 on the POS measure. When the
MADS is considered, the largest mild- vs.
severe-dementia difference is .0.85 vs.
0.49, on the NEG measure. When individual mood subscales were examined in the
severe-dementia subgroup, GDS validity
coefficients were less than 0.5 on the contentment, vigor, affection, and anxiety
subscales. The MADS validity coefficients
were less than 0.5 on the anxiet}', depression, guilt, and hostility subscales.

DISCUSSION
Despite the wide range of extreme mood
states and psychiatric symptoms that oc-

cur in primary progressive dementia, the
formal psychometric assessment of positive mood states has received little attention in the literature. Specific psychiatric
symptoms, such as depression and paranoia have been well documented in patients with dementia, but little is known
about the influence of the full spectrum
of positive and negative moods on shortor long-term psychosocial outcomes. In
order to facilitate research on this topic,
we attempted to determine whether a
commonly used psychometric test of affects balance could be used in a sample of
dementia inpatients recently admitted for
an acute episode of emotional and/or behavioral disturbance. Overall, we conclude that the modified administration
format of the DABS yields reliable and
valid data regarding fluctuating mood
states in both normal elderly subjects and
geriatric psychiatry patients with mild-tomoderate degrees of dementia.
The data obtained from the geriatric
control sample was consistent with previous DABS normative studies supporting
its use in elderly subjects. The mean
subscale scores obtained from our healthy
'geriatric subjects were all within 0.5 SDs
of the mean score obtained from the original standardization sample. 17,18 There..
fore, the current method of obtaining
T-score eqUivalents appears to be valid for
geriatric subjects, as well. In 21 normal
elderly subjects taking the DABS on two
separate occasions separated by a 50-day
interval, test-retest correlations were well
within the acceptable range. The con..
struct validity of the DABS was also replicated in this elderly sample. The last
solution, depicting two well-defined orthogonal constructs representing positive
and negative affects, is consistent with previous research using the DABS, as well as
the structural, two-factor model of
mood. It This stud~ therefore, extends
previous normative research with the
DABS to the geriatric population.
In dementia-depression inpatients,

THE AMERICAN JOURNAL OF GERIATRIC PSYCHIATRY

307

Mood Assessment in Dementia
the psychometric properties of the DABS
were found to be more questionable, althought on the whole, the results support
its use in patients with mild-ta-moderate
dementia. Group comparisons revealed
abnormal scores in this sample of 51 dementia-depression patients. As expected,
patients reported a much lower frequency
of positive mood states and a higher frequency of negative mood states than did
age-matched control subjects. Of course,
these group differences could be caused
by dementia and/or affective symptoma..
tolo~ Yet, when the dementia group was
divided according to GDS score, only
those patients with depression were markedly abnormal on the DABS. These subgroup differences cannot be explained by
the degree of dementia because the subgroups had equivalent MMSE scores. The
ability of the DABS to discriminate dementia patients with and without depression
suggests that the test would be useful in
clinical samples.
As for the reliability of the DABS
among dementia-depression patients,
most of the test-retest correlations were
in the acceptable range t which is remarkable, given that many of the patients were
in an acute state of confusion, agitation,
and/or psychosis when tested the first
time. However, these data are from a very
small sample of patients (n = 13), and so
these findings should be regarded as prelimina~

The construct validity of the DABS was
supported in the analyses using dementia-depression patients. In the principalcomponents analysis, the distinction
between individual DABS subscales was
somewhat less clear than in control subjects, but the final varimax solution was
very similar to that of the control group.
It is possible that with increasing levels of
dementia, the two factors that underlie
frequency ratings are less independent
and more analogous to a single, unidimensional, positive/negative scale. Also,
the cognitive dysfunction associated with
308

increasingdegrees ofdementia might sim..
ply interfere with the test-taking skills of
patients. For instance, it is conceivable
that some patients with severe dementia
would have difficulty understanding certain mood adjectives, or be unable to recall accurately the frequency with which
they have experienced certain moods.
Supporting this conclusion were the discrepancies among validity coefficients favoring those obtained from patients with
MMSE scores greater than 24, compared
with those with MMSE scores lower than
20. We might add that, based on our clinical observations, it is our general impression that the DABS yields extreme ratings,
or is simply not tolerated by patients who
have severe dementia (MMSE scores less
than 17). Overall then, although our data
support the reliability and validity of the
modified administration of the DABS in
patients with mild-to-moderate Alzheimer's or vascular dementia, the DABS
should be interpreted cautiously or not be
used in evaluations ofpatients with severe
dementia.
One should bear in mind that our
findings are based on an adapted version
ofthe DABS, and not the standard administration described by Derogatis. 17 Our interview format, coupled with the visual
scale to assist in patient rankings, was felt
to be instrumental in our ability to obtain
valid results in patients with mild dementia. We had previously found that when
the DABS was administered as a standard
self-report instrument t patients often responded randomly or were confused by
certain mood adjectives, such as the terms
"warm" or "vigor." Orally presenting each
mood adjective helped to facilitate questions regarding unclear items or confusion with the rating scale. Given that the
modified version of the DABS requires
only slightly more time than the standard
version (we have found that the test can
be completed by a patient with moderate
dementia in 15 to 25 minutes), we recommend its use in all geriatric subjects"
VOLUME 4 • NUMBER 4 • FALL 1996

Benedict et at.
Finally; instruments such as the modified DABS, which can assess the full
spectrum of mood balance changes in dementia, may facilitate research designed
to determine the predictive value ofmood
states. An interesting question for future
study is whether mood changes are related to behavioral problems, such as
night wandering, screaming, threats of
violence, and other behaviors that are so
distressing to caregivers. One might hypothesize that in the early stages of progressive dementia, positive mood states
may facilitate psychosocial adaptation.
The patient with mild dementia, even if
aroused and activity-seeking, has a preserved capacity for guiding behavior in a
logical, goal-directed fashion. On the
other hand, for the psychiatrically disturbed patient in the middle or late stages
of DAT or VaD, elevated positive moods
could underlie a general state ofincreased
arousal, which, when combined with cognitive dysfunction, leads to a sequence of
misdirected abnormal and disturbing be-

haviors. Such a patient would likely have
difficulty adjusting to a milieu with few
structured activities and limited staff
resources. We might further predict that
high levels of negative affects t in particular, hostili~ would predict poor psychosocial outcomes in dementia-depression
patients regardless of the degree of dementia. If certain mood states are predictive of psychosocial outcomes in patients
at risk for disruptive behaviors, psychiatric
interventions could be designed to decrease the frequency ofthese troublesome
mood states before patients are discharged. The reliable and valid assessment
of positive as well as negative moods and
the balance between them will provide a
means for investigating these and related
questions.
A copy ofthe visual analog scale used in

the study and further information pertaining to the modified, interview-based
format ofthe DABS is available from Dr.
Benedict on request.

References
I.Alexopoulos GS, Abrams RC: Depression in
Alzheimer's disease. Psychiatr Clin Noeth Am
1991; 14:327-340
2. Burns A: Affective symptoms in Alzheimer's dis·
easc. International Journal of Geriatric Psychia·
try 1991; 6:371-376
3. Burns A: Psychiatric phenomena in dementia of
the Alzheimer type. Int Psychogeriatr 1992;
4:43-54
4. Starkstein SE, Robinson RG: Affective disorders
and cerebrovascular disease. Br J Psychiatry
1989; 154:170-182
5. Roth M: The natural history of mental disorder
in old age. J Ment Sci 1955; 101:281-301
6. Cohen..Mansfleld J: Agitated behaviors in the
elderly, II: Preliminary results in the cognitivcly
deteriorated.) Am Geriatr Soc 1986: 34:722-727
7. Duetsch LH, Rovner BW: Agitation and other
noncognitive abnormalities in Alzheimer's disease. PsychiatrClin NorthAm 1991; 14:341-351
8. George LK, Gwythcr LP: Caregiver ,vell-being,
a multidimensional examination of family
caregivers of demented adults. Gerontologist
1986; 26:253-259
9. Reifler BV, Larson E, Teri L, et al: Dementia of
the Alzheimer's type and depression. J Am
THE AMERICAN JOURNAL OF GERIATIUC PSYCHIATRY

Geriatr Soc 1986; 34:855-859
10. Watson D: Intraindividual and intcrindividual
analyses of positive and negative affect: their
relation to health complaints, perceived stress,
and daily activities. J Pees Soc Psychol 1988;
54: 1020-1030
11. Watson 0, Tellegen A: Toward a consensual
structure of mood. Psychol Bull 1985; 98:219235
12. Beck AT, Ward CH, Mendelson M, ct al: An
Inventory for measuring depression. Arch Gen
Psychiatry 1961; 4:561-571
13. Yesavage JA: The use of self-rating depression
scales in the elderly, in Handbook for Clinical
Memory Assessment of Older Adults. Edited by
Poon LW. Washington, DC, American Psychological Association, 1986, pp 213-217
14.Zung WWK: A self-rating depression scale. Arch
Geo Psychiatry 1965; 12:63-70
15. Parmelee PA, Lawton MP, ~~tz IR: Psychometric
properties of the Geriatric Depression Scale
among the institutionalized aged. Psychological
Assessment: A Journal of Consulting and Clini..
cal Psychology 1989; 1:331-338
16. Snowdon J, Donnelly N: A study of depression
in nursing homes. J Psychiatr Res 1986;
309

Mood Assessment in Dementia
20:327-333
17.Dcrogalis LR: The Dcrogatis Mfects Balance
Scale. Baltimore, MO, Clinical Psychometric Rescarch, 1975
18. Dcrogatis LR. Rutigliano PJ: Derogatis Mfects
Balance Scale: DABS, in Quality of Life and Pharmacoecononlics in Clinical Trials. Edited by
Spilkcr B. Nc\v York, Raven, 1995, in press u:i?4.
19. Rabins PV t Fitting MDt Eastham j, ct al: The
emotional impact of caring for the chronically
ill. Psychsomatics 1990; 31:331-336
20. \VolfTM, Elston Re, Kissling GE: Relationships
of hassles, uplifts, and life events to psychological well-being of freshmen medical students.
Behav Mcd 1989; Spring:37-45
21. Hochn-Sark R, Merchant AF, Keyser ML, et al:
Effccts of clonidinc on anxiety disorders. Arch
Gen Psychiatry 1981; 38:1278-1282
22. American Psychiatric Association: Diagnostic
and Statistical Manual of Mental Disorders, 4th
Edition. 'Vashington t DC, American Psychiatric
Association, 1994
23. McKhann G, Dracholan D, Folstein M, et al:
Clinical diagnosis ofAlzheimer's disease: rcport
of the NINCDS-ADRDA \vork group under the
auspices of the Department of Health and Hu..
man Services Task Force on Alzheimer's Diseasc. Neurology 1984; 34:939-944
24. Roman Ge, Tatemichi TK, Erkinjuntti T, ct al:
Vascular dementia: diagnostic criteria for research studies: report of the NINDS·AIREN
International Workshop. Neurology 1993; 43:
250-260

310

25. Folstcin MF, Falstein S, McHugh PR: Mini-Mental State: a practical method for grading the
cognitive state of patients for the clinician. J
Psychiatr Res 1975; 12:189-198
26. Borko\vski JG, Benton AL, Spreen 0: 'Vord fluency and brain damage. Neuropsychologia
1967; 5:135-140
27. Rosen WG: Verbal fluency in aging and demcntia. Journal of Clinical Neuropsychology 1980;
2:135-146
28. Kaplan EF, Goodglass H, \~cintraub S: Thc Boston Naming Test, 2nd Edition. Philadelphia, PA,
Lea and Fcbiger, 1983
29. Beery KE, Buktcnica NA: Revised Administration, Scoring, and Teaching Manual for the Dcvelopmental Test of Visual-Motor Integration.
Cleveland, 01-1, Modern Curriculum Ilress, 1982
30. Itcitan RM: Validity of the Trail Making Test as
an indicator of organic brain damage. Percept
Mot Skills 1958; 8:271-276
31. Brandt J: The Hopkins Verbal Learning Test:
development of a nc\v memory test \vith six
equivalent forms. Clinical Neuropsychologist
1991; 5: 125-142
32. Benedict IU-IB, Groninger L: Preliminary standardization and validation of a nc\v visuospatial
memory test \vith six alternate forms. Clinical
Neuropsychologist 1995; 9: 11-16
33. Parmelee PA, Katz IR: Geriatric Depression Scale
(letter). J Am Gcriatr Soc 1990; 38:1379
34.Asberg M, Montgomery SA, Perris C, ct al: A
comprehensive psychopathological rating scale.
Acta Psychiatr Scand 1978; 271 (suppl):S-27

VOLUME 4 • NUMBEH 4 • FALL 1996