Insight in Dementia: When Does It Occur? Evidence for a Nonlinear Relationship Between Insight and Cognitive Status

Abstract

Lack of insight or impaired awareness of deficits in patients with dementia is a relatively neglected area of study. The aim of this study was to evaluate insight in a group of demented patients with two assessment scales and to assess their relationship with the cognitive level of disease severity. Sixty-nine consecutive patients affected by Alzheimer's disease (n = 37) and vascular dementia (n = 32) with a wide range of cognitive impairment (MMSE = 17.0 +/- 6.4) were recruited. Insight was evaluated with the Guidelines for the Rating of Awareness Deficits (GRAD)--specifically targeted to memory deficits--and the Clinical Insight Rating scale (CIR), evaluating a broader spectrum of insight (reason for the visit, cognitive deficits, functional deficits, and perception of the progression of the disease). In the whole sample, GRAD and CIR were significantly associated with MMSE (Spearman's coefficient = .51, p < .001; and r = -.55, p < .001) and with Clinical Dementia Rating scale (-.57, p < .001; and r = .57, p < .001) respectively. The shape of the relationship of MMSE with CIR and GRAD scales was assessed with spline smoothers suggesting that the relationship follows a trilinear pattern and is similar for both scales. Insight was uniformly high for MMSE scores > or = 24, showed a linear decrease between MMSE scores of 23 and 13, and was uniformly low for MMSE scores < or = 12. The trilinear model of the association between insight and cognitive status reflects more closely the observable decline of insight and can provide estimates of when the decline of insight begins and ends.

Full text

Journal
of
Gerontology:
PSYCHOLOGICAL SCIENCES
1999,
Vol.
54B, No. 2, P100-P106
Copyright 1999
by
The Gerontological Society
of America
Insight in Dementia: When Does It Occur?
Evidence for a Nonlinear Relationship
Between Insight and Cognitive Status
Orazio Zanetti,
1
Barbara Vallotti,
2
Giovanni B. Frisoni,
1
Cristina Geroldi,
1
Angelo Bianchetti,
1
Patrizio Pasqualetti,
3
and Marco Trabucchi
4
'Alzheimer Disease Unit
IRCCS,
"S.
Giovanni di Dio,"
"S.
Cuore Fatebenefratelli" Hospital, Brescia, Italy.
institute of Geriatrics and Gerontology, University of
Florence,
Florence, Italy.
3
AFaR-CRCCS, Neurological Division,
"S.
Giovanni Calibita" Hospital, Isola Tiberina, Rome, Italy.
4
Geriatric Research Group, Brescia, Italy.
Lack of insight or impaired
awareness
of deficits in patients with dementia is a
relatively
neglected area of study. The aim of
this study
was
to evaluate insight in a group of demented patients with
two
assessment
scales
and
to assess
their
relationship
with the cognitive level of
disease
severity.
Sixty-nine consecutive patients affected
by
Alzheimer's
disease
(n
=
37) and vas-
cular dementia (n
=
32) with a
wide
range of cognitive impairment (MMSE
=
17.0 ±
6.4)
were
recruited.
Insight
was
evalu-
ated with the Guidelines for the Rating of Awareness Deficits
(GRAD)—specifically
targeted to memory deficits—and the
Clinical Insight Rating scale (CIR), evaluating a broader spectrum of insight (reason for the
visit,
cognitive
deficits,
func-
tional
deficits,
and perception of the progression of the
disease).
In the whole
sample,
GRAD and CIR
were significantly
as-
sociated with MMSE (Spearman's coefficient = .57, p < .001; and r = -.55, p < .001) and with
Clinical
Dementia Rating
scale
(-.57,
p <
.001;
and r
=
.57,
p <
.001)
respectively.
The shape of the
relationship
of MMSE with CIR and GRAD
scales
was
assessed with spline smoothers suggesting that the relationship follows a tritinear pattern and
is
similar for
both
scales.
Insight
was
uniformly high for MMSE
scores >
24,
showed
a
linear
decrease
between MMSE
scores
of 23 and
13,
and
was
uniformly low for MMSE
scores
<
12.
The tritinear model of the association between insight and
cognitive
status reflects
more
closely
the
observable
decline of insight and can provide
estimates
of when the decline of insight
begins
and ends.
L
ACK of insight or impaired awareness of deficits in persons
-/ with dementia is a relatively neglected area of study
(McDaniel et al., 1995). Most of the current research on dementia
focuses primarily on biological issues, like etiology or pathogene-
sis,
and more recently on behavioral and therapeutical aspects.
Little attention has been paid to the phenomenology of the de-
mented individual's subjective experience, or serf-perception of
illness (Bahro, Silber, & Sunderland, 1995). It is therefore not sur-
prising that insight is a concept lacking precise operational defini-
tion (Mullen, Howard, David, & Levy, 1996). The terms insight
or awareness of deficits have been used interchangeably, referring
to lack of knowledge or recognition of one's deficits (McGlynn &
Schacter, 1989). Anosognosia similarly means lack of knowledge
of disease, but has been used more often to describe a failure to
acknowledge a particular deficit, usually the motor (McGlynn &
Schacter, 1989; Babinski, 1914). Babinski first used the term in
1914 to describe an absence of awareness in a hemiplegic patient
who had suffered a stroke (Prigatano & Schacter, 1991).
However, the term is now used more generally to include all neu-
ropsychological and neurological deficits (Cotrell, 1997). Some
authors use the terms anosognosia and unawareness interchange-
ably (McGlynn & Schacter, 1989). Recently, David (1990),
analyzing the association between insight and psychosis, has pro-
moted a complex concept of insight as an amalgam of three con-
structs: the ability to identify certain mental events as pathologi-
cal,
the recognition by the individual that he or she has a mental
illness, and the degree of compliance with treatment Insight was
also defined as the ability to judge the presence and the severity of
dementia (DeBettignies, Mahurina, & Pirozzolo, 1990) and, more
recently, as the awareness of and attitudes toward one's mental
symptoms (Mullen et al., 1996). A practical definition considers
insight as the ability to judge both the presence (symptoms) and
the severity (functional impairment) of illness (Babinski, 1914;
Critchley, 1953; DeBettignies et al., 1990; Fisher, 1989; Gainotti,
1972;
Mangone et al., 1991). Foley (1992) defines awareness or
insight as "the capacity to discern the true nature of the situation,
or as applied to dementia, the recognition of the fact, degree, and
implications of one's own illness."
The mechanism of insight in demented patients is still un-
known (Mullen et al., 1996). Attempts to define the causative
mechanisms of insight have shifted from the initial focus on the
biological explanation to a psychodynamic interpretation; re-
cently, interest has shifted back to awareness deficits as a neu-
ropathological problem (Prigatano & Schacter, 1991). Several
studies showed that patients with Alzheimer's disease (AD) and
poor insight had significantly more severe deficits on a frontal
lobe-related neuropsychological test (Auchus, Goldstein, Green,
& Green, 1994; Lopez, Becher, Sumsak, Dew, & Dekosky,
1994;
Mangone et al., 1991; Michon, Deweer, Pillon, Agid, &
Dubois, 1994; Ott et al., 1996; Starkstein,
Federoff,
Price,
Leiguarda, & Robinson, 1992). Other studies did not replicate
these findings, suggesting the main involvement of right hemi-
sphere dysfunction (Migliorelli et al., 1995; Reed, Jagust, &
Coulter, 1993). Usually, impaired insight due to frontal lobe dys-
function is termed confabulation whereas impaired insight after
right hemisphere dysfunction is termed anosognosia. Both con-
P100
by guest on December 21, 2015http://psychsocgerontology.oxfordjournals.org/Downloaded from

INSIGHT IN DEMENTIA
P101
tabulation and anosognosia may contribute to the impaired in-
sight in AD (Mangone et al., 1991).
Schacter and Prigatano (1991) warn that unawareness is not a
unitary entity but probably consists of various forms or types.
Recently, Starkstein, Sabe, Chemerinski, Jason, and Leiguarda
(1996) have suggested that insight may be a complex function:
Unawareness of cognitive deficits and unawareness of behavioral
problems may constitute independent phenomena in Alzheimer's
disease, the former related to the severity of intellectual impairment
and the latter probably associated with the disinhibition syndrome.
The level of insight has been shown to be significantly associated
with the severity of dementia. A number of studies, in fact, confirm
the generally accepted belief that patients with Alzheimer's disease
experience a progressive loss of insight as the severity of dementia
increases (Mangone et al., 1991; McDaniel et al., 1995; Starkstein
et al., 1996). However, as insight is not an "all-or-nothing" phe-
nomenon, little is known about the shape and the kind of relation-
ship between insight and degree of cognitive impairment The aim
of this study was to evaluate the level of insight in a group of de-
mented patients with two assessment scales, one (Verhey,
Rozentaal, Ponds, & Jolles, 1993) specifically targeted to memory
deficits, the other (Ott & Fogel, 1992) evaluating a broader spec-
trum of insight, and to examine their relationship with die level of
cognitive impairment
METHODS
Sixty-nine patients affected by Alzheimer's disease (AD) and
vascular dementia (VD) according to DSM-PV criteria (American
Psychiatric Association, 1994) were consecutively recruited at the
Alzheimer Dementia Unit of Sacro-Cuore Fatebenefratelli Institute
of Brescia, Italy. Thirty-seven patients were affected by AD proba-
ble or possible according to N1NCDS-ADRDA criteria (McKhann
et al., 1984) and 32 were affected by VD probable or possible ac-
cording to NINDS-AIREN criteria (Roman et al., 1993).
Laboratory and neuroimaging evaluation (e.g., computed tomo-
graphic scan, thyroid hormones, folate and vitamin B12 levels,
syphilis serology) were conducted to rule out specific neurology,
neoplastic, infectious and metabolic causes of dementia. Patients
with delirium, aphasia, or comprehension deficits that might com-
promise the administration of insight rating scales were excluded.
Patients with history of severe head injury, alcoholism, major psy-
chiatric illness, and epilepsy were also excluded from the study.
Patients underwent a multidimensional assessment of cogni-
tive,
physical, and social health; for the present study, the follow-
ing variables were assessed:
1.
Sociodemographic variables (gender, age, education) were
recorded.
2.
Cognitive status was measured with the Mini-Mental State
Examination (MMSE) (Folstein, Folstein, & McHugh, 1975).
3.
Clinical Dementia Rating scale (CDR; Hughes, Berg,
Danziger, Coben, & Martin, 1982) was used to provide a score
ranging from 0 to 3 (0 = normal, 0.5 = questionable dementia, 1
= mild, 2 = moderate, 3 = severe dementia), which is based on
combined ratings of
psychic,
social, and functional aspects of the
patient by a clinician and give a global dementia severity level.
4.
Functional status: patients underwent indirect assessment of
their functional competence. The indirect assessment included
six basic and eight instrumental activities of daily living (BADL
and IADL), assessed with the Katz index (Katz, Ford,
Moskowitz, & Jackson, 1963) and the Lawton and Brody (1969)
scale, respectively. Dependency was defined as the inability to
carry out activities of daily living without regular help of another
person. Information was collected from the primary caregiver.
5.
Behavioral disturbances were evaluated with the
Neuropsychiatric Inventory Scale (NPI; Cummings et al., 1994).
6. Depressive symptoms were assessed using the Geriatric
Depression Scale (GDS; Yesavage et al., 1983).
7.
Assessment of insight was done by means of two assessment
instruments: the Guidelines for the Rating of Awareness Deficits
(GRAD; Verhey et al.,
1993;
Verhey, Ponds, Rozendaal, & Jolles,
1995) and the Clinical Insight Rating scale (CIR; Ott and Fogel,
1992;
Ott et al., 19%). Both scales are in a semistructured inter-
view format and are preceded by an interview with the patient's
primary caregiver during which the reason of visit, duration of ill-
ness,
rate of progression of cognitive deficits, functional impair-
ment (BADL and IADL), and behavioral disturbances (NPI) are
investigated. Before insight assessment, the clinician evaluates the
patient's global disease severity (MMSE and CDR). Insight scores
are derived by the comparison of the patient's responses and the
clinical information gathered with the help of the caregiver.
The GRAD is composed of the following four questions: (a)
"Please tell me about the problems you are here for." (b) When
the patient has other complaints, not directly related to demen-
tia: "Do you have other complaints?" (c) When the patient has
no spontaneous complaints about his or her cognitive functions:
"How is your memory functioning? Do you think you have a
poor memory?" And (d) when the patient denies deficits of
memory or other cognitive functions: "So, there are no memory
problems at all? Is everything going all right for you?" After
these questions the complaints are discussed more extensively in
an open interview, in which the clinician tries to get an impres-
sion of the degree and the nature of cognitive symptoms.
Scoring is made directly after the interview. Awareness is judged
to be intact (score = 4) when the cognitive problems were men-
tioned spontaneously by the patient in reply to the opening ques-
tion and when history of the caregiver corresponds to that of the
patient. When the patient commented spontaneously about his
or her memory in reply to the opening question, but there were
apparent discrepancies between the patient's and caregiver's
anamnesis, awareness was scored as mildly impaired (score =
3).
Awareness was scored as 2 (severely impaired) when the pa-
tient uttered no complaints in response to the opening question,
and when there were clear discrepancies between the patient's
and the caregiver's history. When the patient denied any explicit
awareness, insight was scored as 1 (absent).
The CIR is a scale that assesses the patient's awareness re-
garding the following aspects: (a) the reason for the visit to see
doctor; (b) his or her cognitive deficits; (c) his or her functional
deficits; and (d) his or her perception of the progression of the
disease. Each item is ranked from 0 to 2, yielding a total rating
that ranges from 0 (insight fully preserved) to 8 (insight totally
absent). Ratings on the insight scales were carried out by a clin-
ician (B.V.) based on her judgement of the patient's degree of
awareness on each item after an interview with the patient and
the primary caregiver.
The Italian version of the two scales has been previously vali-
dated in a group of 20 demented subjects, 10 affected by AD and
by guest on December 21, 2015http://psychsocgerontology.oxfordjournals.org/Downloaded from

P102 ZANETTIETAL.
10 affected by VD (Vallotti, Zanetti, Bianchetti, & Trabucchi,
1997).
The interrater intra-class correlation coefficient was .73
for the GRAD and
.80
for the
CTR.
The test-retest coefficient was
.73 for GRAD and .89 for
CTR.
Both scales proved to be reliable
for
the
rating of insight.
8. A subgroup of 36 participating patients underwent an ex-
tensive neuropsychological assessment (for details of the tests,
see Binetti et al., 1993). Short-term memory was assessed by
auditory-verbal forward digit span and visuospatial span
(Corsi's block-tapping test). Episodic memory was tested with
the logical memory test (recall of a short story) and with the 20-
minutes delayed recall of the Rey complex figure. A 30-item
version of the Boston naming test provided a measure of nam-
ing and semantic memory. The copy of the Rey figure was used
to assess visuospatial abilities. The Raven's colored progressive
matrices test, the attentional matrices test, and the token test
were also included in the neuropsychogical battery. "Executive"
function was evaluated by the Wisconsin Card-Sorting Test
(WCST), and PEL verbal fluency test. The WCST is a problem-
solving task that measures the ability to identify abstract cate-
gories and shift cognitive set. A shortened version of WCST
with 64 cards was adopted, computing an index of perseveration
(number of errors/number of
perseveration).
The verbal fluency
for letters (PEL) was assessed by recording the number of words
produced in the course of one minute for each letter.
Statistical Analysis
Descriptive and correlation analysis was carried out with
SPSS Software (version 5.0). The relationships among variables
were assessed with Pearson's r and the Spearman correlation co-
efficient. The associations between neuropsychological tests and
insight as measured with GRAD and CIR have been corrected
for Bonferroni multiple (n
=
10) comparisons test, and the criti-
cal p value for statistical significance was set at p = .005. The
following statistical analysis was carried out with S+ (version 4)
for Windows. The relationship of insight scales with cognition
was explored with cubic spline smoothers (Hastie & Tibshirani,
1990).
A smoother is a tool for summarizing the trend of a de-
pendent variable (in the present case, insight scales) as a func-
tion of one or more predictors (here cognitive performance).
Smoothers produce an estimate of the trend that is less variable
than the dependent variable itself (Hastie & Tibshirani,
1990).
In
the case of spline smoothers, points called knots divide the
"real" shape of
the
association, and a number of cubic functions
are fitted between the knots. The algorithm identifies where the
knots that allow maximization of the fitting lie. The higher the
number of
knots,
the higher the fitting of the resulting function,
but the more uneven its shape. The lower the number of knots,
the lower the fitting of the function but the smoother its shape.
The exploration process of the relationship between two vari-
ables consists of finding out the number of knots that gives, at
the same time, the highest possible fitting with the smoothest
function. Similarly to the usual generalized linear models, spline
smoothers allow one to test the effect of covariates on the rela-
tionship between the dependent and the independent variables.
The effect of diagnosis (AD or VD) on the shape of the rela-
tionship between insight scales and MMSE was assessed in a
multivariate generalized additive model with a spline smoother of
MMSE, diagnosis (coded as a dichotomous variable), and their
interaction as independent variables. Spline smoothers can be
compared to conventional linear models by usual methods (test-
ing the significance of
the
increase of fitting, i.e., the decrease of
deviance, by chi-square test). In the present study, we compared
spline smoothers with piecewise linear models. Different piece-
wise linear models were also compared. The optimal point of
discontinuity in piecewise linear models was chosen on die basis
of the best-fit, that is, the lowest deviance. In this analysis the
level of statistical significance was set at/? = .05.
RESULTS
Table 1 shows the clinical and sociodemographic character-
istics of the sample. No statistical differences were found be-
tween the AD and the VD patients, except for age (p < .05). In
particular, no significant differences were found for the GRAD
and the CIR in the two groups. The great majority of patients
(94%) were in the mild to moderate stage of dementia severity.
The goal of
the
first
step of
the
analysis was to confirm the as-
sociation of insight with disease severity and with psychological
variables. This preliminary data analysis was performed assuming
a linear association between insight and cognitive impairment.
In the whole sample, the level of insight as measured by the
GRAD was significantly associated with dementia severity as
measured by the MMSE (Spearman's coefficient =
.51,
p <
.001) and the CDR (Spearman's coefficient = -.57, p < .001).
The CIR scores also showed a good correlation with disease
severity (r
=
-.55,
p
<
.001 and r =
.57,
p
<
.001,
respectively).
No relationship was found between insight and the presence of
depressive symptoms with both the GRAD (r
=
.15,
p = .24)
and the CIR (r
=
-.22,
p
=
.09) scales.
After Bonferroni correction for multiple comparisons, both
the GRAD and CIR scales showed a significant correlation (p <
.005) with tests measuring planning and abstract thinking
(Wisconsin Card-Sorting Test), and language comprehension
(Token test). Moreover, the CIR was significantly associated
with attentional matrix score (r = -.40) and constructional
apraxia (Rey's
figure;
r
=
.41), and the GRAD was significantly
associated with the Boston Naming Test score (r
=
.50).
Table
1.
Sociodemographic
and
Clinical Characteristics
of Alzheimer's Disease
(AD)
and Vascular Dementia
(VD)
Patients
Female
Age (years)
Education (years)
Disease duration (months)
Mini-Mental State Examination
BADL (functions lost)
IADL (functions lost)
Neuropsychiatric Inventory Scale
Geriatric Depression
Scale
a
GRAD
b
CIR
C
Clinical Dementia Rating Scale
1
2
3
AD (n =
37)
28 (75.7)
74.8
± 8.5
6.2 + 3.9
33.4
± 22.5
16.1
+ 6.2
1.2 ± 1.7
3.9 ± 2.7
23.1 ±22.8
8.1 ±5.2
2.7
±1.1
3.1 ±2.6
24 (64.9)
11 (29.7)
2
(5.4)
VD
(n = 32)
Total
(n
=
69)
24 (75.0)
78.8
±6.1
5.2 ± 2.7
30.3
± 16.9
18.0
±6.7
1.7 ± 1.6
4.7
± 2.5
30.4
± 22.3
10.1
±5.7
2.5
± 1.1
2.9
± 2.4
17 (53.2)
13 (40.5)
2
(6.3)
52 (75.4)
76.7
±7.8
5.7 ± 3.4
32.0 ±
2.0
17.0
±6.4
1.4 ±1.7
4.3
± 2.6
26.5
± 22.7
9.1 ±5.5
2.6
± 1.1
3.0 ±2.5
41 (59.4)
24 (34.8)
4
(5.8)
Note:
Data
are
means
+ and SD or
n
(%).
Performed
in
30 AD
and
28 VD
patients.
b
Guidelines
for
the Rating of Awareness
Deficits.
c
Clinical Insight Rating
Scale.
by guest on December 21, 2015http://psychsocgerontology.oxfordjournals.org/Downloaded from

INSIGHT IN DEMENTIA P103
However, as shown in Figures 1A and IB, the association be-
tween insight and cognitive status was not linear. An exploratory
analysis with spline smoothers showed that the relationship was
roughly trilinear, with no association between insight and cogni-
tion at both ends of
the
MMSE distribution, and a linear associa-
tion in between.
The GRAD scale scores distribution into only 4 levels pre-
vented any further analysis assuming a normal distribution of
the response variable. However, the CIR scale scores distribu-
tion, albeit into 9 levels, could be better approximated to a nor-
mal distribution. Therefore, the following analysis was carried
out on the CIR scale only.
A multivariate generalized additive model with the spline
smoother of MMSE and diagnosis (AD or
VD)
as the indepen-
dent variable indicated that the relationship was not different in
the two subgroups (difference of deviances =
1.5;
df= I; p =
.23),
indicating mat the intercept of the smoother was not dif-
ferent in the two groups. The addition in the model of the inter-
action between the spline smoother and MMSE also failed to
reach significance (difference of deviances = 2.2; df
=
2; p =
.33),
indicating that the shape of the smoother was not different
in the two groups. These observations supported the following
joint analyses of the two groups. The fitting of piecewise linear
models of MMSE on CIR was tested and models of increasing
complexity and descriptiveness of the data were compared.
The cutoff points of
12
and 24 for the MMSE were chosen by
comparing two-segment piecewise linear models with cutoffs
ranging from 9 to 15 and from 21 to 27, and by choosing the
cutoff associated with the lowest deviance (Figure 2). Deviance
ranged between 249.3 and 268.5 for MMSE scores between 9
and 15, with the lowest deviance (247.3) being observed for the
MMSE score of 12. Deviance ranged between 256.6 and 249.6
for MMSE scores between
21
and 27, with the lowest deviance
(242.7) being observed for the MMSE score of
24.
Table 2 shows fittings of linear and spline smoothers. The
first model assumes a linear relationship between MMSE and
CIR scales. The second and third models assume that the left-
most and
rightmost
portions of
the
MMSE distribution (MMSE
<12 and MMSE >24, respectively) are flat and that the remain-
ing portion of the MMSE distribution has a linear relationship
with the CIR scale. Both the second and third models had a sig-
nificantly lower deviance (i.e., better fit) than the first one. The
fourth model was built to compound the second and third ones,
and assumes that the relationship between MMSE and CIR
scales is flat at both ends of the MMSE distribution and is lin-
ear in between. Table 2 shows that this trilinear model has a sig-
nificantly better fit than both previous models. The last two to
be tested were generalized additive models with a spline
smoother. Generalized additive models often have a better fit
than linear models because no a priori assumption regarding
the shape of
the
relationship is made. Indeed, the
fifth
model, in
which a spline is fitted, has lower deviance than the trilinear
model, which, however, was not significant. This indicates that
the latter does not perform significantly worse than the best
possible model.
DISCUSSION
This study demonstrates that the association of disease insight
with cognitive functions follows a nonlinear pattern and MMSE
cutoff points can be identified between preserved, moderately
impaired and absent insight. The trilinear model describes the as-
sociation between insight and cognition as proceeding through
three periods: an initial period of stability before detectable de-
cline, a period of decline, and a final period of stability during
8
CO
CO
a:
Alzheimer's
Vascular
All
\
\
\
10 15 20 25
Mini Mental State Examination
30
B
•~ Alzheimer's
Vascular
- All
10 15 20
Mini Mental State Examination
25
30
Figure
1.
Relationship of Clinical Insight Rating scale (A) and Guidelines for the Rating of Awareness Deficits (B) with MMSE score in Alzheimer's and vascular
dementia
patients.
Note:
The lines denote a smoother function (cubic spline, 4 knots) in all patients and in the two subgroups. Error bars of
the
smoother function for
all patients are also shown.
by guest on December 21, 2015http://psychsocgerontology.oxfordjournals.org/Downloaded from

P104
ZANETTIETAL.
0.415
r square
0.355
0.42
11
12 13
MMSE
r square
23
24 25
MMSE
Figure
2.
Identification
of
the
two
cutoff scores with
the
best piecewise linear
fitting of
Mini
Mental State Examination score
on
Clinical Insight Rating scores.
Note:
Higher
r
squares indicate better model fitting.
Table
2.
Comparisons
of
Models Describing
the
Association Between
Mini-Mental State Examination and Clinical Insight Rating Scale
Scores
in
69 Dementia Patients
Model
1.
Linear
3.
Linear
4.
Linear
5.
Additive
Shape Deviance
Nv
251.4
^V 244.5
^\ 230.6
~~"\ 228.9
r
2
.3986
AfiQA
.4151
.4483
df
67
67-i
kooi
Y-n.s.
64 J
P
-.0002
which there
is no
further detectable decline.
Our
data confirm
previous findings
by
Brooks, Kraemer, Tanke,
and
Yesavage
(1993)
who
suggested that
the
trilinear model
for
analyzing
lon-
gitudinal data
in a
sample of Alzheimer's
is
superior
to the com-
monly used linear model that includes
the
flawed assumption
that decline is uniform throughout the course of the disease.
Following
a
simple correlation analysis
and
considering
an a
priori linear association between insight
and
cognition
led to a
confirmation
of
previous research findings (Mangone
et al.,
1991;
Lopez
et al.,
1994;
Ott et al.,
1996). Decreased level
of
insight was significantly associated with severity
of
dementia
as
measured
by
MMSE
and CDR,
irrespective
of its
etiology
(Verhey
et al.,
1995). Some studies, however, have failed
to
show
a
significant relationship between dementia severity
and
reduced awareness
of
deficits (DeBettignies
et
al., 1990; Feher,
Mahurin, Inbody, Crook,
&
Pirozzolo,
1991;
Reed
et
al., 1993).
However, the main results
of
our study suggest that the
a
pri-
ori assumption
of
a linear association between insight
and
cog-
nition,
or
disease severity, cannot
be the
correct
one,
and sug-
gest that this association follows
a
trilinear
pattern.
The trilinear
model reflects more closely
the
observable decline
of
insight.
The second advantage
is
that the trilinear model
can
provide es-
timates of when the decline
of
insight begins and ends.
Insight
as
measured with
CIR is
nearly preserved
in the ini-
tial stages
of
disease;
afterwards, between MMSE scores
of
24
and 12, insight shows
a
linear progressive decay, followed
by a
plateau,
at
MMSE scores
of
12
and
less,
of
severe impaired
in-
sight. Interestingly,
the
cutoff associated with preserved insight
is
set at
MMSE score
of
24
or
more.
A
score
of
23
or
less—a
cutoff score evolving from research findings recommended
in
the original article—has generally been accepted
as
indicating
the presence
of
cognitive impairment (Tombaugh
&
Mclntyre,
1992).
However,
it
should
be
emphasized that, although
per-
sons with MMSE scores between
24 and
12
are
impaired, these
demented patients preserve some degree
of
disease insight.
On
the other hand,
CIR
scores between
0 and
1
can be
associated
with good insight,
CIR
scores between
2 and 4
with relatively
impaired insight,
and CIR
scores higher than
4 are
associated
with poor insight.
These
findings
could have important implications
for
research
with cognitively impaired subjects
and the
emerging problem
of patient autonomy
and
decisional capacity. The criteria
for
in-
formed consent
for
medical research that could
be
beneficial
for the patient imply
a
demanding test
of
capacity because they
envision
a
fully informed
and
reasoned decision
by the
subject.
However, many demented subjects
do not
have
the
capacity
to
satisfy this heroic standard (Bonnie, 1996),
and
investigators
are
left without
a
realistic standard
of
decisional capacity that can
be
used in cases involving subjects with cognitive impairment.
From
a
clinical perspective, knowledge
of
the
point
at
which
decline
of
insight
may
begin would allow health care profes-
sionals
and
patients' families
to
plan
for the
future course
of a
patient's disease (Brooks
et
al.,
1993;
Mullen
et
al., 1996).
Our data confirm that
the
level
of
depressive symptoms
and
lack
of
insight occur in AD and VD with comparable frequency.
Depression
did not
contribute independently
to
disease insight,
confirming previous findings
by
Cummings, Ross, Absher,
Gornbein,
and
Hadjiaghai (1995)
and Ott and
colleagues (1996),
who suggested that depression in AD
is
unrelated
to
patient
self-
awareness of illness. Our data confirm previous
findings
of Verey
and colleagues (1995), who evaluated
48
AD patients and
48
VD
patients, suggesting that depression, insight,
and
personality
do
not favor etiology of AD over that of
VD.
On
the
contrary,
com-
paring AD, VD,
and
controls, Wagner, Spangenberg, Bachman,
& O'Connel (1997) suggest that, independent
of
dementia sever-
ity, unawareness
of
cognitive deficits
is
disease specific. Wagner
and colleagues suggest that there
is a
disproportional degree
of
unawareness associated
with AD
when compared
with
VD.
The main result
of our
study—the nonlinear pattern
of the
association between disease insight
and
cognitive functions—
might explain some
of the
discordant findings
of
previous
re-
search carried
out
with correlation analysis assuming
an a
priori
linear association.
Meanwhile, some limitations
of the
instruments
for
insight
by guest on December 21, 2015http://psychsocgerontology.oxfordjournals.org/Downloaded from

INSIGHT IN DEMENTIA
P105
assessment used in the present study need to be addressed.
Although the GRAD appears to be a measure of awareness of
memory, given the restricted range of scores, the observed pat-
tern of relationship with MMSE might be related to ceiling and
floor effects; moreover, the restricted range of scores makes this
tool not sensitive enough to detect small changes. On the other
hand, the CIR, although less prone to ceiling or floor effects, is
a general measure of awareness because responses to questions
regarding a number of different areas (reason for the visit, cog-
nitive and functional deficits, and perception of the progression
of the disease) are collapsed into a single score.
Schacter and Prigatano (1991) warn that unawareness is not a
unitary entity but probably consists of various forms or types.
Loss of awareness for various skills may decline at different rates
and may be associated with loss of specific cognitive abilities.
This is consistent with the finding that individuals are often aware
of some deficits but not others; moreover, patients may be aware
of the existence of a deficit but unaware of the consequences of
the deficit (Feher et al., 1991; Schacter & Prigatano,
1991).
Along
the same line, Kotler-Cope and Camp (1995) suggest that aware-
ness of behavioral problems may be relatively preserved com-
pared with awareness of cognitive problems. Recently Starkstein
and colleagues (1996) demonstrated the presence of two domains
of insight, one related to cognitive deficits and the other related to
behavioral problems. Starskstein and colleagues (1996), along
with Vasterling, Seltzer, Foss, and Vanderbrook (1995), suggest
that insight may be a complex multifaceted function. The different
pattern of association found in our study, which was obtained as-
suming a linear association between insight and cognitive impair-
ment, supports the view that insight might not be a unitary entity
(Schacter,
1991;
Schacter & Prigatano, 1991). In fact, poor insight
was associated (in both scales) with frontal lobe dysfunction, but
the GRAD also showed association with naming and semantic
memory and the CIR with visuospatial abilities (right parietal dys-
function), which suggests the presence of different domains of in-
sight for memory and insight related to other aspects of disease
such as functional impairment, or awareness of disease progres-
sion. Other authors, however, demonstrated that patients with AD
have impaired awareness of both memory and functional deficits
and suggested that there may be a close relationship between
these two aspects. Ott and colleagues (1996) suggest the presence
of a common cognitive defect of self-monitoring that accounts for
the reduced-awareness phenomenon.
Further studies into the relationship between insight and local-
izing neuropsychological measures with functional brain imag-
ing are needed in order to better identify the nature of neu-
ropathology involved in reduced insight in demented patients
(Ott et al., 1996). Certainly we still need "reliable information
concerning the degree and the quality of awareness in various pa-
tient groups" (McGlynn & Schacter, 1989). As suggested by
Neundorfer (1997), more attention should be paid to the variabil-
ity in awareness within individuals and within diagnostic groups.
In conclusion, our data suggest that the shape of the associa-
tion between insight and cognitive impairment is not a linear
one but, on the contrary, follows a trilinear pattern with pre-
served insight in the mild stages of the dementia, followed by a
progressive loss of insight, and finally by a plateau of severe in-
sight impairment in the latest stages of dementia. Our findings
suggest the presence of MMSE cutoff points that can be used in
association with insight scales, in order to differentiate full in-
sight from declining to absent insight. The trilinear model of
the association between insight and cognitive status reflects
more closely the observable decline of insight and can provide
estimates of when the decline of insight begins and ends. These
results could be useful in developing instruments devised to as-
sess patient autonomy.
ACKNOWLEDGMENTS
Address correspondence to Orazio Zanetti, MD, Alzheimer Disease Unit
I.R.C.C.S. "S. Giovanni di Dio," "S. Cuore - Fatebenefratelli" Hospital, Via
Pilastroni, 4, 25123 Brescia, Italy. E-mail: ozanetti@master.cci.unibs.it
REFERENCES
American Psychiatric Association.
Diagnostic
and
statistical
manual of mental
disorders
(4th
ed.).
Washington,
DC:
Author.
Auchus, A. P., Goldstein, F. L., Green, J., & Green, R. C. (1994). Unawareness
of cognitive impairments in Alzheimer's disease. Neuropsychiatry,
Neuwpsychology,
and
Behavioral
Neurology,
7,
25-9.
Babinski, M. J. (1914). Contribution e l'etude des troubles mentaux dans
rhemiplegie organique cerebrale (anosognosie) [Contribution to the study
of mental disorders in hemiplegia (anosognosia)]. Revue Neurologique
(Paris),
12,
845-848.
Bahro, M., Silber, E., & Sunderland, T. (1995). How do patients with
Alzheimer's disease cope with their
illness? A
clinical report.
Journal
of the
American
Geriatrics
Society,
43,
41-46.
Binetti, G., Magni, E., Padovani, A., Cappa, S.F., Bianchetti, A., & Trabucchi,
M. (1993). Neuropsychological heterogeneity in mild Alzheimer's disease.
Dementia,
4, 321-6.
Bonnie, R. J. (1996). Research with cognitively impaired subjects. Unfinished
business in the regulation of human research.
Archives
of
General
Psychiatry,
54,
105-111.
Brooks, J. O., Kraemer, H. C, Tanke, E. D., & Yesavage, J. A. (1993). The
methodology of studying decline in Alzheimer's disease. Journal of
the
American
Geriatrics
Society,
41, 623-628.
Cotrell, V. C. (1997). Awareness deficits in Alzheimer's disease: Issues in as-
sessment and intervention.
Journal
of Applied
Gerontology,
16,
71-90.
Critchley, M. (1953). The parietal
lobes.
London: Edward Arnold.
Cummings, J. L., Mega, M., Gray, K., Rosenberg-Thompson, S., Carusi,
D.
A.,
& Gornbein, J. (1994). The neuropsychiatric inventory: Comprehensive as-
sessment of psychopathology in dementia.
Neurology,
44, 2308-2314.
Cummings, J. L., Ross, W., Absher, J., Gornbein, J., & Hadjiaghai, L. (1995).
Depressive symptoms in Alzheimer disease: Assessments and determi-
nants.
AlTheimer
Disease
and
Associated
Disorders,
9,
87-93.
David, A. S. (1990). Insight and psychosis. British Journal of
Psychiatry,
161,
599-602.
DeBettignies, B. H., Mahurin, R. K., & Pirozzolo, F. J. (1990). Insight for im-
pairment in independent living skills in Alzheimer's disease and multi-
infarct dementia. Journal of
Clinical
and Experimental Neuwpsychology,
12,
355-363.
Feher, E. P., Mahurin, R. K., Inbody, S., Crook, T., & Pirozzolo, F. J. (1991).
Anosognosia in Alzheimer's disease.
Neuropsychiatry,
Neuwpsychology,
and
Behavioral
Neuwlogy,
4, 136-146.
Fisher, C. M. (1989). Neurologic fragments n. Remarks on anosognosia, con-
fabulation, memory and other topics and an appendix on self-observations.
Neurology,
39, 127-132.
Foley, J. M.
(1992).
The experience of being demented. In R. H. Binstrock, S. G.
Post, & P. J. Whitehouse (Eds.), Dementia and
aging:
Ethics, values, and
policy
choices
(pp.
30-43). Baltimore: Johns Hopkins University Press.
Folstein, M. F., Folstein, S. E., & McHugh, P. R. (1975). "Mini-Mental State":
A practical method for grading the cognitive status of patients for the clini-
cian.
Journal
of
Psychiatric
Research,
12, 189-198.
Gainotti, G. (1972). Emotional behaviour and hemispheric side of
lesion.
Cortex,
8,
41-55.
Hastie, T. J., & Tibshirani, R. J. (1990). Generalized additive models. London:
Chapman & Hall.
Hughes, C. P., Berg, L., Danziger,
W.
L., Coben, L. A., & Martin, R. L. (1982).
New clinical scale for staging of dementia
British Journal
of
Psychiatry,
140,
566-572.
Katz, S., Ford,
A.
B., Moskovitz, R. W., Jackson, B. A., & Jaffe, M. W. (1963).
Studies of illness in the aged. The index of
ADL:
A standardized measure
of biological and psychological function. Journal of
the American
Medical
Association,
185,
914-919.
by guest on December 21, 2015http://psychsocgerontology.oxfordjournals.org/Downloaded from

P106
ZANETTIETAL.
Kotler-Cope, S., & Camp, C. J. (1995). Anosognosia in Alzheimer disease.
Alzheimer
Diesease
and Associated
Disorders,
9, 52-56.
Lawton,
M.
P.,
& Brody,
E.
M.
(1969).
Assessment of older
people:
Self-maintain-
ing and instrumental activities of daily
living.
The
Gervntologist,
9,
179-186.
Lopez, O. L., Becher, J. T., Sumsak, D., Dew, M. A., & Dekosky, S. T. (1994).
Awareness of cognitive deficits and anosognosia in probable Alzheimer's
disease.
European
Neurology,
34, 277-282.
Mangone, C. A., Hier, D. B., Gorelick, P. B., Ganellen, R. J., Langerberg, P.,
Boarman, R., & Dollear, W. C. (1991). Impaired insight in Alzheimer's
disease.
Journal
of
Geriatric Psychiatry
and
Neurology,
4, 189-193.
McDaniel, K. D., Edland, S. D., Heyman, A., & the CERAD Clinical Investi-
gators. (1995). Relationship between level of insight and severity of demen-
tia in Alzheimer's disease. Alzheimer
Disease
and
Associated
Disorders,
9,
101-104.
McGlynn, S. M., & Schacter, D. L. (1989). Unawareness of deficits in neu-
ropsychological syndromes. Journal of Clinical and Experimental
Neuropsychology,
11, 143-205.
McKhann, G., Drachman, D., Folstein, M., Katzman, R., Price, D., & Stadlan,
E. M. (1984). Report of
the
NINCDS-ADRDA Work Group under the aus-
pices of Department of Health Services task force on Alzheimer's Disease.
Neurology,
34, 939-944.
Michon, A., Deweer, B., Pillon, B., Agid, Y, & Dubois, B. (1994). Relation of
anosognosia to frontal lobe dysfunction in Alzheimer's disease. Journal of
Neurology,
Neurosurgery,
and
Psychiatry,
57, 805-809.
Migliorelli, R., Teson,
A.,
Sabe, L., Petracca, G., Leiguarda, R., & Starkstein, S.
E. (1995). Anosognosia in Alzheimer's disease: A study of associated fac-
tors.
Journal
of
Neuropsychiatry
and
Clinical
Neurosciences,
7, 338-344.
Mullen, R., Howard, R., David, A., & Levy, R. (1996). Insight in Alzheimer's
disease.
International Journal
of
Geriatric
Psychiatry,
11,
645-651.
Neundorfer, M. M. (1997). Awareness of variability in awareness. Alzheimer
Disease
and
Associated
Disorders,
11, 121-122.
Ott, B. R., & Fogel, B. S. (1992). Measurement of depression in dementia: Self
vs clinician rating. International Journal of Geriatric Psychiatry, 7,
899-904.
Ott, B. R., Lafleche, G., Whelihan,
W.
M., Buongiorno, G. W., Albert, M. S., &
Fogel, B. S. (1996). Impaired awareness of deficits in Alzheimer's disease.
Alzheimer Disease
and Associated
Disorders,
10,
68-76.
Prigatano, G., & Schacter, D. (1991). Introduction. In Prigatano and Schacter
(Eds.),
Awareness of deficits after brain injury (pp. 3-16). New York:
Oxford University Press.
Reed, B. R., Jagust, W. J., & Coulter, L. (1993). Anosognosia in Alzheimer's
disease, relationship to depression, cognitive function and cerebral perfu-
sion.
Journal
of
Clinical
and
Experimental
Neuropsychology,
15,
231-244.
Roman, G. C, Tatemichi, T. K., Erknjutti, T., Cummings, J. L., Masdeu, J. C,
Garcia, J. H., Amaducci, L., Orgogozo, J. M., Bran, A., Hofmann, A.,
Moody, D. M.,
O'Brien,
M.
D.,
Yamaguchi, T., Grafmann, J., Drayer, B. P.,
Bennett, D. A., Fisher, M., Orgata, J., Kokmen, E., Bermejio, F,
Wolf,
P.
A., Gorelick, P. B., Bick, K. L., Pajeau, A. K., Bell, M. A., DeCarli, C,
Culebras, A., Korcyn, A. D., Bogousslavsky, J., Hartmann, A., &
Scheinberg, P. (1993). Vascular dementia: Diagnostic criteria for research
studies. Report of NINDS-AIREN International Workgroup. Neurology,
43,
250-260.
Schacter,
D.
(1991). Unawareness of deficit and unawareness of knowledge in pa-
tients with memory disorders. In G. Prigatano & D. Schacter (Eds.),
Aware-
ness
of
deficits after brain injury
(pp.
127-151).
New
York:
Oxford University
Press.
Schacter, D., & Prigatano, G. (1991). Forms of
unawareness.
In G. Prigatano &
D.
Schacter (Eds.),
Awareness
of
deficits
after
brain
injury (pp. 258-262).
New York: Oxford University Press.
Starkstein, S. E.,
Fedoroff,
J. P., Price,
T.
R., Leiguarda, R., & Robinson, R. G.
(1992).
Anosognosia in patients with cerebrovascular lesions.
Stroke,
23,
1446-1453.
Starkstein, S. E., Sabe, L., Chemerinski, E., Jason, L., & Leiguarda, R. (1996).
Two domains of anosognosia in Alzheimer's disease.
Journal
of Neurology,
Neurosurgery,
and
Psychiatry,
61,
485-^490.
Tombaugh, T. N., & Mclntyre, N. J. (1992). The Mini-Mental State
Examination: A comprehensive review.
Journal
of
the American Geriatrics
Society,
40, 922-935.
Vallotti, B., Zanetti, O., Bianchetti, A., & Trabucchi, M. (1997).
L'insight
nel
paziente demente: Riproducibilita di due strumenti di valutazione
[Reproducibility of two scales to rate the insight in demented patients].
Giornale
di
Gerontologia,
45,
341-345.
Vasterling, J. J., Seltzer, B., Foss, J. W., & Vanderbrook, V. (1995).
Unawareness of deficits in Alzheimer's disease. Domain-specific differ-
ences and disease correlates. Neuropsychiatry, Neuropsychology, and
Behavioral
Neurology,
8, 26-32.
Verhey, F. R. J., Ponds, R. W. H. M., Rozendaal, N., & Jolles, J. (1995).
Depression, insight and personality changes in Alzheimer's disease and vas-
cular dementia.
Journal
of
Geriatric Psychiatry
and
Neurology,
8,
23-27.
Verhey,
F.
R. J., Rozentaal,
N.,
Ponds,
R.
W.
H. M.,
&
Jolles,
J. (1995). Dementia,
awareness and depression.
International Journal
of
Geriatric
Psychiatry,
8,
851-856.
Wagner, M. T, Spangenberg, K. B., Bachman, D. L., & O'Connel, P. (1997).
Unawareness of cognitive deficit in Alzheimer disease and related demen-
tias.
Alzheimer
Disease
and
Associated
Disorders,
11,
125-131.
Yesavage,
J.
A.,
Brink,
T.
L.,
Rose,
T.
L., Lum, O., Huang, V,
Adey,
M.,
&
Leirer,
O. (1983). Development and validation of a geriatric depression screening
scale:
Preliminary
report.
Journal
of
Psychiatric
Research,
17,
37-49.
Received February
6,
1998
Accepted
October
27,
1998
by guest on December 21, 2015http://psychsocgerontology.oxfordjournals.org/Downloaded from