Assessment of Behavioral
Symptoms in CommunityDwelling Dementia Patients
Myron F. Weiner, M.D.
Brent Williams, M.A.
Richard C. Risser, M. S.
The autbors compared the CERAD Behavior Rating Scale
for Dementia (CBRSD) with the Cohen-Mansfield Agitation Inventory (CMN) for their ability to detect behavioral
symptoms in community-dwelling dementia patients
with mild-to-moderate global impairment. Both instruments were administered to caregivers of 33 cognitively
impaired patients seen in a dementia clinic at initial
evaluation or folloui-up visit. Endorsement of a higher
percentage of items on the CBRSD than the CMAI suggests
greater sensitivity of this instrument to the behavioral
symptoms seen in community-dwelling patients. There
was good correlation between the number of items endorsed on both scales but not between subscales of the
CMAI and factors of the CBRSD that appeared related to
agitation. Thus, the CBRSD and CMAI both seem to measure behaviors that occur in dementia patients, but the
CBRSD's two agitation-related factors do not appear to
measure agitation as defined by the eMAI. (American] OUfnal of Geriatric Psychiatry 1997; 5:26-30)

T

he quantification of behavioral symptoms in dementia patients is important for the assessment of behavioral
intervention strategies and for the assess..
ment of psychotropic and cognition-enhancing drug effects on such symptoms.
Scales for the evaluation of psychiatric
symptoms in cognitively intact persons

are not useful for assessing behavioral
symptoms in dementia patients because
they are based on self-report or clinical observation. Scales based on direct clinical
observation are not useful because the behaviors are often infrequent. Self-report
scales are not useful because ofcognitively
impaired patients' inability to report on

Received August 2, 1995; revised December 12, 1995; accepted May 9, 1996. From the Departments of
Psychiatry and Neurology, University of Texas Southwestern Medical Center, DaIL'1s, TX. Address correspondence to Dr. Weiner, 5323 Harry Hines Blvd., Dallas, TX 75235-9070.
Copyright © 1997 American Association for Geriatric Psychiatry

26

VOLUME 5 • NUMBER 1 • WINTER 1997

Weiner et al.
their own behavior.' For these reasons, informant-based scales, such as the CERAD
Behavior Rating Scale for Dementia
(CBRSD) and the Cohen-Mansfield Agitation Inventory (CMAI) have been developed.
We compared the CBRSD, a new
instrument, with a commonly used instrument, the CMAI, to determine the sensitivity of each instrument to behaviors in
community-dwelling Alzheimer's disease
(AD) patients with mild-to-moderate dementia. The CMAI 2 measures so-called agitated behaviors in nursing home residents
with and without a formal diagnosis of dementia. It was chosen as the instrument
for comparison because it was developed
without reference to psychopatholop or
psychiatric syndrome. The CBRSn3 , was
developed by the Consortium to Establish
a Registry for Alzheimer's Disease
(CERAD)5 as a broad-based scale to quantify behavioral and emotional symptoms
in community..d welling AD patients.

METHODS
Subjects

Subjects were a convenience sample
of 33 consecutive caregivers of community-dwelling individuals seen for initial
evaluation or in regular follow-up visits at
the Clinic for Alzheimer's and Related Diseases at the University ofTexas Southwestern Medical Center. Our evaluation
process is described els ewhe re ,"
Caregivers included spouses and children
of patients. The CMAI7 sB and CBRSD were
administered to caregivers during the
course of their patient's visit by a clinician
(BW)well trained in the administration of
the CMAI and CBRSD. For every caregiver,
the CMAIwas administered first. The origi4
nal version of the CBRSn was used. Responses were coded to maintain patient
confidentiality: Weestimated that a sample
of about 30 subjects would provide better
THE AMERICAN JOURNAL OF GERIATlUC PSYCHIATRY

than 90% power to detect moderately high
correlations (at least r = 0.70) between
measures as opposed to low-incidence
(1· = 0.30 or less) alternatives.
Descriptive statistics concerning age,
duration of symptoms, onset age, MiniMental State Exam (MMSE)9 score, and
Clinical Dementia Rating (CDR) 10 score
for the 33 patients are provided in Table
1. As suggested by MMSE and CDR scores,
the patients were mildly to moderately impaired. The 33 patients included 19
women (58%) and 14 men. Using NINCDS
criteria for the diagnosis of AD, 24 of the
33 patients (73%) were classified as probable (18) or possible (11) AD. Some patients had various other diagnoses,
including prodromal AD (3), unclassifiable dementia (2), progressive dysphasic dementia (1), and cognitive
impairment without dementia (2).

Instruments
The CMAI is a compilation of "agitated" behaviors observed in nursing
home residents. In the development of
this scale, agitation was defined as "inappropriate verbal, vocal, or motor activities
not explained by apparent needs or confusion.,,2 The scale is well anchored, and
its reliability and validity are well estab..
lished. 6,7 The informant is the patient's
caregiver. Symptoms are assessed for the
preceding 2 weeks. There are 37 items and
3 agitationsubscales: physically aggressive
(directed against a person or object),
physically nonaggressive (not directed
against a person or object, such as pacing
and wandering), and verbal. Items are
scored on a seven-point scale: 1 = never;
2 = < once a week; 3 = 1-2 times/week;
4 = several times/week; 5 = 1-2 times
per day; 6 = several times/day; 7 = several times/hour. Severity is not rated on
this scale.
The CBRSDwas designed to assess AD
patients. It samples a wide range ofbehaviors and psychopathology and is intended
27

Behavioral Assessment in Dementia
TABLE 1.

Descriptive statistics: dementia clinic outpatients
Minimum

Age, years
Age at onset of illness
Duration of illness, years
MMSE score
CDR scale
Note:

SD

TABLE 2.

73.6
69.5
4.1
16.5
1.3

CDR scale

= Clinical Dementia Rating.

CERAD Behavior Rating Scale for Dementia (CBRSD): summary scale scores of items
overall and for each factor

Total
Depressive
Psychotic
Defective
Self-Regulation
Irritation/Agitation
Vegetative
Apathy
Aggression
Affective Lability

Maximum
Attainable Score

Maximum
Score

Minimum
Score

Mean ± SD

48
7
6
10

168
28
20
37

22
0
0
0

112
26
23
28

48.8
9.7
6.1
10.5

±
±
±
±

20.3
6.0
4.8
6.6

4
4
4
4
4

16
10
10
16
13

0
0
1
0
0

16
10
10
12
12

7.4
5.6
7.2
2.9
5.5

±
±
±
±
±

4.7
3.0
2.5
3.1
3.5

SD = standard deviation.

TABLE 3.

Cohen-Mansfield Agitation Inventory (CMAI): summary scale scores of items overall
and for each subscale
No. of
Items

Total
Physical Aggression
Non-physical Aggression
Verbal Agitation

37
12

10
8

Maximum
Attainable Score
222
72
60
48

for use as a structured caregiver interview:
Items were selected from a literature review and consultation with experts in the
field. The items are well-anchored, and
most are homogeneously scaled. The
authors of the scale chose to quantify only
frequency of behaviors because severity
judgments are more difficult to anchor
and thus appear to be less reliable. Based
on a pilot study of 303 subjects with
NINCDS probable AD,11 the scale is under
evaluation in further studies and will then
be released for general use.
28

90
87
15
25
3

6.2
7.2
3.0
5.6
0.7

= standard deviation; MMSE = Mini-Mental State Exam;

No. of
Items

Note:

±
±
±
±
±

Maximum

59
52
1
5
1

Mean ± SD

Maximum
Score

Minimum
Score

58
2
26
24

1
0
0
0

Mean ± SD
25.2
0.1
6.4
9.9

±
±
±
±

13.2
0.4
6.6
5.9

The scale assesses behavior over the
preceding month, but also notes behaviors over 1 month ago and since the onset
of dementia. It is administered to a caregiver. Of the 48 items, 40 are rated as 0 =
has not occurred since illness began; 1 =
present 1-2 days in the last month; 2 =
3-8 days; 3 = 9-15 days; and 4 = > 16
days. Factor analysis of the initial study
suggested 8 factors: depressive symptoms,
psychotic symptoms, defective self-regulation, irritability/agitation, vegetative features,
apathy, aggression, and affective lability.
VOLUME 5 • NUMBER 1 • WINTER 1997

Weiner et al.
There are no firmly established traditions for the scoring of the CMAI, and no
recommendations have yet been made for
scoring tile CBRSD. For the purpose of
analysis, we summarized the responses
for both instruments in terms of number
of items endorsed instead of using the
scale values associated with the items.
Standard correlation analysis was used to
compare responses for the two instruments.

RESULTS
Descriptive statistics for the endorsement
ofCBRSD items and its subscales (factors)
and the CMAIand its subscales are shown
in Tables 2 and 3, respectively. There was
a significant correlation between the
number of items endorsed on both scales
(r = 0.480; P = 0.005). Total CBRSD en..
dorsements correlated significantly with
endorsements on the CMAI Physically
Nonaggressive subscale (r = 0.457; P =
0.007) and Verbal Agitation subscale (r =
0.397; P = 0.022). There were too few re ..
sponses among the CMAI Physically Aggressive scale items (only two of the 33
subjects endorsed items for this subscale)
to produce a meaningful correlation with
the CBRSD total and subscales.
Correlations between the number of
CMAI items endorsed and the endorse..
ments to CBRSDsubscales indicated some
TABLE 4.

significant and some nonsignificant relationships.. The CMAI total correlated sig..
nificantly with CBRSD subscales for
psychotic features (1· = O.411;P = 0.017)
and defective self-regulation (r = 0.387;
P = 0.0269). Other subscale correlations
were not significant. Notably; the CMAI endorsements displayed a marginal correlation with endorsements on the CBRSD
scale for irritability/agitation (r = 0.332;
P = 0 . 06). Also, the CMAI subscale endorsements did not correlate with the
CBRSD Irritability/Agitation factors (r =
0.101; P = 0.58) or for the CMAI Nonphysical Aggression subscale (r = 0.263;
P = 0.139) for the CMAI Verbal Agitation
subscale. Finally; we examined the correlations of the endorsement of CBRSD
items pertaining to aggression and agitation across items on the CMAI. Table 4
shows the average as well as the highest
inter-item correlations for these seven
CBRSD items with all CMAI items and the
items contained in the CMAI Verbal
Agitation and Nonphysical Aggression
subscales. For example, the CBRSD item
"agitated or upset" had an average correlation of-0.020 across all CMAI items and
an average correlation of-O.064 across the
eight CMAI Verbal Agitation items. These
various inter-item correlations indicate
that there are frequently low as well as
negative correlations among endorse..
ments to the agitation..s pecific CBRSD
items and the CMAI items in general.

Average (and highest) inter-item correlations (r) pertaining to Agitation and Aggression:
average (highest) CBRSD correlation with CMAI items
All CMAI
Items

Endorsed
CBRSD Item:
Easily irritated
AgitatedIUpset
Verbally aggressive
Physically aggressive
Abandonment
Uncooperative
Sudden changes in emotion

0.105 (0.410)
-0.003 (0.467)
0.116 (0.492)
0.066 (0.348)
0.147 (0.609)
0.168 (0.495)
0.047 (0.398)

THE AMERICAN JOURNAL OF GEIUATIUC PSYGI=lIA-TRY..

CMAIVerbal
Agitation Items
Endorsed
0.104
-0.027
0.138
0.124
0.152
0.216
0.035

(0.320)
(0.220)
(0.492)
(0.348)
(0.605)
(0.495)
(0.276)

CMAI Non-physical
Aggression Items
Endorsed
0.052
-0.129
0.045
-0.030
0.279
0.092
-0.023

(0.284)
(0.250)
(0.340)
(0.231)
(0.609)
(0.175)
(0.398)

29

Behavioral Assessment in Dementia
DISCUSSION
The median number of CBRSD items en..
dorsed overall and for its agitation factor
were higher relative to the number of
items (20 of 48 and 3 of4) than the median
number of CMAI items endorsed relative
to the total number of items (7 of 37).
Thus, the CBRSD appears sensitive to
lesser degrees of behavioral disturbance
than the CMAI. This sensitivity may be re ..
Iated to the I-month time frame of the
CBRSD, compared with the 2-week observational window of the CMAI.
We fail to find an overwhelming cor..
relation between the endorsements of the
CBRSD items pertaining to agitation and
the items endorsed on the CMAI and its
subscales. Our sample of 33 caregivers
did, however, provide adequate power to
find moderately high correlations between measures against low-incidence al..
ternatives. We documented significant
correlations between the two total scores
and between some subscales/factors, Because direct physical aggression against
persons or objects was rare in these 33
subjects, we could not identify a relation-

ship between the instruments for physical
aggression. Thus, the modest concurrent
validity found is partly because of our
community-dwelling population of persons with mild-to-moderate deme.ntia and
only mild behavioral symptoms. It is possible that administering the CMAI first and
the CBRSD second and the use of a single
interviewer introduced bias that limits the
generalizability of our study:

CONCLUSION
The significant correlation of the CBRSD
with the CMAI suggests that the CBRSD
measures behavioral symptoms that occur
in community..dwelling dementia patients
with mild..to-moderate global impairment. It remains to be established
whether the CBRSD is sensitive to behavioral changes brought about by various
treatment strategies.

This work was supported in part by National Institute on Aging Grant I-P30AG12300-01.

References
1. Weiner MF, Koss E, Wild KY, et at: Measures of
psychiatric symptoms in Alzheimer's disease pa-

tients: a review. Alzheimer Dis Assoc Disord
1996; 10:20-30
2. Cohen-Mansfield 1: Agitated behaviors in the
elderly, II: preliminary results in the cognitively
deteriorated. J Am Geriatr Soc 1986; 34:
722-727

3. Tariot PN, Mack 1L, Patterson MB, et al: The
CERAD Behavior Rating Scale for Dementia

(BRSD) (abstract). Gerontologist 1992; 32:160
4. Tariot PN, Mack JL, Patterson MB, et al: The
Behavior Rating Scale of the Consortium to Establish a Registry for Alzheimer's Disease. Am J
Psychiatry 1995; 152:1349-1357
5. Morris JC, Heyman A, Mohs Re, et a1: The consortium to establish a registry for Alzheimer's
disease (CERAD), part I: clinical and neuropsychological assessment of Alzheimer's disease.
Neurology 1989; 39:1159-1165
6. Weiner MF, Bruhn M, Svetlik OS, et a1: Experiences with depression in a dementia clinic. J

30

Clin Psychiatry 1991; 52:234-238
7. Finkel S, Lyons IS, Anderson RL: Reliability and
validity of the Cohen-Mansfield Agitation Inventory in institu tionalized elderly. International
Journal of Geriatric Psychiatry 1992; 7:487-490
8. Miller RJ, Snowdon J, Vaughan R: The use of the
Cohen-Mansfield agitation inventory in the assessment of behavioral disorders in nursing
homes. J Am Geriatr Soc 1995; 43:546-549
9. Folsteln MF, Foistcin SE, 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
10. Hughes CP, Berg L, Danziger WL:A new clinical
scale for the staging of dementia. Br J Psychiatry
1982; 140:566-572
11.McKhann G, Drachman D, Folstein M, et al:
Clinical diagnosis ofAlzheimer's disease: Report
of the NINCDS-ADRDA Work Group under the
auspices of the Department of Health and Human Services Task Force on Alzheimer's Disease. Neurology 1984; 34:939-944
VOLUME 5 • NUMBER 1 • WINTER 1997