Clinical Hub,UW Health Clinical Tool Search,UW Health Clinical Tool Search,Questionnaires,Related

Six-Item Screener to Identify Cognitive Impairment Among Potential Subjects for Clinical Research

Six-Item Screener to Identify Cognitive Impairment Among Potential Subjects for Clinical Research - Clinical Hub, UW Health Clinical Tool Search, UW Health Clinical Tool Search, Questionnaires, Related

Six-Item Screener to Identify Cognitive Impairment Among
Potential Subjects for Clinical Research




OBJECTIVE. To design a brief cognitive
screener with acceptable sensitivity and speci-
ficity for identifying subjects with cognitive
DESIGN. Cohort one is assembled from a
community-based survey coupled with a
second-stage diagnostic evaluation using for-
mal diagnostic criteria for dementia. Cohort
two is assembled from referrals to a specialty
clinic for dementing disorders that completed
the same diagnostic evaluation.
SETTING. Urban neighborhoods in Indianap-
olis, Indiana and the Indiana Alzheimer Dis-
ease Center.
PATIENTS. Cohort one consists of 344
community-dwelling black persons identified
from a random sample of 2212 black persons
aged 65 and older residing in Indianapolis;
cohort two consists of 651 subject referrals to
the Alzheimer Disease Center.
MEASUREMENTS. Formal diagnostic clinical as-
sessments for dementia including scores on the
Mini-mental state examination (MMSE), a six-item
screener derived from the MMSE, the Blessed
Dementia Rating Scale (BDRS), and the Word List
Recall. Based on clinical evaluations, subjects were
categorized as no cognitive impairment, cognitive
impairment-not demented, or demented.
RESULTS. The mean age of the community-
based sample was 74.4 years, 59.4% of the
sample were women, and the mean years of
education was 10.1. The prevalence of demen-
tia in this sample was 4.3% and the prevalence
of cognitive impairment was 24.6%. Using a
cut-off of three or more errors, the sensitivity
and specificity of the six-item screener for a
diagnosis of dementia was 88.7 and 88.0, re-
spectively. In the same sample, the corre-
sponding sensitivity and specificity for the
MMSE using a cut-off score of 23 was 95.2 and
86.7. The performance of the two scales was
comparable across the two populations studied
and using either cognitive impairment or de-
mentia as the gold standard. An increasing
number of errors on the six-item screener is
highly correlated with poorer scores on longer
measures of cognitive impairment.
CONCLUSIONS. The six-item screener is a brief
and reliable instrument for identifying sub-
jects with cognitive impairment and its diag-
nostic properties are comparable to the full
MMSE. It can be administered by telephone or
face-to-face interview and is easily scored by a
simple summation of errors. (Med Care 2002;
Scientists interested in enrolling older adults in
clinical research studies often seek to identify
subjects with cognitive impairment as an initial
assessment in the consideration of more specific
From the *Indiana University Center for Aging Re-
search, the

Regenstrief Institute for Health Care, and

Department of Psychiatry, Indiana University
School of Medicine, Indianapolis, Indiana.
Supported by NIA grants R01 AG 00956, P30 AG
10133, and K07 AG 00868, and a grant from the John A.
Hartford Foundation.
Address correspondence and reprint requests to:
Christopher M. Callahan, MD Indiana University Center
for Aging Research, 1050 Wishard Blvd, RG6 Indianap-
olis, IN 46202. E-mail ccallahan@regenstrief.org
Received August 1, 2001; initial review October 15,
2001; accepted February 18, 2002.
DOI: 10.1097/01.MLR.0000024610.33213.C8
Volume 40, Number 9, pp 771–781
2002 Lippincott Williams & Wilkins, Inc.

inclusion or exclusion criteria. Excluding subjects
with cognitive impairment may be desirable when
the study relies on self-reports of functioning,
mood, health-related quality of life, or health
services utilization as outcome measures. Other
investigators may adjudge that adherence to spe-
cific self-care behaviors, study protocols, or other
complex tasks require intact cognitive function.
Scientists studying dementing disorders often
seek to efficiently screen a large numbers of sub-
jects in a first-stage assessment to identify those
patients most likely to meet criteria for dementia
in a second-stage assessment. Despite the fre-
quent goal to efficiently identify older adults with
cognitive impairment or identify those with a high
probability of dementing disorders, there is no
consensus on how to best balance the need for
accuracy with limited resources and time.
Clearly, these issues are not limited to research.
Clinicians faced with the resource constraints of
daily clinical practice also seek screening tests,
which can balance accuracy with efficiency. There
are already numerous measures of cognitive im-
pairment developed for use in clinical settings.
These instruments typically range from 10 to 30
items. Most of these questionnaires have demon-
strated sensitivity and specificity as an aid to the
diagnosis of dementia. Unfortunately, these in-
struments can take from 7 to 15 minutes to
complete and some require props, paper, and
pencil, or other face-to-face interactions. In addi-
tion, these longer scales do not always perform
with greater accuracy in comparison to shorter
One solution to the time burdens of these
longer questionnaires has been two-stage screen-
ing. For example, Lachs et al
have suggested
using three-item recall as an initial screen for
cognitive impairment followed by the Mini-
Mental State Examination (MMSE) for those pa-
tients unable to recall all three items. This first-
stage screen is reported to have excellent
sensitivity (97%), but poor specificity (43%) which
makes it useful as an initial screen to identify those
subjects unlikely to have the condition.
However, in some clinical trials, investigators
may be more interested in optimizing specificity.
For example, in the design of an ongoing multisite
study of late life depression, investigators were
faced with the challenge of balancing the need to
exclude older adults who would be unable to
provide self-reports or adhere to the protocol with
the competing goal to include older adults who
might have poor cognitive performance because of
a treatable depression.
Indeed, it is often difficult
to determine what magnitude of cognitive impair-
ment renders a potential subject ineligible for
meaningful participation. Many patients with mild
cognitive impairment may be capable of providing
self-reports and following study protocols. An
overzealous exclusion of subjects with mild cogni-
tive impairment might unnecessarily reduce the
generalizability of a study. Thus, different studies
would be expected to make different choices in
balancing the competing needs for sensitivity and
We sought to develop a brief screen for cogni-
tive impairment that would balance diagnostic
accuracy with the logistic demands of screening a
large group of subjects in an efficient manner. This
report provides a detailed description of the sen-
sitivity, specificity, and predictive value of a six-
item screener for cognitive impairment among
older adults. There are several advantages of this
six-item screener over existing scales in addition to
its brevity. First, each of the six items comes from
the MMSE, which allows for comparison among
the many studies utilizing this longer question-
naire. Second, the six-item screener can be admin-
istered over the telephone and it is scored simply
by summing the number of errors. Third, the
diagnostic performance of the scale can be varied
by choosing a cut-off score to match the study
goals. The six-item screener is offered as an effi-
cient tool to identify patients with cognitive im-
pairment either as a one-stage screen with accept-
able specificity to exclude those with moderate to
severe impairment, or as the first stage of a
two-stage screen to identify probable cases of
Materials and Methods
Study Samples
The impetus for this study was the need for a
brief cognitive screener to efficiently exclude pa-
tients with moderate to severe cognitive impair-
ment in a multisite study of late life depression.
The data for this study come from two projects
funded by the National Institute on Aging that are
investigating the prevalence, incidence, risk fac-
tors, and treatment of dementia. The first source of
subjects is a study on the prevalence of dementia
among a community-based sample of black per-
sons. The second source is from the subjects

assembled from referrals to the Indiana Alzheimer
Disease Center. Both groups of subjects complete
the same clinical evaluation process by the same
group of clinicians associated with the Indiana
Alzheimer Disease Center. However, in the first
sample, subjects are identified by a community-
based screening program and in the second sam-
ple, subjects are referred to the Center. The two
samples are described below followed by a de-
scription of the common clinical evaluation.
For the community-based sample, the geo-
graphic target area consisted of 29 contiguous
census tracts with a total population of 82,387 and
total households of 32,954 in the 1990 US Census.
Black persons comprised 86% of this population,
which also represents more than two-thirds of
Indianapolis’ elderly black population. A random
sample of 60% of residential addresses was con-
structed by the Indianapolis Water Company using
all residential addresses in the target area, and
identified homes were then visited by interviewers
from May 1, 1992-April 30, 1993. Patients residing
in nursing homes are not included in this sample.
Eligible subjects had to be (1) a resident at a
sampled address, (2) black, and (3) age 65 years or
older. A total of 7590 households were ap-
proached, 4915 of which did not have an eligible
resident. Of the 2582 eligible persons, 2212
(85.7%) agreed to participate. These subjects were
screened with the Community Screening Instru-
ment for Dementia (CSI-D).
Details of the development, content, scoring,
and psychometric properties of the CSI-D have
been previously published.
Briefly, the CSI-D is
composed of two parts: a 33-item scale assessing
the subject’s cognitive performance and a 24-item
scale assessing a relative’s perception of a decline
in the subject’s functional or social abilities. Items
for the CSI-D were selected from several widely
used screening instruments including the Cam-
bridge Mental Disorders in the Elderly Examina-
the Mini-Mental State Examination,
Dementia Rating Scale,
the Comprehensive As-
sessment and Referral Evaluation,
and the East
Boston Memory Test.
The items selected test
cognitive function across multiple domains but
specifically exclude literacy dependent items. A
discriminant function was derived in developmen-
tal work on the CSI-D to establish an empirically
derived cut-off score that best differentiated be-
tween demented and nondemented with a struc-
tured clinical assessment as the gold standard.
Subjects were classified into “poor,”“intermedi-
ate,”or “good”performance groups based on their
discriminant function score. In a community prev-
alence study, the sensitivity of the CSI-D was 87%
and the specificity was 83%.
A stratified sample of the community-based
subjects was selected for full clinical assessments
based on their performance on the CSI-D. All
subjects who scored poorly on the CSI-D were
invited for clinical assessments and we also se-
lected a 50% sample of those with intermediate
performance, and a 5% sample of those with good
performance. Patients aged 75 and older were
over-sampled in the 5% sample so that 75% of the
patients with good performance on the CSI-D
would be 75 years of age or older. Rates of
cognitive impairment, dementia, and Alzheimer’s
disease among this community-based sample
have been previously published.
The impact of
age, gender, education, and occupation on cogni-
tive performance in this sample has also been
previously published.
There were 351 patients
selected for full clinical assessments but seven
were too severely impaired to complete the stan-
dardized questionnaires. Data for the remaining
344 (98%) subjects are included here.
The second set of subjects comes from patient
referrals to the Alzheimer Disease Center at the
Indiana University School of Medicine. The differ-
ences in sampling strategies for these two samples
are considerable and are reflected in the demo-
graphic and clinical characteristics provided in
Table 1. Patients are referred to this Center both for
diagnosis and for treatment and it is the only
Center of its kind in Indiana. Notably, patients
from this sample are not initially screened but
referred by family, caregivers, or providers for
evaluation. Thus, the CSI-D is not performed as
the first stage assessment of the clinical sample.
The clinical sample is not limited to black persons
who were the focus of the community-based study
described above. There were 662 subjects referred
for the clinical assessment, but eleven were too
severely impaired to complete the standardized
questionnaires. Data for the remaining 651 (98%)
subjects are included here.
Clinical Assessments
All clinical assessments of subjects from the
community-based cohort were made blinded to
the screening status. A geriatric psychiatrist or
neurologist conducted a complete physical and

neurologic examination. Cognitive assessments
included the MMSE, the cognitive performance
portion of the CAMDEX, and the Consortium for
Establishment of Registry for Alzheimer Disease
(CERAD) battery.
In addition to the MMSE, the
CERAD battery includes the Animal Fluency Test
(a measure of semantic fluency in which subjects
generate as many names of animals as possible in
60 seconds), the Boston Naming Test (a 15-item
test of confrontation naming of line drawings of
objects), Constructional Praxis (a test of grapho-
motor skill in which subjects copy geometric fig-
ures), and the Word List Recall (a 10-item word list
is presented three times with free recall and rec-
ognition assessed after a brief, filled interval).
Where possible, a relative of the subject was also
interviewed. A research nurse met with a spouse
or other relative and completed the semi-
structured Informant Interview. The interview pro-
vides information on the presence, duration, and
severity of symptoms of memory, language, judg-
ment and reasoning, and personality change. In-
formants are also asked to characterize the sub-
ject’s performance of instrumental and basic
activities of daily living (ADLs). The CERAD-
modified version of the Blessed Dementia Scale
was calculated from the Informant Interview for
those subjects where an informant could be inter-
viewed. The Blessed consists of 11 items assessing
memory, comprehension, shopping/money man-
agement, performance of household chores, dress-
ing, feeding, and toileting.
On the basis of the above evaluation, partici-
pants were classified as normal, cognitive
impairment-not demented, or demented. Patients
were diagnosed as cognitive impairment-not de-
mented if: (1) the informant reported a clinically
significant decline in cognition; (2) the physician
detected a clinically significant impairment in cog-
nition; or (3) the participant’s scores on cognitive
testing fell below the 7th percentile; and if there
was no clinically important impairment in the
performance of activities of daily living.
The 7th
percentile is approximately equivalent to 1.5 stan-
dard deviations (SD) below the mean, the level of
impairment specified by Mayo Clinic in their cri-
teria for mild cognitive impairment. For a diagno-
sis of dementia both DSM-III-R and ICD-10 cri-
teria had to be satisfied.
On the basis on this
clinical assessment, patients were dichotomized
into demented and nondemented groups. Patients
with dementia were then further categorized into
those with and without possible or probable Alz-
heimer disease as defined by NINCDS/ADRDA
For the purposes of the current study, we
focus on the diagnosis of normal, cognitive
impairment-not demented, or dementia. In all
tables, the cognitive impairment group includes
both patients with the “cognitive impairment-not
demented”diagnosis and the dementia diagnosis.
Design of Six-Item Screener
In designing the six-item screener, we sought to
balance the instrument’s diagnostic properties
with brevity, ease of administration, and validity.
Because investigators working on different
projects might seek to optimize sensitivity as
opposed to specificity or vice versa, we also sought
TABLE 1. Characteristics of Patient Samples
Alzheimer Disease
Center Sample
Sample size 344 651 �0.001
Mean age (range) 74.4 (65–99) 69.6 (21–92) �0.001
% women 59.4 57.1 �0.698
% black 100 16.1 �0.001
Mean years of education (range) 10.4 (0–16) 12.5 (0–20) �0.001
% cognitively impaired 26.4 61.3 �0.001
% with dementia 4.3 53.0 �0.001
Mean errors on six-item screener 1.3 2.6 �0.001
Mean score on MMSE 26.1 21.7 �0.001
Mean score on Word List Recall 13.8 12.6 0.012
Mean score on Blessed Dementia Rating Scale 4.3 7.4 �0.001

to design a screener that would allow a variety of
“cut-off points.” The hallmark of dementia is a
deficit in short-term memory. The MMSE is
heavily loaded with memory items though some
are more sensitive than others. For example, tem-
poral disorientation occurs before disorientation to
place. Within temporal orientation, problems with
day of the week, month, and year are rarely seen in
those not experiencing dementia (high specificity).
Three-object recall is the best assessment of new
learning ability in the MMSE and has consistently
been identified as having excellent discrimination
for identification of subjects with cognitive impair-
ment (high sensitivity). Three-object registration
has more to do with language, hearing, and atten-
tion. Although registration is a necessary step in
successful recall, it does not in itself discriminate
well between those with and without dementia.
The rest of the MMSE items tap language, atten-
tion, or praxis and while any of these may be
impaired in any given patient with dementia, no
one domain or item is reliably implicated, some of
these items are more sensitive to education, and
some require props or motor skills not assessable
by telephone. Thus, we chose the three-item recall
(apple, table, penny) and three-item temporal
orientation (day of the week, month, year) to
design the six-item screener. Notably, the three-
item recall question in the CSI-D is “boat, house,
and fish” consistent with prior work on this
We present the sensitivity, specificity, predictive
value, and area under the receiver operating char-
acteristic (ROC) curve for the six-item screener
using cognitive impairment as the gold standard
and then with dementia as the gold standard.
Analyses of the community-based sample analy-
ses are weighted, with individual weights being
inversely proportional to the sampling proportion
in that stratum. To compare the performance of
the six-item screener with the full MMSE, we
present the diagnostic properties of the MMSE in
this same population and report the mean scores
and ranges on the MMSE, Word List Recall, and
Blessed Dementia scale at each level of subject
performance on the six-item screener. As noted
above, approximately 2% of both sample popula-
tions could not be tested on the MMSE because of
the severity of their impairment. Among the sub-
jects adjudged to be testable, coding of responses
to the MMSE required that the respondent provide
the correct answer or the item was coded as
incorrect. However, 21% of the community-based
sample and 8% of the clinical sample either re-
fused or could not perform the Word List Recall.
Also, 53% of the community-based sample and
31% of the clinical sample did not have an infor-
mant and therefore do not have scores on the
Blessed Dementia Rating Scale.
Table 1 provides the clinical characteristics of
the two samples. As would be expected from the
differences in sampling strategy, the community-
based sample consists of black persons who are
older, less educated, and less likely to have cogni-
tive impairment or dementia as compared with the
Alzheimer Disease Center sample.
Tables 2 to 5 present the diagnostic properties of
the six-item screener as compared with the MMSE
TABLE 2. Sensitivity, Specificity, and Predictive Value of Six-Item Screener Among the
Community-Based Sample
Six-item Screener Cognitive Impairment as Gold Standard Dementia Diagnosis as Gold Standard
Errors N Sens Spec PPV NPV Sens Spec PPV NPV
�0 344 100.0 0.0 26.4 100.0 0.0 4.3
�1 273 97.7 49.2 40.8 98.3 100.0 38.4 6.7 100.0
�2 190 74.2 80.2 57.4 89.6 96.8 68.6 12.1 99.8
�3 120 50.4 97.4 87.2 84.5 88.7 88.0 24.8 99.4
�4 75 27.8 99.4 93.9 79.3 75.2 95.2 40.9 98.9
�5 45 14.8 100.0 100.0 76.6 56.1 98.4 61.1 98.1
6 18 4.7 100.0 100.0 74.5 24.2 99.8 83.3 96.7
MMSE � Mini-mental state examination; Sens � sensitivity; Spec � specificity; PPV � positive predictive value;
NPV � negative predictive value; ADC � Alzheimer’s Disease Center.

using cognitive impairment or dementia as the
gold standard in both the community-based and
clinic-based patient populations. It must be
stressed that these two instruments are being
compared in the same population(s) of patients
against a separate gold standard clinical diagnosis.
In addition to sensitivity and specificity, we
present the positive and negative predictive val-
ues. Predictive value is a property both of the
sensitivity and specificity of the test and the prev-
alence of the disease in the population under
study. A test with higher sensitivity optimizes
negative predictive value whereas a test with
higher specificity optimizes positive predictive
As demonstrated in Tables 2 to 5, the six-item
screener performs well in comparison with the
longer MMSE. In both populations and using
either gold standard, one can identify a cut-off
score on the six-item screener that would compare
favorably with the MMSE in terms of diagnostic
accuracy. Indeed, as a first stage screening tool
among a community-based population to identify
subjects with cognitive impairment the six-item
screener performs at least as well as the MMSE.
The six-item screener performs less well in com-
parison to the full MMSE when one compares the
instruments in a population with a high preva-
lence of disease and using dementia as the gold
standard. However, even in this population, one
can choose a cut-off score that optimizes sensitiv-
ity and specificity. Table 6 compares the area under
the ROC curves for the six-item screener as com-
pared with the MMSE.
Table 7 compares the mean scores of three other
commonly used instruments to screen for cogni-
tive impairment with scores on the six-item
screener. Mean MMSE, Word List recall, and
Blessed Dementia Scale scores progressively
worsen as the number of errors on the six-item
screener increase. This finding is consistent across
all three comparison scales and at each level of
TABLE 3. Sensitivity, Specificity, and Predictive Value of the MMSE Among the
Community-Based Sample
MMSE Cognitive Impairment as Gold Standard Dementia Diagnosis as Gold Standard
Score N Sens Spec PPV NPV Sens Spec PPV NPV
�27 269 91.5 56.2 42.9 94.9 100.0 45.6 7.6 100.0
�26 241 76.5 12.9 50.4 89.6 98.4 62.5 10.5 99.9
�25 206 71.5 87.3 66.9 89.5 98.4 74.9 14.9 99.9
�24 172 53.3 92.1 70.9 84.6 98.4 83.6 21.1 99.9
�23 149 44.4 93.2 70.1 82.4 95.2 86.7 24.2 99.8
�22 123 38.9 94.8 72.8 81.2 87.1 89.1 26.3 99.4
�21 108 36.1 95.8 75.5 80.7 87.1 90.7 29.4 99.4
MMSE � Mini-mental state examination; Sens � sensitivity; Spec � specificity; PPV � positive predictive value;
NPV � negative predictive value; ADC � Alzheimer’s Disease Center.
TABLE 4. Sensitivity, Specificity, and Predictive Value of Six-Item Screener Among the
ADC Clinical Sample
Six-item Screener Cognitive Impairment as Gold Standard Dementia Diagnosis as Gold Standard
Errors N Sens Spec PPV NPV Sens Spec PPV NPV
�0 651 100.0 0.0 61.3 100.0 0.0 53.0
�1 477 93.7 59.1 78.4 85.6 96.8 53.3 70.0 93.7
�2 372 84.0 85.3 90.1 77.1 89.6 79.4 83.1 87.1
�3 306 74.2 96.0 96.7 70.1 80.6 90.9 90.9 80.6
�4 245 60.9 99.2 99.2 61.6 67.5 96.1 95.1 72.4
�5 173 43.1 99.6 99.4 52.5 49.0 98.7 97.7 65.2
6 107 26.6 99.6 99.1 46.1 30.4 99.4 98.1 55.9
MMSE � Mini-mental state examination; Sens � sensitivity; Spec � specificity; PPV � positive predictive value;
NPV � negative predictive value; ADC � Alzheimer’s Disease Center.

performance on the six-item screener. Using this
table, an investigator can extrapolate mean scores
on the six-item screener to corresponding scores
on the longer scales if one seeks to compare levels
of cognitive impairment to studies using the
longer scales. As shown in Table 8, the number of
errors on the six-item screener is highly correlated
with performance on the other three scales.
We propose the six-item screener as an efficient
and accurate method to screen subjects for cogni-
tive impairment. The scale was specifically devel-
oped for studies that must screen large numbers of
subjects and for studies that rely on subjects’
cognitive ability to participate in a complex inter-
vention and/or provide self-reports. A specific
inclusion criterion in such a study is often the
requirement that a patient have the cognitive
capacity to understand questions about their cur-
rent symptoms, emotion, or function, and be able
to follow the study protocol. Although this scale
was originally conceived for use in research stud-
ies, the diagnostic characteristics are comparable
to the MMSE or the Blessed Dementia Rating
Scale and thus the six-item screener could also be
used in clinical practice as a first stage assessment
for cognitive impairment.
There are several important logistic features of
the six-item screener that make it particularly
well-suited for use in research studies compared
with other brief screens recently developed.
First, the scale is short and unobtrusive so that it
can be readily incorporated into an initial patient
assessment of eligibility. The scale takes only 1 to
2 minutes to complete as compared with 7 to 15
minutes for longer scales.
Second, the scale
does not include any visuospatial or motor skill
tasks, it does not require any props or visual cues,
and scoring requires only the simple addition of
the number of errors.
Thus, the six-item
screener can be easily administered by telephone
or in face-to-face interviews. Third, the investiga-
tor can alter the cut-off score to match the goals of
the study and the targeted population.
We have demonstrated the diagnostic charac-
teristics of the six-item screener in a community-
based sample where the screening scale used
(CSI-D) was independent from the six-item
TABLE 5. Sensitivity, Specificity, and Predictive Value of MMSE Among the ADC Clinical Sample
MMSE Cognitive Impairment as Gold Standard Dementia Diagnosis as Gold Standard
Score N Sens Spec PPV NPV Sens Spec PPV NPV
�27 445 93.0 70.6 83.4 86.4 98.0 65.0 76.0 96.6
�26 393 88.2 83.7 89.6 81.8 94.5 78.1 83.0 92.6
�25 357 82.7 89.3 92.4 76.5 89.3 84.0 86.3 87.4
�24 322 77.2 94.4 95.7 72.3 84.6 90.2 90.7 83.9
�23 301 73.4 96.8 97.3 69.7 81.5 93.5 93.4 81.7
�22 279 68.9 98.4 98.6 66.7 76.8 95.4 95. 78.5
�21 261 64.9 99.2 99.2 64.1 73.0 97.1 96.6 76.2
MMSE � Mini-mental state examination; Sens � sensitivity; Spec � specificity; PPV � positive predictive value;
NPV � negative predictive value; ADC � Alzheimer’s Disease Center.
TABLE 6. Area Under ROC Curves for MMSE Compared with Six-Item Screener
Gold Standard Six-item screener MMSE
Community-based sample
Cognitive Impairment 0.86 0.84
Dementia 0.95 0.96
Clinical sample
Cognitive Impairment 0.91 0.93
Dementia 0.92 0.95

screener described here. Notably, the six-item
screener’s performance is based on a gold stan-
dard diagnosis of cognitive impairment or demen-
tia rather than its ability to predict a total score on
the full MMSE. This is important because the
MMSE typically performs in the range of 80% to
85% sensitivity and specificity;
in other words,
the MMSE does not provide a gold standard for
cognitive impairment or dementia. The six-item
screener’s performance was excellent in both of
the populations studied in this report. The scale
performed nearly as well as the MMSE in these
patient populations and showed a high level of
validity when compared with other commonly
used screens for cognitive impairment.
Although the six-item screener performed well
in these two populations in terms of diagnostic
accuracy for identifying older adults with cognitive
impairment or dementia, it is important to note
the differences in the two patient populations as
described in Table 1. The community-based sam-
ple is representative of urban, black older adults,
but these results may not generalize to other racial
groups. Subjects in the clinical sample completed
the same evaluation as the community-based
sample and this sample comprises both white
persons and black persons. Taken together, the
two samples thereby represent a fairly broad spec-
trum of older adults but simply combining the
results of these two samples does not create a
cohort necessarily generalizable to all older adults.
Although our use of a community-based sample
of black persons improves upon prior studies
relying only on clinical samples, exploring the
generalizability of our findings is an important
area for future research.
Because both clinicians and researchers seek a
brief and accurate method to identify patients or
subjects with cognitive impairment, multiple pre-
vious investigators have reported on the sensitivity
TABLE 7. Means, Medians, and Ranges of other Screening Instruments by Number of Errors on
Six-item Screener Among Community-Based Sample and Alzheimer Disease Center Sample
No. of
Errors Sample
MMSE Word List Recall Blessed Dementia Scale
Mean Median Range Mean Median Range Mean Median Range
0 Comm 28.4 29.0 17–30 16.0 16.0 7–24 3.6 3.5 3.0 –7.5
Clinical 28.9 29.0 23–30 20.0 20.0 9–30 3.7 3.0 2.8–12.0
1 Comm 27.0 27.0 17–29 14.6 15.0 5–23 4.4 4.0 3.0–11.6
Clinical 26.9 27.0 20–29 15.9 15.0 5–28 4.9 4.0 3.0–11.5
2 Comm 25.8 26.0 16–28 12.1 12.0 4–20 3.8 3.5 3.0–14.3
Clinical 24.8 25.0 15–28 13.1 13.0 6–24 6.4 6.5 3.0–17.6
3 Comm 22.4 25.0 10–27 10.6 12.0 0–16 3.8 3.5 3.0–10.4
Clinical 20.6 21.0 9–27 9.3 9.0 0–22 7.3 7.0 3–14.5
4 Comm 19.4 19.0 12–24 8.6 10.0 0–14 5.1 5.5 3.0–13.8
Clinical 18.9 20.0 5–26 8.8 9.0 0–17 8.1 8.0 3.5–16.5
5 Comm 14.4 16.0 3–23 7.1 6.0 0–13 6.8 6.0 4.4–10.5
Clinical 14.7 15.5 4–24 5.7 5.5 0–15 9.7 9.5 3.5–18.7
6 Comm 8.9 7.5 0–21 3.3 1.0 0 –9 10.2 9.4 3.9–17.1
Clinical 10.0 10.0 0–23 4.1 3.0 0–15 10.8 10.0 4.5–22.0
Comm � community-base sample; Clinical � Alzheimer’s Disease Center clinical sample.
TABLE 8. Regression Coefficients Comparing Screening Scores Versus Number of Errors on
Six-Item Screener
Community-based Sample Alzheimer’s Disease Center Sample
Value R
Value R
344 MMSE �2.4 �0.001 59.7% 651 �3.1 �0.001 75.4%
273 Word List Recall �1.9 �0.001 35.9 599 �2.7 �0.001 64.2%
158 Blessed 0.5 �0.001 19.6% 452 1.2 �0.001 45.2%

and specificity of shorter scales. Initially, these
attempts included instruments of 10-to-15 items
(eg, Short Portable Mental Status Questionnaire)
rather than the 30-item scales such as the Mini-
Mental State Examination or Blessed Dementia
Scale. By the early 1980s, scientists were exploring
scales as short as six items. These early efforts were
limited by the use of small clinical samples of
nursing home residents or medical inpatients,
and the developers were typically predicting scores
on longer screening tests rather than predicting
the actual clinical determination of cognitive im-
pairment or dementia.
In the 1990s, several authors reporting from
Alzheimer Disease Research Centers were able to
report on the sensitivity and specificity of a re-
duced item Mini-mental state examination.
The studies by Galasko and Fillenbaum were
limited to patients with Alzheimer’s disease who
had been referred to the clinical center whereas
the study by Wells coupled data from an Alzhei-
mer Disease Research Center with data from the
Epidemiologic Catchment Area study. Although
all three of these studies demonstrated that a
reduced-item Mini-mental State examination had
acceptable sensitivity and specificity for identifying
patients with Alzheimer’s disease, the study by
Wells requires the calculation of a discriminant
function score and includes a total of nine items.
Although limited to Alzheimer disease subjects,
the studies by Fillenbaum and Galasko
previously demonstrated that three-item recall
and orientation items provide excellent discrimi-
nation for normal subjects as compared with those
with cognitive impairment or Alzheimer Disease.
More recently Buschke et al,
reported the
performance of a 4-minute, four-item, delayed
free- and cued-recall test of memory impairment.
The study sample included 286 volunteers re-
cruited from physician offices and senior centers
and 197 subjects from the local community iden-
tified through Medicare lists. All subjects com-
pleted a neurologic evaluation to establish a diag-
nosis of dementia. These authors reported a
sensitivity of 86% and a specificity of 91% in
diagnosing dementia using a cut-off score of 5
(range of possible scores 0–8). This level of diag-
nostic accuracy has not been demonstrated in an
unselected community-based population. How-
ever, the primary drawback of this test for screen-
ing large research populations is the requirement
that patients read a visual cue card containing the
four items to be recalled, and that testing of recall
be delayed from 3 to 4 minutes after reading the
card. This makes completion by telephone or
in-person more cumbersome.
One of the primary advantages of the six-item
screener compared with the other brief cognitive
screens mentioned above is its suitability for ad-
ministration over the telephone. There are at least
three other instruments reported in the literature
that are designed specifically to assess cognitive
function via telephone administration. These in-
clude the Telephone Interview for Cognitive Status
the Minnesota Cognitive Acuity Screen
and the Structured Telephone Inter-
view for Dementia Assessment (STIDA).
three instruments have reported acceptable sensi-
tivity and specificity although the MCAS and
STIDA studies did not target a representative
community-base sample of older adults. The pri-
mary disadvantage of these three instruments is
their length. Although all three instruments elim-
inate items that would require props or face-to-
face administration, the length of these instru-
ments approximate that of the MMSE and thereby
require 10 to 20 minutes to complete. A short-
STIDA has also been described but this instrument
still requires 10 min to complete and the reported
specificity falls to 0.77. Each of these longer tele-
phone assessments could readily be considered as
a second-stage cognitive screen to be used in
tandem for those older adults scoring positive on
the six-item screener.
In conclusion, we have demonstrated that a
brief six-item screener that can be readily admin-
istered face-to-face or by telephone has
diagnostic-test characteristics comparable with the
MMSE and other longer scales designed to iden-
tify cognitive impairment or dementia. Sensitivity
and specificity change precipitously but predict-
ably as one varies the number of errors used as a
cut-off point. This scale, which is a subset of the
full MMSE, provides investigators with an efficient
and accurate mechanism to identify patients with
probable cognitive impairment.
1. Stuss DT, Meiran N, Guzman A, et al. Do long
test yield a more accurate diagnosis of dementia than
short tests? Arch Neurol 1996;53:1033–1039.

2. Lachs MS, Feinstein AR, Cooney LM, et al. A
simple procedure for general screening for functional
disability in elderly patients. Ann Intern Med
3. Klein LE, Roca RP, McArthur J, et al. Diagnos-
ing dementia: univariate and multivariate analyses of the
mental status examination. J Am Geriatr Soc
4. Unutzer J, Katon W, Williams JW, et al. Im-
proving primary care for depression in late life: the
design of a multi-center randomized trial. Med Care
5. Hall KS, Hendrie HC, Brittain HM, et al. The
development of a dementia screening interview in two
distinct languages. Int J Meth Psychiatric Res 1993;3:1–
6. Hall KS, Ogunniyi AO, Hendrie HC, et al. A
cross-cultural community based study of dementias:
Methods and performance of the survey instrument
Indianapolis, USA, and Ibadan, Nigeria. Int J Meth
Psychiatric Res 1996;6:129–142.
7. RothM, Tym E, Mountjoy CO, et al. CAMDEX:
a standardized instrument for the diagnosis of mental
disorder in the elderly with special reference to the early
detection of dementia. Br J Psychiatry 1986;149:698–709.
8. Folstein MF, Folstein SE, McHugh SE. Mini-
mental state. A practical method for grading the cogni-
tive state of patients for the clinician. J Psychiatr Res
9. Blessed G, Tomlinson BE, Roth M. The asso-
ciation between quantitative meausures of dementia and
of senile change in the cerebral gray matter of elderly
subjects. Br J Psychiatr 1968;114:797.
10. Gurland BJ, Kuriansky J, Sharpe L, et al. The
Comprehensive Assessment and Referral Evaluation
(CARE) rationale. Rationale, development and reliability.
Int J Aging Hum Dev 1977;1:9–42.
11. Evans DA, Funkenstein HH, Albert MS, et al.
Prevalence of Alzheimer’s disease in a community pop-
ulation of older persons: higher than previously re-
ported. JAMA 1989;262:2551–2556.
12. Hendrie HC, Osuntokun BO, Hall KS, et al.
Prevalence of Alzheimer’s disease and dementia in two
communities: Nigerian Africans and African Americans.
Am J Psychiatry 1995;152:1485–1492.
13. Hendrie HC, Ogunniyi A, Hall KS, et al.
Incidence of dementia and Alzheimer disease in 2 com-
munities: Yoruba residing in Ibadan, Nigeria, and black
persons residing in Indianapolis, Indiana. JAMA
14. Callahan CM, Hall KS, Hui SL, et al. Rela-
tionship of age, education, and occupation with demen-
tia among a community-based sample of African-
Americans. Arch Neurol 1996;53:134–140.
15. Unverzagt FW, Hall KS, Torke AM, et al.
Effects of age, education, and gender on CERAD neuro-
psychological test performance in an African American
sample. Clin Neuropsychol 1996;10:180–190.
16. Morris JC, Heyman A, Mohs RC, et al. The
Consortium to Establish a Registry for Alzheimer’s Dis-
ease (CERAD). Part I. Clinical and neuropsychological
assessment of Alzheimer’s disease. Neurology
17. Unverzagt FW, Gao S, Baiyewu O, et al.
Prevalence of Cognitive Impairment without Dementia:
Data from the Indianapolis-Ibadan Dementia Project.
Neurology 2001;57:1655-1662.
18. American Psychiatric Association: Diagnostic
and Statistical Manual of Mental Disorders, Third Edi-
tion, Revised (DSM-III-R). Washington DC, American
Psychiatric Association, 1987.
19. ICD-10. The International Statistical Classifica-
tion of Diseases and Related Health Problems. American
Psychiatry Press. Volumes 1 and 2, 1991;V:1992:3.
20. McKhann G, Drachman D, Folstein M, et al.
Clinical diagnosis of Alzheimer’s disease. Neurology
21. Borson S, Scanlan J, Brush M, et al. The
mini-cog: a cognitive “vital signs”measure for dementia
screening in multi-lingual elderly. Int J Geriatr Psychiatry
22. Buschke H, Kuslansky G, Katz M, et al.
Screening for dementia with the Memory Impairment
Screen. Neurology 1999;52:231–238.
23. Wells JC, Keyl PM, Chase GA, et al. Discrimi-
nant validity of a reduced set of Mini-Mental Status
Examination items for dementia and Alzheimer’s Dis-
ease. Acta Psychiatr Scand 1992;86:23–31.
24. Solomon PR, Hirschoff A, Kelly B, et al. A
7-minute neurocognitive screening battery highly sensi-
tive to Alzheimer’s Disease. Arch Neurology
25. Pfeiffer E. A short portable mental status ques-
tionnaire for the assessment of organic brain deficit in
elderly patients. J Am Geriatr Soc 1975;23:433–441.
26. Schulman KI. Clock-drawing: is it the ideal
cognitive screening test? International J Geriatr Psychi-
atry 2000;15:548–561.
27. Scanlan J, Borson S. The mini-cog: receiver
operating characteristics with expert and naïve raters. Int
J Geriatr Psychiatry 2001;16:216–222.
28. Tombaugh TN, McIntyre NJ. The Mini-
Mental State Examination: a comprehensive review.
J Am Geriatr Soc 1992;40:922–935.
29. Katzman R, Brown T, Fuld P, et al. Validation
of a short Orientation-Memory-Concentration Test of

cognitive impairment. Am J Psychiatry 1983;140:734–
30. Magaziner J, Bassett SS, Hebel JR. Predicting
performance on the mini-mental state examination. Use
of age- and education-specific equations. J Am Geriatr
Soc 1987:996–1000.
31. Braekhus A, Laake K, Engedahl K. The Mini-
Mental State Examination: identifying the most efficient
variables for detecting cognitive impairment in the el-
derly. J Am Geriatr Soci 1991;40:1139–1145.
32. Galasko D, Klauber MR, Hofstetter R, et al.
The mini-mental state examination in the early diagnosis
of Alzheimer’s disease. Arch Neurol 1990;47:49–52.
33. Fillenbaum GG, Wilkinson WE, Welsh KA,
et al. Discrimination between stages of Alzheimer’s
Disease with subsets of mini-mental state examination
items. Arch Neurol 1994;51:16–21.
34. Brandt J, Spencer M, Folstein M. The tele-
phone interview for cognitive status. Neuropsychiatry
Neuropsychol Behav Neurol 1988;1:111–117.
35. Knopman DS, Knudson D, Yoes ME, et al.
Development and standardization of a new telephonic
cognitive screening test: the Minnesota Cognitive Acuity
Screen (MCAS). Neuropsychiatry Neuropsychol Behav
Neurol 2000;13:286–296.
36. Go RC, Duke LW, Harrell LE, et al. Devel-
opment and validation of a Structured Telephone
Interview for Dementia Assessment (STIDA): the
NIMH Genetics Initiative. J Geriatr Psychiatry Neurol
APPENDIX A. TABLE 1. Six-Item Screener
1. I would like to ask you some questions that ask you to use your memory. I am going to name three objects.
Please wait until I say all three words, then repeat them. Remember what they are because I am going to ask
you to name them again in a few minutes. Please repeat these words for me: APPLE—TABLE—PENNY.
(Interviewer may repeat names 3 times if necessary but repetition not scored.)
Did patient correctly repeat all three words? Yes No
Incorrect Correct
1. What year is this? 0 1
2. What month is this? 0 1
3. What is the day of the week? 0 1
What were the three objects I asked you to remember?
4. Apple � 01
5. Table � 01
6. Penny � 01