Archivesof Clinical Neuto~ychology,Vol. I I, No. 6, pp.

529-539, 1996
Pergamon Colayright© 1996National Academyof Nemol~ycholo~
Printed in the USA.All rights reserved
0887-6177/96 $15.00 + .00
SSDI 088/-6177(95)00041-0

Screening for Dementia of the Alzheimer Type in

the Community: The Utility of the Clock
Drawing Test
Deborah A. Cahn, David P. Salmon, Andreas U. Monsch, Nelson Butters,
W. C. Wiederholt, and Iody Corey-Bloom
Departments of Neurosciences and Psychiatry, University of California, San Diego

Elizabeth Barrett-Connor
Department o f Family & Preventive Medicine, University of California, San Diego

This study reports the sensitivity and specificity o f the Clock Drawing Test (CDT) for detecting
dementia of the Alzheimer type in a coramunity-dwelling sample of elderly subjects. Forty-two
patients with clinically diagnosed Alzheirner's disease and 237 cognitively intact subjects were
administered the CDT as part of an epidemioiogical study of aging and dementi~ Three individual
measures of clock drawing performance (quantitative score, qualitative score, and combined quanti-
tative and qualitative score) were determined for each participant. When qualitative elements such as
errors and strategies were incorporated into the CDT score, the sensitivity was 84% and the specifici-
ty was 72%. The findings suggest that a CDT score which evaluates qualitative and quantitative fea-
tures provides reasonably good discrimination between normal elderly individuals and DAT patients.
However, the CDT appears to have limited _uti!i_~as a single screening instrument in the community.
Instruments such as the Dementia Rating Scale (Maltis, 1976) provide better discrimination of DAT,
indicating that functions such as memory and verbal fluency need to be assessed during screening.

The Clock Drawing Test (CDT) (Goodglass & Kaplan, 1972) has been proposed as a useful
neuropsychological screening instrument for dementia of the Alzheimer type (DAT)

Address correspondence to: David P. Salmon, PhD, Department of Neurosciences (0948), U.C. San
Diego, La Jolla, CA 92093-0948.
This work was funded in part by National Institute on Aging Grants AG0781 and AGO513 l-OgSl.
A.U.M. was supported by a La Roche Foundation (Switzerland) postdoctoral fellowship. This article
is based in part on the first author's dissertation submitted to the University of California, San Diego
and San Diego State University Joint Doctoral Program in Clinical Psychology in partial fulfillment of
the requirements for the degree of Doctor of Philosophy in Clinical Psychology. We thank Mel
Klauber, Sharon Edelstein, Sarah Cahn, and Edith Kaplan for their helpful comments and assistance
during the conduct of this study. Presented in part at the 22nd Annual Meeting of the International
Neuropsychological Society, Cincinnatti, OH, February 1994.

529
530 D. A. Cahn et al.

because of its time and cost efficiency and its sensitivity to the constructional, visuopercep-
tual, and conceptual deficits that are commonly observed in this disorder (Shulman,
Shedletsky, & Silver, 1986; Sunderland et al., 1989; Mendez, Ala, & Underwood, 1992;
Watson, Arfken, & Birge, 1993). Several studies have shown that performance on the CDT
is highly effective for discriminating between patients with DAT and normal elderly (NE)
individuals (Wolf-Klein, Silverstone, Levy, & Brod, 1989; Tuokko, Hadjistavropoulos,
Miller, & Beattie, 1992; Feedman et al., 1994). For example, Wolf-Klein et al. (1989)
reported rates of sensitivity (i.e., percentage of correctly identified patients) and specificity
(i.e., percentage of correctly identified normal control subjects) for the diagnosis of DAT of
86% and 92%, respectively, using only the CDT. Similarly, Tuokko et al. (1992) found that
the CDT discriminated between DAT and NE subjects with sensitivity of 92% and specifici-
ty of 86%.
Although previously reported diagnostic accuracy of the CDT is quite impressive, the effec-
tiveness of this test as a general screening instrument for dementia in the community remains
questionable. All of the previous studies employed clinically-based samples of self-selected or
pre-screened subjects who may not have been representative of the community-dwelling popu-
lation. The use of these clinically-based ,samples may have led to an overestimation of the true
diagnostic accuracy of the CDT when it is applied to a community-based sample in which
other medical disorders and subtle cognitive imp rment are prevalent. The true effectiveness
of the CDT for screening for DAT in the community can only be assessed by determining its
diagnostic accuracy in a representative ,sample of community-dwelling individuals (Beardsall
& Huppert, 1991).
The efficacy of the CDT as a screening instrument for DAT in the community may be
enhanced by considering the qualitative aspects of a subject's performance, in addition to
the single, quantitative rating of the drawing that is usually obtained. Qualitative analyses of
the types of errors produced on other neuropsychological tests have proven to be particular-
ly helpful in distinguishing subjects in the earliest stages of DAT from NE individuals
(Bondi et al., 1994).
Furthermore, qualitative assessment of the types of errors committed on the CDT have
been shown to be useful in differentiating between demented and nondemented individuals
(Kozora & Cullum, 1994) and between patients with different dementing disorders
(Rouleau, Salmon, Butters, Kennedy, & McGuire, 1992). Rouleau and her colleagues
demonstrated that DAT patients were more likely than NE subjects and patients with
Huntington's disease (HD) to make errors that reflected a deficit in adequately representing
the attributes, features, and meaning of a clock. In contrast, HD patients were more likely
than DAT and NE subjects to make errors indicative of visuospatial processing deficits. The
results of this study suggest that DAT patients have a propensity to make a particular and
unique pattern of errors on the CDT, and that consideration of the errors produced on the
task when screening for DAT in the community may augment the test's diagnostic effective-
ness, particularly when attempting to classify individuals who manifest very subtle cogni-
tive impairment.
In view of the lack of knowledge concerning the efficacy of the CDT as a screening
instrument for DAT in the community, and the possibility that consideration of qualitative
aspects of performance may enhance diagnostic accuracy, the present study had two goals.
The first was to empirically derive for three measures from the CDT (i.e., a quantitative
score, a qualitative score, and a combined quantitative-qualitative score) the effective cut-
off scores for differentiating between DAT patients and normal elderly individuals in a com-
munity-based population sample.
Because all previous studies of the efficacy of the CDT for identifying patients with DAT
employed clinically-based, self-selected, or pre-screened subjects, the present study was the
 Screeningfor Dementia With the Clock Drawing Test 531

first to examine the true diagnostic accurracy of this test in the general population. This
information is critical if the CDT is to be used as a screening instrument for DAT in general
clinical practice or in large scale epidemiological studies of dementia.
The second goal of the present study was to examine the CDT performance of subjects
who were identified by a diagnosing neurologist as being "at risk" for DAT because of subtle
cognitive impairment (detected on neuropsychological evaluation or mental status testing) in
the absence of clear functional decline. Although it is usually not possible to diagnose such
subjects on the basis of psychometric performance alone (Storandt & Hill, 1989), the qualita-
tive aspects of their performance on the CDT may provide important information about the
underlying nature of their mild cognitive deficits. An error profile similar to that of patients
with DAT may be indicative of the early stages of a dementing disorder, whereas a profile
similar to that of normal elderly may indicate a more benign age-associated cognitive decline.

METHOD

Study Population

The Rancho Bernardo Study population has been described in detail elsewhere (Criqui,
Barrett-Connor, & Austin, 1978; Wiederholt et al., 1993). Briefly, between 1972 and 1974,
5052 adults (representing 82% of the population of all adult residents aged 30 to 79) of
Rancho Bernardo, CA, were surveyed for heart disease risk factors. The entire cohort has
been followed with regard to health status to the present. Beginning in 1988, the 1800 local-
ly-residing surviving members of the cohort, who were aged 65 and older, were invited to a
clinic visit, and of these, 1349 completed a screening evaluation for cognitive functioning.
One hundred ninety-nine subjects who were identified by the screening evaluation as having
possible cognitive impairment according to operationaly defined criteria (Wiederholt et al.,
1993) attended a subsequent clinic visit to determine the actual clinical diagnoses. To con-
trol for the possibility of false negative errors, a sample (n = 203) of NE subjects was ran-
domly selected for the comprehensive evaluation from those subjects who had no cognitive
impairment according to the screening evaluation, with the same age and gender distribution
as the sample who had possible cognitive impairment according to the screening evaluation.
Thus, a total of 402 individuals attended a clinic visit where they were given the comprehen-
sive evaluation, which included neurological examination, neuropsychological assessment,
standard medical history and examination, and, in some cases, CT scans of the brain.

Determination of Neurological Diagnosis

The diagnosis of DAT was made according to the NINCDS-ADRDA criteria (McKhann
et al., 1984). The examining neurologist made a diagnosis based on his/her examination,
medical history, functional status, and the results of laboratory tests and CT scans. A second
neurologist independently reviewed all data from the clinical examination and the laborato-
ry results. Both examiners were blind to the specific scores of the neuropsychological tests
but were informed of a neuropsychologist's opinion as to whether the subject had two or
more areas of cognitive deficit. Any discrepancy in the diagnosis between the examining
neurologist and the reviewing neurologist was discussed in a monthly meeting attended by
all participating investigators, and a consensus diagnosis was determined. The diagnosis of
"at risk" (AR) for DAT was made when the subject demonstrated impairment in two or
more areas of cognition on either the neurological or neuropsychological examination or
both but had no obvious impairment in activities of daily living and had not demonstrated
532 D. A. Calm et al.

functional decline by clinical assessment or by the score on the Pfeffer Outpatient Disability
Scale (Pfeffer, Kurosaki, & Harrah, 1981).
Fifty-two individuals met NINCDS-ADRDA criteria for possible or probable Alzheimer's
Disease. Of the individuals diagnosed with DAT, 48 had been identified as possibly cogni-
tively impaired by the screening evaluation and 4 had been identified as not cognitively
impaired. Two-hundred-thirty-nine individuals were diagnosed as being cognitively-intact.
Of these NE individuals, 168 had been identified as not cognitively impaired by the
screening evaluation, and 71 had been identified as possibly cognitively impaired. Eighty-
two individuals were identified by the comprehensive examination as being AR for
Alzheimer's disease.
Of these subjects, 21 had been identified by the screening evaluation as not cognitively
impaired, and 61 had been identified as possibly cognitively impaired. Twenty-nine patients
met criteria for dementia due to other neurological conditions. The current study focuses on
the clock drawing performance of the NE, DAT, and AR individuals.
Ten DAT, 8 AR, and 2 NE subjects were excluded from the current analyses because of
incomplete data. Table 1 shows the demographic characteristics and Mattis Dementia
Rating Scale (DRS) (Mattis, 1976) scores for the NE, DAT, and AR groups. One-way analy-
sis of variance (ANOVA) revealed that the three groups differed significantly in age,
F(2, 350) = 24.18, p < 0.0001, and level of dementia as determined by the DRS, F(2, 350) =
190.11, p < 0.0001, but not in years of education, F < 1. Post-hoc Tukey tests (~ = 0.05)
revealed that the DAT patients and AR subjects did not differ significantly in age, and that
both of these groups were significantly older than the NE group (p < 0.05). The mean DRS
score of the NE group was significantly higher than that of the AR group (p < 0.05) and that
of the DAT group (p < 0.05). The mean DRS score of the AR group was significantly high-
er than that of the DAT group (p < 0.05).

Procedure

As part of the extensive neuropsychological test battery, the CDT was administered to
each subject with the instructions to "Draw a clock, put in all of the numbers, and set the
hands for 10 after 11 ." Two trained independent raters scored the drawings according to the
criteria developed by Rouleau et al. (1992). Briefly, a quantitative score is derived by
adding the scores of three separate features: the clockface (0-2 points), the placement of the
hands (0-4 points), and the placement of the numbers (0-4 points). Thus, a "perfect" clock
would receive a quantitative score of 10. A qualitative analysis of error types is also per-
formed in which the examiner notes the presence of conceptual, perseverative, stimulus
bound, and spatial arrangement errors. A description of these different error types is pre-
sented next.

TABLE 1
Demographic Characteristics and Mean Scores on the Dementia Rating Scale
(DRS) of Normal Elderly (NE) Subjects, Individuals at Risk for DAT, and
Patients With Dementia of the Alzheimer ~ype (DAT)

NE At Risk DAT
(n = 237) (n = 74) (n = 42)
Gender (F/M) 140/97 45/29 25/I7
Age in years 78.4 (6.8) 83.3 (5.5) 83.5 (4.4)
Yearsof education 13.9 (2.7) 13.9 (2.7) 13.6 (3.0)
DRS (max = 144) 136.9 (5.3) 127.9 (6.6) 115.8 (12.5)
Note. Standard deviationsare presented in parentheses.
 Screening for Dementia With the Clock Drawing Test 533

1. Stimulus bound response. The tendency of the drawing to be dominated or guided by a

single stimulus. There may be three types of stimulus bound errors:
(a) The hands may be set for 10 to 11 instead of 10 after 11.
(b) The time is written beside the 11 or between 10 and 11 on the clock.
(c) The hands are absent or are pointed toward "10" and/or "! 1". This type of error is
also considered to be a conceptual error.
2. Conceptual deficit. This error type reflects a loss or deficit in accessing knowledge of
the attributes, features, and meaning of a clock. Included in this category are
misrepresentation the clock itself (a clockface without numbers or inappropriate use
of numbers) and misrepresentation of the time on the clock (the hands are either
absent or inadequately represented or the time is written on the clock).
3. Perseveration. The continuation or the recurrence of activity without an appropriate
stimulus. In clock drawing, this error is seen in the presence of more than two hands
and abnormal prolongation of numbers (writing beyond "12").
4. Neglect of left hemispace. All attributes of the clock are written on the right half of the
clock face. Possible neglect of right hemispace was also evaluated, but this type of
error was never observed.
5. Planning deficit. This error type is represented by gaps before the 12, 3, 6, or 9,
depending on the strategy used in drawing.
6. Nonspecific spatial error. A deficit in the spatial layout of numbers, without any
specific pattern in spatial disorganization.
7. Numbers written on the outside of the clock. Numbers written either around the
perimeter of the circle or on the circle itself.
8. Numbers written counterclockwise. Arrangement of the numbers with "12" at the top
of the clockface and then continuing around in a counterclockwise fashion.

In the present study, the following three measures were derived:

1. Quantitative CDT score. Assesses the presence and correctness of the features of the
clock (Maximum = 10 points).
2. Qualitative CDT score. The number of error types, as described by Rouleau et al.
(1992) that were committed (Maximum = 8 error types). Because the error types were
scored only for their presence or absence, the total number of error types committed
provided a summary score of the qualitative errors.
3. Global CDT score. The CDT quantitative score minus the qualitative error score. This
score takes into account not only presence and correctness of the features of the clock, but
also the number of error types made and strategies used in the construction of the clock.

The two raters demonstrated high interrater reliability (r = 0.95, p < 0.0001) for the
Quantitative CDT score. This level of interrater reliability is consistent with the results of
previous studies (Nussbaum, Fields, & Starratt, 1992). A kappa statistic was calculated for
each of the eight error types by comparing the observed agreement with the agreement
expected by chance (50%) alone.
Of the 8 kappa values, 7 were in the range indicating excellent agreement beyond
chance, and 1 (nonspecific spatial arrangement) was in the range of fair to good agreement
beyond chance (Landis & Koch, 1977).

Statistical Methodology
A Receiver Operating Characteristic curve (ROC) analysis was performed to quantify
test accuracy and to characterize the distribution of test scores in the study samples. For
534 D. A. Calm et al.

the present analyses, each C D T score was treated as a possible cut-off score and the cor-
responding sensitivity (i.e., percentage of DAT patients scoring below this score) and
specificity (i.e., percentage o f NE subjects scoring above this score) for the diagnosis of
DAT were calculated. An ROC curve was then generated for each of the three C D T mea-
sures by plotting sensitivity against specificity for every possible cut-off score. Because
it was decided a p r i o r i that sensitivity and specificity were to be of equal importance,
the optimal cut-off score was the point at which the sum of both measures reached a
m a x i m u m value. In the present study, the criterion standard for correct classification as a
DAT patient or NE subject was the clinical diagnosis made by the two independent
senior staff neurologists.

RESULTS

Because the DAT, NE, and AR subject groups differed in age, the possible effect of age
on C D T performance was of some concern. Therefore, simple regression analyses were per-
formed to examine the relationship between age and CDT performance in cognitively intact
NE subjects.
Because these analyses revealed no significant relationship between age and quantitative
C D T score (r = -0.08, p < 0.05), qualitative error score (r = 0.07, p < 0.05) or global C D T
score (r = -0.08, p < 0.05), all subsequent analyses were performed using the raw scores.
The means and SDs of the three C D T scores achieved by each group are presented in
Table 2. Separate ANOVA comparing the DAT, NE, and AR groups on each measure
revealed significant main effects of Group for the Quantitative CDT score, F(2, 350) = 59.2,
p < 0.0001; the Qualitative C D T score, F(2, 350) = 47.6, p < 0.0001; and the Global C D T
score, /7(2, 350) = 64.6, p < 0.0001. Posbhoc Tukey tests (ct = 0.05) indicated that DAT
patients scored significantly worse than NE and AR subjects, and AR subjects scored signif-
icantly worse than NE subjects on all three C D T measures (all p ' s < 0.05).
T h e results o f the R O C analyses indicated that the optimal c u t - o f f score for the
Quantitative C D T measure (_<7) produced 88% sensitivity and 63% specificity for the diag-
nosis of DAT. The optimal cut-off score for the Qualitative measure (_>1 error) was similarly
effective, producing 81% sensitivity and 68% specificity. The Global C D T score, which
takes into account both quantitative and qualitative aspects of performance, was the most
effective of the three C D T measures for the diagnosis of DAT with an optimal cut-off score
of <6 producing 83% sensitivity and 72% specificity. Thus, the combination of qualitative
and quantitative measures on the CDT resulted in a statistically significant increase in speci-
ficity relative to a simple quantitative rating from 63% to 72% (g 2 = 4.26; p < 0.04), with
no appreciable decline in sensitivity (88-83%).

TABLE 2
Mean Score of the Normal Elderly (NE), At Risk, and Dementia of the Alzheimer
1~pe (DAD Subjects on the Three Clock Drawing Test Measures

NE At Risk DAT
Quantitative CDT Score 8.7 (I.2) 7.4 (I.7) 6.4 (2.1)
(max = 10)
Qualitative CDT Score 0.43 (0.72) I. I ( I. 1) 1.8 (1.5)
(max = 8)
Global CDT Score 8.3 (I.8) 6.3 (2.5) 4.5 (3.2)
(max = 10)
Note. Standard deviations are presented in parenthesis.
 Screeningfor Dementia With the Clock Drawing Test 535

The clock drawing performance of the AR subjects was examined to determine their
diagnostic classification based upon the three CDT cut-off scores derived for the DAT and
NE subjects. The Quantitative CDT cut-off score of _<7 classified 50% of the AR subjects as
normal and 50% as DAT. The Global CDT cut-off score of _<6 classified 51% of AR sub-
jects as normal and 49% as DAT. When qualitative features alone were evaluated, a cut-off
score of _>l error classified 35% of AR subjects as normal and 65% as DAT. Thus, AR sub-
jects were as likely to be classified as DAT as they were normal on the basis of their
Quantitative and Global CDT scores. However, the Qualitative CDT score classified more
AR subjects as DAT than as normal. In addition, the pattern of error types produced by the
AR subjects was very similar to that of the DAT patients.
The percentage of NE, DAT, and AR subjects that made each type of error on the CDT
are presented in Figure 1. Chi-square analyses comparing the three groups revealed that
there was a significant effect of diagnosis (p < 0.001) for the following error types:
Conceptual, Perseverative, Stimulus-bound, Planning, and Nonspecific Spatial. The three
groups did not differ significantly in the number of errors reflecting neglect of left hemis-
pace nor in writing the numbers outside of the clockface. Too few subjects made counter
clockwise errors to perform an analysis. Post-hoc Chi-square analyses comparing the DAT
and NE subjects demonstrated that more DAT patients than NE subjects committed Planning
(p < 0.01), Nonspecific spatial (p < 0.0001), Stimulus bound (p < 0.0001), Conceptual
(p < 0.0001 ), and Perseverative (p < 0.0001 ) errors. The AR and DAT groups differed only
on nonspecific spatial (p < 0.05) errors, with a greater percentage of DAT than AR subjects
making this type of error. Comparison of the AR and NE subjects showed that more AR sub-
jects than NE subjects committed Conceptual (p < 0.01), Stimulus bound (p < 0.0001),
Perseverative (p < 0.01 ), and Planning (p < 0.0001 ) errors. Examples of the types of errors
made on the CDT by four different AR subjects are presented in Figure 2. An example of a
Conceptual deficit, in which the subject was unable to convey the time properly on the clock
is shown in the upper left quadrant. A Stimulus bound error, in which the subject drew the
hands pointing at "10" and "l 1" is shown in the upper right quadrant. Perseverative behavior
is shown in the lower left quadrant; the lower right quadrant illustrates planning difficulty.

DISCUSSION

The results of the present study indicate that a clock drawing score that takes into account
both qualitative and quantitative features can provide reasonably good sensitivity and speci-
ficity for discriminating DAT from normal elderly individuals in a community-based popula-
tion. The Global CDT score, based on the scoring procedure described by Rouleau et al.
(1992), provided sensitivity of 83% and specificity of 72% for differentiating DAT patients
from normal elderly subjects. These findings suggest that when used as a single screening
instrument, the CDT will be only marginally effective in detecting DAT in the community.
The sensitivity and specificity for the diagnosis of DAT obtained with the optimal cut-off
scores in the present community-based sample are lower than previously reported with clin-
ically-based or volunteer samples (Wolf-Klein et al., 1989; Tuokko et al., 1992; Freedman
et al., 1994). As with other neuropsychological tests (Cahn et al., 1995), the lower discrimi-
nation between DAT and NE observed here may have occurred because NE subjects in the
community demonstrate greater variability in CDT performance than healthy, self-selected
volunteer subjects. It is likely that this increased variability in CDT performance is due to
the presence of NE individuals in the community who have medical disorders that may
affect test performance or who have subtle cognitive decline that falls between dementia
and normal cognitive functioning.
536 D. A. Cahn et al.

O NE (n=237)
[] AR ( n = 7 4 )

50- DAT (n--42)

40-

30-
4#

O 20-
00

10-

Error type
a: DAT>NE (p<.01)
b: AR>NE (p<.01)
c DAT>AR (p<.01)

FIGURE 1. Percentage of dementia of the Alzheimer type (DAT), normal dderly (NE), and at risk for DAT
(AR) subjects comltt/ng each e r ~ r type on the Clock Drawing Test

The CDT measure that provided the best discrimination between DAT and NE individuals
incorporated both a quantitative rating and an analysis of the qualitative features of the CDT.
This finding suggests that evaluation of the types of errors produced on the CDT may aid
in the early detection of dementia. Such an error analysis has been shown previously to be
extremely useful for differentiating DAT from normal aging and from other dementing dis-
orders (Fuld, 1983; Jacobs, Salmon, Trtster, & Butters, 1990; Rouleau et al., 1992;
Freedman et al., 1994). The predominant types of errors committed by the DAT patients in
the present study were Conceptual, Stimulus bound, and nonspecific spatial arrangement of
numbers. Rouleau and colleagues (1992) have suggested that Conceptual errors, in which
the subject demonstrates a misrepresentation of the clock or difficulty in setting the hands
may occur because of loss of semantic knowledge concerning the meaning and attributes of
the clock (Butters, Granholm, Salmon, Grant, & Wolf, 1987; Roulean et al., 1992; Chan et
 Screening for Dementia With the Clock Drawing Test 537

I J

ir

.....

,/"

FIGURE 2. Upper left quadrant: Conceptual deficit in the clock drawing of an at risk (AR) subject, Upper
right quadrant: Stimulus-bound error in the clock drawing of an AR subject, Lower left quadrant:
Perseveratlve behavior in the dock drawing of an AR subject, Lower right quadrant: Planning dil~culty in
the clock drawing of an AR subject Screening for DemenUia.

al., 1993). Stimulus bound errors in which the patient is pulled to the perceptual features of
the stimulus (i.e., setting the hands at 10 and 11), may represent a tendency to process infor-
mation on a more perceptual than semantic level (Kaplan, 1988). This type of error has been
attributed to frontal system dysfunction which disrupts abstract thinking.
In the current study, the optimal cut-off score of the quantitative CDT measure classified
50% of AR subjects as DAT and 50% as normal. Furthermore, when quantitative ratings
and qualitative features were considered together, the classification of AR subjects remained
relatively unchanged. The inconsistency with which the CDT quantitative rating classified
AR subjects is similar to the observation of Storandt and Hill (1989) that the performances
of questionably demented subjects overlaps with those of normal control and mildly-
demented subjects on objectively, or quantitatively, .scored neuropsychological tests.
When the types of errors produced by the AR subjects on the CDT were used for classifi-
cation, the majority of these subjects were classified as DAT. Close examination of the error
types revealed that the AR subjects produced a pattern of errors similar to that of the DAT
patients. In particular, the DAT and AR subjects were equally prone to make Planning,
Stimulus-bound, Conceptual, and Perseverative errors. Furthermore, both the DAT and AR
groups committed more of these types of errors than the NE subjects. It is possible that
changes in such subtle qualitative features of clock drawing may be an early indication of a
538 D. A. Cahn et al.

progressive dementing disorder. It should be emphasized, however, that only longitudinal

follow-up of the AR subjects will determine whether these error types are predictive of
future cognitive deterioration.
Although the CDT by itself was effective in differentiating between DAT and NE indi-
viduals in the present study, more accurate classification has been reported with other single
or composite neuropsychological measures. For example, measures of memory and, in par-
ticular, rapid forgetting, may be more appropriate screening instruments as memory impair-
ment is the most prominent and early manifestation of the disorder (Eslinger, Damasio,
Benton, & Allen, 1985; Moss & Albert, 1988; Welsh, Butters, Hughes, Mohs, & Heyman,
1991; Welsh, Butters, Hughes, Mohs, & Heyman, 1992). Also, the Mattis Dementia Rating
Scale (Mattis, 1976) has been recently demonstrated to have excellent classification accura-
cy, particularly because it assesses memory and language functions (Monsch et al., 1995).
Finally, it is important to note the limitations to the generalizability of the current find-
ings. The Rancho Bemardo community is a predominantly white, highly-educated, middle
to upper-middle class population. The cut-off scores and rates of sensitivity and specificity
that were derived with the three CDT measures might not be applicable to less well-educat-
ed or non-White communities. While some investigators suggested that performance on
clock drawing tasks is independent of socioeconomic factors (Shulman, Shedletsky, &
Silver, 1986; Watson, Arfken, & Birge, 1993) recent evidence suggests that educational
background may affect clock drawing ability (Ainslie & Murden, 1993; Cahn, Wiederholt,
Salmon, & Butters, 1994). Additional studies should be conducted to evaluate the diagnos-
tic accuracy of the CDT in community-dwelling populations with different socio-demo-
graphic characteristics.

REFERENCES

Ainslie, N. K., & Murden, R. A. (1993). Effect of education on the clock-drawing dementia screen in nondemented
elderly persons. Journal of the American Geriatric Society, 41,249-252.
Beardsall, L., & Huppert, E A. (1991). A comparison of clinical, psychometric and behavioural memory tests:
Findings from a community study of the early detection of dementia. International Journal of Geriatric
Psychiatry, 6, 295-306.
Bondi, M. W., Monsch, A. U., Galasko, D., Butters, N., Salmon, D. P., & Delis, D. C. (1994). Preclinical cognitive
markers of dementia of the Alzheimer type. Neuropsychology, 8, 374-384.
Butters, N., Granholm, E., Salmon, D. P., Grant, I., & Wolfe, J. (1987). Episodic and semantic memory: A compar-
ison of amnesic and demented patients. Journal of Clinical and Experimental Neuropsychology, 9, 479-497.
Cahn, D. A., Salmon, D. P., Butters, N., Wiederholt, W. C., Corey-BIoom, J., Edelstein, S. L., & Batrett-Connor, E.
(1995). Detection of dementia of the Alzheimer type in a community dwelling sample: Neuropsychological test
performance. Journal of the International Neuropsychological Society, 1,252-260.
Cahn, D. A., Wiederholt, W. C., Salmon, D. P., & Butters, N. (1994). Detection of cognitive impairment in
Spanish-speaking elderly with the clock drawing test. Archives of Clinical Neuropsychology, 9, 113-114
(Abstract).
Chan, A. S., Butters, N., Paulsen, J. S., Salmon, D. P., Swenson, M. R., & Maloney, L. T. (1993). An assessment of
the semantic network in patients with Alzheimer's disease. Journal of Cognitive Neuroscience, 5, 254-261.
Criqui, M. H., Barrett-Connor, E., & Austin, M. (1978). Differences between respondents and nonrespondents in a
population-based cardiovascular disease study. American Journal ofEpidemiology, 1 ~ , 367-372.
Eslinger, P. J., Damasio, A. R., Benton, A. L., & Allen, M. V. (1985). Neuropsychological detection of abnormal
mental decline in older persons. Journal of the American Medical Association, 253, 670-674.
Freedman, M., Leach, L., Kaplan, E., Winocur, G., Shulman, K. I., & Delis, D. C. (1994). Clock drawing: A neu-
ropsychological analysis. New York: Oxford University Press.
Fuld, P. A. (1983). Word intrusions as a diagnostic sign in Alzheimer's disease. Geriatric Medicine Today, 2,
33-41.
C_nx~glass, H., & Kaplan, E. (1972). The assessment of aphasias and related disorders. Philadelphia: Lea and Febiger.
Jacobs, D., Salmon, D. P., Trtster, A. I., & Butters, N. (1990). Intrusion errors in the figural memory of patients
with Alzheimer's and Huntington's disease. Archives of Clinical Neuropsychology, 5, 49-57.
 Screening f o r Dementia With the Clock Drawing Test 539

Kaplan, E. (1988). A process approach to neuropsychological assessment. In T. Boll & B. K. Bryant (eds.), Clinical neu-
ropsychalogy and brain function: Research, measurement, and practice. Washington, DC: American Psychological
Association.
Kozora, E., & Cullum, C. M. (1994). Qualitative features of clock drawings in normal aging and Alzheimer's dis-
ease. Assessment, 1, 179-187.
Landis, S. E., & Koch, C. G. (1977). The measurement of observer agreement for categorical data. Biometrics, 33,
159-174.
Mattis, S. (1976). Mental status examination for organic mental syndrome in the elderly patient. In R. Bellak, &
B. Katusu (eds.), Geriatric psychiatry: A handbook for psychiatrists and primary care physicians (pp. 77-121).
New York: Grune & Stratton.
McKhann, G., Drachman, D., Folstein, M., Katzman, R., Price, D., & Stradlan, E. M (1984). Clinical diagnosis of
Alzheimer's disease: Report of the NINCDS-ADRDA work group under the auspices of department of health
and human services task force on Alzheimer's disease. Neurology, 34, 939-944.
Mendez, M. E, Ala, T., & Underwood, K. L. (1992). Development of scoring criteria for the clock drawing task in
Alzheimer's disease. Journal of the American Geriatric Society, 40, 1095-1099.
Monsch, A. U., Bondi, M. W., Salmon, D. P., Butters, N., Thai, L. J., Hansen, L. A., Wiederholt, W. C., Cahn,
D. A., & Klauber, M. R. 0995). Clinical validity of the Mattis Dementia Rating Scale in detecting dementia of
the AIzheimer type: A double cross-validation and application to a community-dwelling sample. Archives of
Neurology, 52, 899-904.
Moss, M. B., & Albert, M. S. (1988). Alzheimer's disease and other dementing disorders. In M. S. Albert, & M. B.
Moss (eds.), Geriatric neuropsychology. New York: Guilford Press.
Nussbaum, P. D., Fields, R. B., & Starratt, C. (1992). A comparison of thrce scoring procedures for the clock draw-
ing test. Journal of Clinical and E.g~erimental Neuropsychology, 14, 44 (Abstract).
Pfeffer, R. 1., Kurosaki, T. T., & Harrah, C. H. (1981). A survey diagnostic tool for senile dementia. American
Journal of Epidemiology, 114, 515-527.
Rouleau, I., Salmon, D. P., Butters, N., Kennedy, C., & McGuire, K. (1992). Quantitative and qualitative analyses
of clock drawings in Alzheimer's and Huntington's disease. Brain and Cognition, 18, 70-87.
Shulman, K. I., Shedletsky, R., & Silver, 1. L. (1986). The challenge of time: Clock-drawing and cognitive function
in the elderly. International Journal of Geriatric Psychiatry, 1, 135-140.
Storandt, M., & Hill, R. D. (1989). Very mild senile dementia of the Alzheimer type: Psychometric test perfor-
mance. Archives of Neurology, 46, 383-386.
Sunderland, T., Hill, J. L., Mellow, A. M., Lawlor, B. A., Gundersheimer, J., Newhouse, P. A., & Grafman, J. H.
(1989). Clock drawing in Alzheimer's disease: A novel measure of dementia severity. Journal of the American
Geriatric Society, 37, 725-729.
Tuokko, H., Hadjistavropoulos, T., Miller, J. A., & Beattie, B. L. (1992). The clock test: A sensitive measure to dif-
ferentiate normal elderly from those with Alzheimer disease. Journal of the American Geriatric Society, 40,
579-584.
Watson, Y. 1., Arfken, C. L., & Birge, S. J. (1993). Clock completion: An objective screening test for dementia.
Journal of the American Geriatric Society, 41, 1235-1240.
Welsh, K. A., Butters, N., Hughes, J., Mobs, R., & Heyman, A. (1991). Detection of abnormal memory decline in
mild cases of Alzheimer's disease using CERAD neuropsychological measures. Archives of Neurology, 48,
278-281.
Welsh, K. A., Butters, N., Hughes, J., Mohs, R., & Heyman, A. (1992). Detection and staging of dementia in
Alzheimer's disease: Use of the neuropsychological measures developed for the Consortium to Establish a
Registry for Alzheimer's Disease (CERAD). Archives of Neurology, 49, 448-452.
Wiederholt, W. C., Cahn, D. A., Salmon, D. P., Butters, N., Kritz-Silverstein, D., & Barrett-Connor, E. (1993). The
effects of age, education, and gender in an elderly community cohort. Journal of the American Geriatric
Society, 41,639-647.
Wolf-Klein, G. P., Silverstone, E A., Levy, A. P., & Brod, M. S. (1989). Screening for Alzheimer's Disease by
Clock Drawing. Journal of the American Geriatric Society, 37, 730-734.