”AL?

 

 

UBRARY
Michigan State
University

 

 

This is to certify that the
dissertation entitled

Classifying Neuropsychiatric Symptoms in Patients with

Doctoral

Alzheimer’s Disease

presented by

Saw-Myo Tun

has been accepted towards fulﬁllment
of the requirements for the

degree in Psychology

 

 

Mask

Major Profesgor’s Signature
é'ia/oc

 

Date

MSU is an Aﬁirmative Action/Equal Opportunity Institution

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DAIEDUE

DAIEDUE

DAIEDUE

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:/ClRC/DateDue.indd-p.1

 

CLASSIFYING NEUROPSYCHIATRIC SYMPTOMS IN PATIENTS WITH
ALZHEIMER’S DISEASE

By

Saw-Myo Tun

A DISSERTATION

Submitted to
Michigan State University
in partial fulﬁllment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY
Department of Psychology

2006

ABSTRACT
CLASSIFYING NEUROPSYCHIATRIC SYMPTOMS IN PATIENTS WITH
ALZHEIMER’S DISEASE
By
Saw-Myo Tun

Objective: The aim of the study was to conceptualize neuropsychiatric symptoms in
Alzheimer’s disease patients, as distinct symptom proﬁles with differential disease
outcomes. Five outcomes of interest in the study were caregiver distress, quality of life,
functional impairment, nursing home placement, and survival. Method: Cluster analysis
was used to categorize 122 patients with Alzheimer’s disease, based upon their
neuropsychiatric symptoms, as assessed by the Neuropsychiatric Inventory (NPI). The
presence, severity, and frequency of symptoms were considered. After identiﬁcation of
the subgroups, the predictive validity of the categorization was tested on the following: 1)
group differences in caregiver distress at baseline using AN COVA; 2) group differences
in quality of life and ﬁmctional impairment over a 2-year period using repeated measures
AN OVA; and 3) group differences in time to nursing home placement and time to death
over a 3-year period using Cox proportional hazard models. Results: Based on the
presence of neuropsychiatric symptoms, three subgroups were identiﬁed: Minimally
Symptomatic, Highly Symptomatic, and Predominantly Apathetic. At baseline, the scores
on a caregiver distress measure differed signiﬁcantly between the clusters (p_= 0.00).
Over a 2-year period, the subgroups were predictive of quality of life (p = 0.00).
Similarly, over the same 2-year period, functional impairment was differentially

predicted by the subgroups (p = 0.00). As for time to nursing home placement over a 3-

year period, the results were signiﬁcant (p < 0.05) with the Highly Symptomatic group
showing the highest risk. In addition, the rates of survival were signiﬁcantly predicted by
the subgroups (p < 0.05), with the Minimally Symptomatic group having the lowest risk.
Based on the severity and frequency of neuropsychiatric symptoms, 2-cluster and 4-
cluster solutions were produced. The 4-cluster solution provided a better differentiation
of the symptom proﬁles than the 2-cluster solution. The 4 clusters were: Minimally
Symptomatic, Affective/Apathetic, Predominantly Apathetic, and Highly Symptomatic
with Psychotic Features. Cross-sectionally, these four subgroups differentially predicted
caregiver distress (p = 0.00). Over a 2-year period, the clustering predicted signiﬁcant
differential outcomes in quality of life (9 = 0.00), and in functional impairment (2 <
0.01). Moreover, cluster membership was predictive of nursing home placement (9 < .05)
and survival (2 < 0.01) over a 3-year period. Conclusions: Neuropsychiatric subgroups,
using the cluster analysis method, were able to predict differential outcomes, and identify
those with an increased risk for a worse prognosis. Speciﬁcally, the clustering based on
the presence, severity and ﬁequency of symptoms, were able to predict outcome in
caregiver distress, quality of life, functional impairment, nursing home placement, and
survival. Thus, the results highlighted the importance of detecting and treating

neuropsychiatric symptoms in Alzheimer’s disease patients.

ACKNOWLEDGMENTS

First and foremost, I would like to thank the members of my dissertation committee, Dr.
Bogat, Dr. Hambrick, Dr. Larson, Dr. Murman, Dr. Symonds, and Dr. von Eye, for their
thoughtful input and guidance throughout the researching and writing of my dissertation.
In particular, my deep gratitude goes to my committee co-chairs, Dr. Bogat and Dr.
Murman, for their expertise, skillful mentoring, and unwavering support. I am truly
grateful for a tremendously rewarding learning experience.

I would also like to thank the CHANCE project for making my research possible.

Speciﬁcally, I would like to thank Dr. Murman, Dr. Colenda, and Heidi Long.

iv

TABLE OF CONTENT

LIST OF TABLE ................................................................................ vii
LIST OF FIGURES ............................................................................... viii
INTRODUCTION ............................................................................... 1
Speciﬁc Aims ................................................................... 1
Background and Signiﬁcance ................................................ 2
Rationale ........................................................................ 10
METHOD ........................................................................................ 15
Participants ..................................................................... 1 5
Measures ....................................................................... 16
Procedure ....................................................................... 20
RESULTS ........................................................................................ 22
Cluster Analysis of Presence of Symptoms ................................ 22
Outcomes Based on the Presence of Symptoms ................. 23
Caregiver Distress ........................................... 23
Quality of Life ............................................... 24
Functional Impairment ..................................... 25
Nursing Home Placement..................................... 25
Survival ....................................................... 27
Cluster Analysis of Severity and Frequency of Symptoms .............. 28
Outcomes Based on the Severity and Frequency of Symptoms29
Caregiver Distress ........................................... 29
Quality of Life ................................................ 30
Functional Impairment ...................................... 30
Nursing Home Placement ................................... 31
Survival ........................................................ 32
DISCUSSION .................................................................................... 33
Summary of the Findings on Cluster Analyses ...................................... 33
Summary of the Findings on Caregiver and Patient Outcomes .................. 36
Caregiver Distress ............................................................. 37
Quality of Life .................................................................. 39
Functional Impairment ........................................................ 40
Nursing Home Placement .................................................... 41
Survival ......................................................................... 43

Implications of the Study .............................................................. 44

Limitations of the Study ............................................................... 47

Suggestions for Future Research ..................................................... 48

Conclusions ............................................................................. 49
APPENDIX: Tables and Figures ............................................................. 51
LIST OF RFERENCES ........................................................................ 80

vi

Table 1

Table 2

Table 3

Table 4

Table 5

Table 6

Table 7

Table 8

Table 9

LIST OF TABLES

Cluster Mean Proﬁles for the Presence of Symptoms .................. 51
HUI Mean Scores for the Presence of Symptoms ....................... 52
Dependence Scale Mean Scores for the Presence of Symptoms ...... 53
Pair-wise Compairsons between Clusters ................................. 54
Cluster Mean Proﬁles for the 4-cluster Solution based on the severity
and Frequency of Symptoms ............................................... 55
HUI Mean Scores for the Severity and Frequency of Symptoms. . ...56
Dependence Scale Mean Scores for the Severity and Frequency of
Symptoms .................................................................... 57
Clusters based on the Severity and Frequency of Symptoms as
Predictors of Nursing Home Placement over a 3-year

Period ......................................................................... 58
Clusters based on the Severity and Frequency of Symptoms as Predictors

of Mortality over a 3-year Period ......................................... 59

vii

Figure 1

Figure 2

Figure 3

Figure 4

Figure 5

Figure 6

Figure 7

Figure 8

Figure 9

Figure 10

LIST OF FIGURES

Cluster Means of Neuropsychiatric Symptoms based on the Presence
of Symptoms ................................................................ 60
Quality of Life over a 2—Year Period based on the Presence of
Symptoms ................................................................... 61
Emotional Impairment over a 2-Year Period based on the Presence of
Symptoms ................................................................... 62
Estimated Probability of Nursing Home Placement based on the
Presence of Symptoms .................................................... 63
Estimated Probability of Death based on the Presence of

Symptoms ................................................................... 64
4-Cluster Means of Neuropsychiatric Symptoms based on the
Severity and Frequency of Symptoms .................................. 65
Quality of Life over a 2-Year Period based on the Severity and
Frequency of Symptoms .................................................. 66
Functional Impairment over a 2-Year Period based on the Severity
and Frequency of Symptoms ............................................. 67
Estimated Probability of Nursing Home Placement based on the
Severity and Frequency of Symptoms ................................. 68
Estimated Probability of Death based on the Severity and

Frequency of Symptoms .................................................

viii

Classifying Neuropsychiatric Symptoms in Patients with Alzheimer’s Disease

Speciﬁc Aims

Neuropsychiatric symptoms, such as apathy, agitation, hallucination, delusion,
and depression, have long been acknowledged as the secondary resultants of
neuropathological changes associated with dementia. Fmtherrnore, it has been recognized
that the presence of such neuropsychiatric disturbances contributes to a worse prognosis,
including a more rapid functional and cognitive decline, and an increase in caregiver
distress (Ballard et a1., 2000; Cummings, 2003a; Norton, Malloy, & Salloway, 2001;
Teri, 1997). Nevertheless, as the importance of neuropsychiatric symptoms in dementia
becomes more apparent, it seems to highlight the limitations in the present
conceptualization of the symptoms. Such limitations in conceptualization can have
detrimental effects in that they can potentially hamper our understanding of the
pathophysiology of the symptoms as well as their resulting clinical correlates. Thus, in
the current study, the primary purpose was to determine whether neuropsychiatric
disturbances in dementia patients could be better conceptualized as distinct symptom
proﬁles that predict differential disease courses. In doing so, the study attempted to
overcome some of the methodological limitations of previous studies by using statistical
methods that allowed for grouping of dementia patients with psychiatric symptoms into
meaningful clusters. In addition, we examined the presence, severity, and frequency of
symptoms. A critical component of the study was that predictive validity of membership
of such groupings was tested on four patient outcomes and one caregiver outcome. Thus,

the research has two speciﬁc aims. (1) To identify groups of dementia patients with

similar proﬁles of neuropsychiatric symptoms based upon caregiver’s responses on the
Neuropsychiatric Inventory (N PI). The group membership was based on the presence,
and the severity and frequency of the symptoms. (2) To test the ability of group
membership to predict caregiver and patient outcomes. Possible covariates such as
severity of functional and cognitive impairments, severity of Parkinsonism, severity of
other medical problems, and age were adjusted at baseline.
Background and Siggiﬁcance

As one ages, the risk of dementia grows at an incremental rate in such a way that
by the time an individual reaches the age of 80, he or she has a 1 in 6 chance of
developing dementia (J orm, 1991). Of the signiﬁcant number of older adults who develop
dementia, a sizable portion also goes on to develop neuropsychiatric symptoms in
addition to the cognitive impairments. According to recent estimates, neuropsychiatric
disturbances are believed to be present in 74-75% of the individuals with dementia
(Lyketsos et al., 2002; Weiner et al., 2005). Moreover, disturbances of this nature
substantially increase the likelihood of a worse outcome in a number of domains,
including a decrease in activities of daily living, an increase in cost of care, and more
likelihood of nursing home placement (Murman et al., 2002; Gilley et al., 2004; Norton,
Malloy, and Salloway, 2001; Kopetz et al., 2000; Lyketsos et al., 1997). Thus, a
combination of a high prevalence rate and the debilitating nature of the psychological
disturbances have propelled researchers to search for an underlying pathophysiology of
the symptoms and their clinical correlates. It has been suggested that such undertakings
could pave the way for the development of more effective interventions that could lessen

the hardship on the patients and their caregivers (Aalten et al., 2003; Frisoni et al., 1999).

In recent years, however, concerns have been raised regarding the methodologies
employed in the study of neuropsychiatric symptoms. Traditionally, the study of
neuropsychiatric disturbances entailed focusing exclusively on one symptom, such as
depression, and examining its impact on the patient (e. g. Lyketsos et al., 1997; Payne et
al., 2002). However, it has been postulated that such an approach disregards the high co-
occurrence of neuropsychiatric symptoms in dementia patients (F risoni et al., 1999;
Lyketsos et al., 2001b). For example, a study by Lyketsos and colleagues (2002) found
55% of the dementia patients endorsed 2 or more neuropsychiatric disturbances. Hence,
given a particularly high co—morbidity of symptoms, the exclusion approach of previous
studies has a limited capacity for ﬁnding consistent patterns in outcome research and for
offering clinically meaningful causal relationships (Aalten et al., 2003; Frisoni et al.,
1999: Lyketsos et al., 2001b; Holthoff et al., 2005).

In order to better reﬂect the clinical picture of neuropsychiatric disturbances in
dementia, two alternative strategies have been employed in recent studies. In one such
approach, the method of factor analysis is used to create meaningful groups of symptoms
that are most likely to co-occur as assessed by a scale measuring neuropsychiatric
symptoms. For example, this approach has been used with the BEHAVE-AD, the
Neuropsychiatry Inventory (N PI), and the Present Behavioral Examination. Despite
variations in symptoms included in each factor, most studies found 3 factors, with mood,
psychosis, and hyperactivity/frontal factors being the most consistent ﬁndings (Aalten et
al., 2003; Frisoni et al., 1999; Hope et al., 1997; McShane, 2001; Schrenizer et al., 2005).
Nevertheless, although this method addresses the problem of focusing too narrowly on a

particular symptom, factor analysis does not place patients into subgroups. Instead, it

places patients on a continuum of a continuous factor score. If subgroups are to be
derived, then an arbitrary cutoff score must be deﬁned on each factor score.
Furthermore, since the method of factor analysis allows for individual patients to be
included in more than one factor, it does not create well-deﬁned groups of individuals
with distinct symptom proﬁles (Lyketsos et al., 2001b). Thus, it is not an ideal approach
for examining the differential outcomes of individuals based upon their neuropsychiatric
symptom proﬁles.

Another strategy that has been employed in recent studies of neuropsychiatric
disturbances is to create groups of individuals, with differing symptom proﬁles, using a
latent class analysis method (e. g. Lyketsos et al., 2001b; Moran et al., 2004). Although
solutions derived from latent class analysis correspond to that of factor analysis, the
purpose of this approach is to identify meaningﬁil groups of individuals rather than
groups of variables (von Eye & Bergman, 2003). Hence, the latent class analysis
approach avoids the problem of the same individuals being included in multiple
subgroups. This particular feature makes latent class analysis more suitable for use in
neuropsychiatric outcome studies.

Based on the results of their latent class analysis study (2001b) and that of an
epidemiological study (2000), Lyketsos and colleagues (2001a) derived a classiﬁcation
system for categorizing neuropsychiatric symptoms in Alzheimer’s disease patients.
Although their effort toward the development of an empirically-driven classiﬁcation
system should be applauded, their methodologies did not account for the severity of
neuropsychiatric symptoms. As aptly observed by Meyers (2001), the failure to account

for severity and frequency of symptoms is a serious ﬂaw in the study of neuropsychiatric

disturbances in that a patient who occasionally experiences a mild symptom of agitation
is deemed to have a similar degree of pathology as that of a patient endorsing persistent
and severe agitation. Along the same line, such disparate individuals will be expected to
have comparable clinical outcomes. However, such a line of reasoning is intrinsically
problematic, and outcome research based on such a conceptualization will have limited
applicability (Meyers, 2001).

The present study addressed some of the limitations of previous studies in the
ﬁeld in the following ways: 1) Psychiatric symptoms of neurological origin in dementia
patients were conceptualized as clusters of symptoms that co-occur together, rather than
as disparate symptoms. The presence of symptoms as well as the severity and frequency
of disturbances were explored. 2) To maximize the predictive validity of groups of
individuals on outcomes, individual patients were placed in non-overlapping categories
rather than in multiple categories.

Clustering of Symptoms

To accomplish the above-stated aims, it was proposed that the method of cluster
analysis is a suitable approach for such undertaking. First, if one’s goal was to have a
classiﬁcation system that could predict outcomes in individual patients, then it would be
preferable to use a grouping method that is person-oriented in nature than a variable-
centered approach (von Eye & Bergman, 2003). After the decision was made to use a
person-oriented method, the investigator was left with an array of choices in grouping
methods, including cluster analysis and latent class analysis. Of those possible

classiﬁcation methods, the method of cluster analysis allowed for the most ﬂexibility in

using both continuous and categorical variables, and also in correcting for the high
intercorrelations among some of the NPI domains.

By employing this ﬂexible approach, the study was able to consider both the
presence of symptoms, which is categorical in nature, and the severity and the frequency
of symptoms, which are continuous. Also, cluster analysis has been suggested as a
suitable method for looking at differences in proﬁles between individuals while avoiding
the loss of critical information (von Eye, Mun & Indurkhya, 2004). On the other hand,
von Eye and co-authors (2004) note that the use of cluster analysis sometimes raises the
concern of arbitrariness. To address such a concern, the authors outlined a series of
decision making steps to ensure the choice of cluster analysis method is based on a well-
justiﬁed position rather than a blind application of the method.

Here, a few of the critical steps in the decision making process are reported. First,
there is a decision concerning whether there are disjointed or overlapping clusters. Given
that one of the main objectives of this research is to identify non-overlapping symptom
proﬁles, a disjoint cluster approach was a natural choice. A second critical decision
concerns the hierarchical versus non-hierarchical structures of the clusters. In this regard,
an assumption of the study, that the clustering of patients into smaller groups will have an
interpretable meaning, led to the choice of a hierarchical method. The third, and a
fundamental decision of the study, relates to the choice of base measures. Because the
study was interested in examining the presence of neuropsychiatric symptoms, which is
categorical, as well as their severity and frequency, which is continuous, the use of
Pearson’s correlation was the most suitable, as it allowed for an examination of both

categorical and continuous data.

A decision was made to analyze the presence of symptoms, and the severity of
symptoms in parallel, as it allowed the study to accomplish two aims. First, in previous
studies, the classiﬁcation was based only on the presence of symptoms. Thus, by
including the presence of symptoms in the study, it made it feasible to compare the
ﬁndings of the present research with those of prior studies. Second, as Meyers (2001)
noted, the severity and frequency of symptoms are potentially important considerations in
neuropsychiatric research. However, to our knowledge, there has not been any study to
consider this aspect. Hence, we incorporated the severity and frequency of symptoms into
the study on an exploratory basis. Findings from both approaches are presented. A
precedent for the use of a similar methodology has been set in previous studies, including
a study on substance use and adjustment in adolescents (Tubman et al., 1991).

Predictive Validity of Clusters

In the second segment of the study, the clinical utility of the classiﬁcation was
tested. When classifying individuals into subgroups, the usefulness of the procedure
depends not only on the identiﬁcation of distinct groups, but also on establishing the
predictive validity of these groups (von Eye & Bergman, 2003). Therefore, this research
examined the differential impacts of neuropsychiatric symptom proﬁles on various
domains of clinical interest.

Caregiver burden. The ﬁrst domain was caregiver burden. Greater numbers of
older adults continue to live in the community as they age, and when they stay in their
own homes, they often rely on their family members to be the primary caregivers
(Gonzalez-Salvador et al., 1999). As such, understanding the role of the caregivers and

the challenges they face becomes essential block in order to have a complete clinical

picture of the dementia patient. In the case of neuropsychiatric disturbances, evidence is
accumulating that caregivers do indeed experience a greater degree of burden and a
decreased quality of life in the presence of these symptoms (Ballard et al., 2000;
Gonzalez-Salvador et al., 1999; Lyketsos et al., 1997; Shin et al., 2005; Teri, 1997). In
fact, the distress associated with caring for AD patients has been linked speciﬁcally to
neuropsychiatric disturbances, but not to cognitive or functional impairments (Craig et
al., 2005). However, the results of the previous studies were based on a traditional
approach of looking at a single symptom and examining its impact. Given the highly co-
morbid nature of neuropsychiatric symptoms, there is a potential that the ﬁndings are still
an underestimation of the actual extent of caregiver burden. Therefore, this research
investigated how co-occurring neuropsychiatric disturbances contributed to caregiver
distress.

Functional impairment. The second domain we considered is functional
impairment of the dementia patient. Although the role of cognitive deterioration on
functional impairment has been well-established, our knowledge of the contribution of
neuropsychiatric disturbances to functional impairment is limited. Here again, this
limitation can partly be attributed to research that examines the effects of symptoms in
isolation. Based on the available data, however, it is known that neuropsychiatric
disturbances have a negative impact on activities of daily living (Hargave, Reed, &
Mungas, 2000; Norton, Malloy, & Salloway, 2001; Tekin et al., 2001; Weiner et al.,
2005). Therefore, in the present study, we sought to gain a fuller understanding of the
impact of neuropsychiatric symptoms on functional impairment by examining it

longitudinally over a 2-year period of time. It is recognized that the longitudinal design of

the study may raise the question of the stability of neuropsychiatric disturbances over
time. Nevertheless, several ﬁndings are in agreement that neuropsychiatric symptoms,
once present, are highly likely to persist over the course of the dementia (Devanand et al.,
1997; Levy et al., 1996; Steinberg et al., 2003, 2004).

Quality of life. The third domain we examined was the quality of life in dementia
patients. As intuition might lead one to expect, neuropsychiatric symptoms adversely
impact quality of life in dementia patients. For instance, in a study by Gonzalez-
Salavador and colleagues (2000) that looked at quality of life in dementia patients
residing in a long-term care setting, depression was associated with a lower quality of
life. An interesting ﬁnding, however, is that a follow-up study of the same population by
Lyketsos and associates (2003) found that depression did not ﬁirther contribute to a
decline in quality of life during the 2-year follow-up period. The ﬁndings suggested that
when depression is detected, it was possible to provide necessary interventions to prevent
further deterioration in quality of life. Similar to ﬁndings by Lyketsos and co-workers
(2005), a recent study also reported quality of life to be stable over a l-year period in AD
patients (Selwood, Thorgrimsen, & Orrell, 2005). However, this study reported anxiety,
in addition to depression, to be factors in predicting lower quality of life. Since these two
studies were the ﬁrst of their kind to consider the effects of depression and anxiety on
change in quality of life over time, further research is needed to test these ﬁndings, and to
incorporate the possible impact of co-morbid neuropsychiatric symptoms on quality of
life. In the current study, the effects of neuropsychiatric symptoms on quality of life in

dementia patients were considered over a 2-year period.

Nursing home placement and survival rate. As a purely longitudinal component
of the study, we also examined the rate of nursing home placement and survival rate. It is
recognized in the ﬁeld that these two aspects are important considerations in the lives of
patients with dementia for a number of reasons. First, the decision to place a family
member into a nursing home setting can have a tremendous psychological impact on both
the caregiver and the patient (Thomas et al., 2004). Second, the decision to
institutionalize a dementia patient is also a matter of ﬁnancial consideration. As estimated
by Butler (1995), a delay in institutionalization by one month for every person over the
age of 65 in the US could save $60 billion a year in the cost of care. Third, availability of
information on approximate survival rates could aid end of life decisions for the family
and the patient. Therefore, the study considered the differential impact of
neuropsychiatric symptoms on the rates of nursing home placement and survival.

It should be noted that in considering the rate of nursing home placement, it is
recognized that differences in caregiver characteristics, such as martial status, could
potentially inﬂuence the outcome. For instance, it is conceivable that an AD patient with
a spouse as a primary caregiver would have a decreased likelihood of being placed in a
nursing home compared to a patient with an adult child as a caregiver. Thus, the present
study controlled for marital status in examining the rate of nursing home placement.
Rationale

The primary aim of the current study was to reﬁne the conceptualization of
neuropsychiatric disturbances in dementia; however, the results of the study may be
applicable to both scientiﬁc research and clinical applications. First and foremost, it is

hoped that by conducting this project, we would be able to arrive at a more satisfactory

10

conceptualization and categorization of neuropsychiatric disturbances that can predict
patient and caregiver outcomes. Such a conceptualization would also aid in conducting
further research at the neurobiologic level. For example, a natural step that could
potentially follow the present study is to apply the classiﬁcation in examining whether
genetic polymorphism associated with neuropsychiatric symptoms in patients with
dementia are associated with the categorization of patients based upon neuropsychiatric
disturbances. Moreover, identiﬁcation of subgroups of dementia patients with distinct
neur0psychiatric symptom proﬁles may contribute to detection of subtypes of dementia
with possibly different neurobiologic underpinnings (Aalten etal., 2003; Lyketsos et al.,
2001b). Overall, if the categorization of symptoms shows predictive validity, it may
provide evidence for applying the subgroups to neurobiologic research.

Potentially, the study also has more immediate clinical implications. As noted by
previous studies, the current Diagnostic and Statistical Manual of Mental Disorders —- I V
(DSM-IV) does not contain speciﬁc criteria for denoting neuropsychiatric symptoms
associated with dementia (Aalten et al., 2003; Lyketsos et al., 2001b). However, a new
edition of the DSM is expected to redress the situation and an empirically-driven
classiﬁcation system can help contribute to that effort.

Another potential clinical application is that if the results of the study show
detrimental impact of neuropsychiatric disturbances in a number of domains, it may
highlight the importance of detection and intervention to healthcare professionals.
Furthermore, if NPI proves to be useﬁil in classiﬁcation of the symptoms as well as
providing external validity, it would lend further support for its use in the clinical setting.

By completion of the 15-minute NPI battery, the healthcare provider would be able to

11

identify patients at risk for poor outcomes based upon their neuropsychiatric proﬁle. This
would allow healthcare providers to target these patients for psychological and
pharmacological interventions that may improve their neuropsychiatric symptoms. It is
anticipated that better management of neuropsychiatric symptoms would result in an
improvement in the patient’s functional abilities and quality of life and a decrease in
caregiver’s burden. Further studies could examine the impact of neuropsychiatric
symptom proﬁles on costs of care in our current healthcare system and measure the
economic impact of focused interventions to improve neuropsychiatric disturbances in
this population of patients.

The present study was a secondary data analysis study with cross—sectional and
longitudinal components. Speciﬁcally, data from the initial time point were used to
classify AD patients into groups based upon the presence, severity, and frequency of
neuropsychiatric symptoms. After the classiﬁcation, the groupings were used to predict
the level of caregiver distress cross-sectionally. Also, the groups were used to predict the
functional impairment and quality of life longitudinally. In addition, the longitudinal
design of the study enabled the investigator to examine the between-group differences in
time to nursing home placement and in survival over the 3- year period.

Since the present study was the ﬁrst study to categorize neuropsychiatric
symptoms using cluster analysis, and to incorporate severity and frequency of symptoms,
there were no a priori hypotheses regarding the outcome of the categorization. Similarly,
the current study was the ﬁrst to consider the impact of neuropsychiatric symptoms based

on clusters of symptoms. In addition, given that the second segment of predicting

12

caregiver and patient outcomes rested on the results of categorization, there were no
speciﬁc a priori hypotheses regarding subgroup differences.

However, more general hypotheses can be made. (1) It is predicted that there will
be well-defmed patterns of neuropsychiatric symptom proﬁles among dementia patients,
as tested by cluster analysis. Previously, grouping based on latent class analysis, which is
the closest approximation to cluster analysis, had identiﬁed three groups: absent/mono-
neuropsychiatric symptom group, predominantly affective, and psychotic syndrome.
Similar groupings are expected. (2) Based on the class membership as identiﬁed in part 1,
groups will differ in terms of the level of caregiver distress croo-sectionally as tested by
analysis of covariance. Previous reports of speciﬁc symptoms associated with caregiver
distress have varied. However, it is expected that more symptomatic groups, such as
predominantly affective or psychotic syndrome group, will lead to greater caregiver
distress than non- or mono-symptomatic group. (3) Class membership will predict the
rate of decline on the quality of life scale over a 2-year period as tested by repeated
measures analysis of variance. Based on available literature, subgroups with affective
symptoms will be expected to have lower quality of life than the non- or low
symptomatic groups. (4) Class membership will predict the rate of functional decline
over a 2-year period as tested by repeated measures analysis of variance. Previous studies
indicated that apathy and psychotic symptoms predict functional impairments. Therefore,
we identify subgroups with similar characteristics, those subgroups will be expected to
have greater functional decline. (5) Class membership will predict the differential time to
nursing home placement in dementia patients over a 3-year period as tested by survival

analysis. Based on previous literature, groups with high symptoms are expected to be at

13

greater risk for nursing home placement. (6) Class membership will predict the
differential time to death in dementia patients over a 3-year as tested by survival analysis.
Limited information is available on the relationship between neuropsychiatric symptoms
and survival. However, one study has shown depression to be associated with greater risk
of death. Therefore, subgroups with affective symptoms are expected to have lower

chances of survival.

14

Methods
Participants

The participants were drawn from an ongoing study, the Cost of Health in
Alzheimer’s disease Relative to Gained Effectiveness (CHARGE), at Michigan State
University (Murman et al., 2002; Murman et al., 2003). Participants who met the study
criteria were recruited by mail from six neurology practices and three geriatric medicine
practices in Michigan. The inclusion criteria for the study were a clinical diagnosis of
probable Alzheimer’s disease using the Alzheimer’s Disease and Related Disorders
Association-National Institute of Neurological Communicative Disorder and Stroke
(N INCDS-ADRDA) criteria, and the availability of an informant who was in direct
contact with the patient at least once a week. Given that different types of dementia may
have distinctive neuropsychiatric proﬁles, only patients with the AD diagnosis were
included to maintain the homogeneity of the sample.

Of the 692 eligible participants, 128 patients (18%) responded and were
subsequently included in the study. However, at the stage of data analysis, 6 participants
with MMSE score of 0 were excluded from the study, as this score indicated very severe
dementia, and the patient may have became mute by this stage of dementia. Therefore,
the informant may not be able to detect symptoms such as hallucination and delusions,
and may subsequently underreport these symptoms. Thus, the ﬁnal sample used for the
study was 122.

Ninety-seven percent of the participants in the study were Caucasian. The mean

age of the participants was 76.2 years (S_D = 9.0), and the mean level of education was

15

12.9 years (S_D = 3.1). F iffy-ﬁve percent of the participants were female. In many cases,
spouses served as the informants for the study (60%). In the remainder of the cases,
daughters (30%), other relatives (7%), and friends of the patient (3%) were the
informants. Twenty-two percent of the sample resided in a long-term care setting at
baseline. Of these, 41% lived in an assisted living setting, 30% in a foster care home, and
18% in a nursing home. The type of long-term care for 3 (11%) of the participants was
not known. Also of note, 2 participants (2%) dropped out from the study after the ﬁrst
wave of data collection, and their data were not used for the longitudinal portion of the
study.
Measures

The study used 5 questionnaires, 3 yearly interviews, a cognitive measure and a
neurological rating scale. Of those, the Neuropsychiatric Inventory, the Cumulative
Illness Rating Scale, the Mini-Mental Status Exam, and the Modiﬁed Colombia
University Parkinson’s Disease Rating Scale were given at the initial time point. The
Dependence Scale and the Health Utilities Index were given at the initial time point, and
at each of the yearly follow-up. The interviews were given at each yearly follow-up.

Neuropsychiatric Inventory (NPI). NPI was used to assess the AD related
behavior symptoms that were present in the past month. This structured interview was
developed by Cummings and colleagues (1994) to examine the presence, severity, and
frequency of ten commonly observed neuropsychiatric symptoms in dementia patients.
The ten domains included in the interview are delusion, hallucination, agitation,
depression, anxiety, euphoria, apathy, disinhibition, irritability, and aberrant motor

behavior. The NPI uses an informant to report on symptoms that they have observed in

16

the patient in a speciﬁc time period. Furthermore, the design of the measure focuses only
on those symptoms that are present, thus allowing for an efﬁcient use of the examiner’s
time without sacriﬁcing information. For example, one of the questions used to assess
hallucinations asked, “Does the patient talk to people who are not there?” If a behavior is
present, further inquires are made as to the behavior severity and frequency.

In a report by Cummings (1997), the NPI was noted to have a strong inter-rater
reliability with r ranging from 0.94 to 1.0, and a test-retest reliability ranging from 0.79 to
0.86. Moreover, it was reported that NPI has a good construct and content validity, and
therefore is commonly used in dementia research (Lyketsos et al., 2001b; Schneider,
2001; Holthoff et al., 2005). In the current study, the score indicating the presence of a
neuropsychiatric symptom as well as the domain score that incorporates the severity and
frequency of the symptom were used. In the presence of a neuropsychiatric disturbance,
the domain score can range from 1 to 12. For example, if the informant endorses a
symptom, such as depression, he/she will rate the severity of that symptom, which ranges
from 1 to 4, and rate the frequency of the symptoms, which can range from 1 to 3. The
domain score is the severity of symptom x the frequency of symptom.

Screen for Caregiver Burden. This is a 25-item measure constructed to assess the
objective prevalence of the patient’s distressing behaviors and the level of subjective
distress the caregiver experiences in response to the behaviors (Vitaliano et al., 1991).
For example, it includes items, such as, “My spouse doesn’t cooperate with the rest of the
family”, and “I feel so alone as if I have the world on my shoulder.” The informant
reports as to whether the statements are true, and rates the distress associated with each

Statement.

17

For this measure, Vitaliano and colleagues (1991) reported an alpha coefﬁcient of
0.89 and a test-retest reliability of 0.70 for a 15-month interval. In the study, the total
score on the measure was used to evaluate the distress experienced by the caregiver.

Dependence Scale. This scale is designed to measure the functional decline
associated with progression of dementia, as reported by an informant who is familiar with
the patient’s daily activities (Stern et al., 1994). The measure contains 13 items, and the
items are converted to determine the level of dependency ranging from 0 to 5, with 0
being the lowest level of dependency. Examples of items include, “Does the patient have
to be fed?” and “Does the patient need to be escorted when outside?” Stern and
colleagues (1994) reported the scale to have an intraclass correlation of 0.90, indicating
strong interrrater reliability, and a good validity with other functional impairment scales.
This measure was included to examine the patient’s ﬁmctional impairment.

Health Utilities Index (HUI). The HUI is a reliable, general health-related quality
of life measure that can be administered in a survey format to informants. It consists of
six domains applicable to patients with AD that are combined in a multiplicative ﬁrnction
to derive a single, preference-weighted quality of life score on a zero to one continuous
scale (F eeny et al., 1995). For example, the informant is asked to rate on patient’s
mobility, attribute sensation, emotion, fertility, cognition, self-care, and pain. The HUI
correlates signiﬁcantly with important signs and symptoms of AD, including severity of
the patient’s cognitive impairment, neuropsychiatric symptoms, functional impairment,
and Parkinsonism (Murman et al., 2002). The scale provides an assessment of health
status, and, in addition, it provides a utility-based estimate of the value of health state

based upon population-based studies using the standard gamble technique.

18

F ollow—up Interviews. At each yearly follow-up interview, the informant was
asked whether the patient had been placed in a nursing home facility, which included
traditional nursing home setting, an assisted living facility, and an adult foster care home,
or had died. The interviewer recorded the date of such event, which was used to calculate
the time to events over the follow-up period.

Measures of confounding variables:

Mini-Mental Status Exam (MMSE). The MMSE is a commonly used measure for
determining the level of cognitive functioning in an individual. It has been reported that
MMSE is appropriate for use in examining the progression of cognitive decline and is
able to capture a wide range of cognitive impairment in patients with AD (Folstein,
Folstein, & McHugh, 1975). The score ranges from 0 to 30, with scores below 24
indicating some cognitive deﬁcits. In the present study, score on the MMSE was used to
adjust for cognitive impairments at baseline.

Cumulative Illness Rating Scale (CIRS). In order to control for the potentially
confounding impacts of co-morbid medical conditions on outcome measures, the CIRS
was used to evaluate the presence and the severity of co-morbid medical conditions in the
patient. This measure calculates the severity and the presence of co—morbid medical
conditions based upon the functioning of the 13 organ systems (Linn, Linn, & Gurel,
1968), and has been used successfully to predict hospitalization and death (Miller et al.,
1992). Examples of the organ systems assess include cardiac and respiratory functioning.
The score ranges from 14 to 70. The higher score on the scale indicates a more severe co-

morbid medical condition.

19

Modiﬁed Colombia University Pmson’s Disease Rating Scale. This scale is
used to quantify the presence of signs of Parkinsonism that are commonly observed in
patients with Parkinson’s disease and AD. Examples of symptoms assessed include
resting tremor, rigidity, bradykinesia, and posture abnormalities. Each symptom is rated
on a four-point scale of absent, slight, mild-moderate, and severe. The score ranges from
0 to 28. The reliability of this scale is respectable, as has been previously reported
(Richards et al., 1991). Severity of Parkinsonism predicts rate of functional decline and is
correlated with some neuropsychiatric symptoms (e.g. psychosis). Thus, we used the total
score on this scale to adjust for group differences in Parkinsonism at baseline that may
inﬂuence outcomes.

Procedure

After obtaining informed consent, trained examiners interviewed the patient and
his or her informant separately for approximately 45 minutes. During the interview with
the patient, he or she was asked to undergo a neurological exam and an assessment of
neuropsychological functioning. The informant, meanwhile, provided the examiner with
information on patient’s symptoms, co-morbid medical conditions, level of functional
impairment, quality of life, cost of care for the patient, and the level of distress
experienced as the caregiver. A year after the initial interview, the informant was
contacted yearly for 3 years by mail to complete surveys and by phone for an interview.
In these follow-up contacts with the informant, he or she reported on the patient’s level of
functioning, quality of life, changes in symptoms, and signiﬁcant events in the past year

such as hospitalization, institution or death. The participants were compensated with a

20

$25.00 check after the initial interview, and with a $15.00 check after each of the yearly

follow-up interviews.

21

Results
Cluster Analysis

To account for both the presence and the severity of neuropsychiatric symptoms,
2 sets of cluster analyses were conducted. For both of the analyses, weighted scores were
used to correct for high intercorrelations observed between some of the NPI domains. In
the ﬁrst set of analyses, we considered the presence of symptoms on the NPI. Given that
the presence of symptoms is a categorical variable, we used the Ward’s minimum sum of
squares method with Pearson’s correlation as a base measure. In the second set of
analyses, the severity x frequency of symptom score on the NPI was considered. Here
again, Ward’s method with Pearson’s correlation as a base measure was used. The use of
Pearson’s correlation as a base measure in both sets of analyses allowed for examination
of overall pattern of clustering, such as curvature similarities (von Eye, Mun, &
Indurkhya, 2004). Since Ward’s method, like other types of hierarchical agglomerative
methods, is sensitive to outliers, a non-exhaustive approach was used (Everitt et al.,
2001)

At the conclusion of the two analyses, it is possible that groupings based upon
presence of symptoms and groupings based upon severity x frequency of symptoms
would yield different symptom proﬁles that have unique outcome predictabilities worth
examining. Therefore, the study conducted two parallel analyses on outcome measures. A
precedent for such a methodological approach has been set in previous studies (Tubman
etal., 1991). The SPSS 13.0 (2005) and the SYSTAT 10.0 (2002) statistical software
packages were used to conduct the descriptive analyses, correlations, and cluster

analyses.

22

Presence of Neuropsychiatric Symptoms

Based on the presence of neuropsychiatric symptoms, cluster analysis suggested a
3-cluster solution. Cluster mean proﬁles are presented in Table 1. Cluster 1 (_N_ = 38) had
the lowest mean score on the NPI, and thus was deemed to be Minimally Symptomatic.
Cluster 2 (_N = 53) had the highest mean score on the NPI, and the group as a whole
appeared to be Highly Symptomatic. Cluster 3 (_N = 31) had a moderate degree of
symptoms, and an examination of the distribution of scores indicated that the group was
Predominantly Apathetic. (See Table 1, Figure 1).

Hence, for the second segment of the study, the predictive validity of these three
subgroups was examined. Severity of ﬁmctional and cognitive impairments, severity of
Parkinsonism, co-morbid medical conditions, martial status and/or age were adjusted at
base line. The decision to adjust for a covariate was based on previous reports in the
literature of a relationship with the outcome, and on examination of cluster differences
for a particular variable. For instance, if a particular variable was previously found to be a
signiﬁcant predictor of the outcome variable of interest, we examined whether there were
signiﬁcant differences between the clusters, and if so, the variable was included in the
model.

Caregiver Distress. At baseline, caregiver distress was assessed using the Screen
for Caregiver Burden as reported by the informants. On this scale, the Minimally
Symptomatic group had a total mean score of 12.79 (SD = 11.02), the Highly
Symptomatic group had a total mean score of 23.51 (E = 11.55), and the Predominantly

Apathetic group had a total mean score of 17.10 (S_D = 10.27).

23

Using ANCOVA, the signiﬁcance of group differences in distress scores was
tested. Initially, in the overall models, the covariates of co-morbid medical conditions, the
presence of Parkinsonism, age, and the severity of cognitive and ﬁmctional impairments
were included. However, only functional impairment as a covariate was a signiﬁcant
predictor. Thus, in the ﬁnal model, only ﬁmctional impairment was included as a
covariate. After controlling for the covariate, functional impairment, the three subgroups
differed signiﬁcantly on the total mean score for caregiver distress (E = 5.93, p = 0.00;
9% explained variance).

Post-hoe analyses indicated that the Minimally Symptomatic group and the
Highly Symptomatic group differed (p= 0.00), in such a way that the caregivers of the
Minimally Symptomatic group had lower distress than the Highly Symptomatic group.
Furthermore, the Highly Symptomatic group and the Predominantly Apathetic group
differed signiﬁcantly (p < .05) in such a way that the caregivers of the Predominantly
Apathetic group had a lower score on the caregiver distress scale.

Quality of Life. Over a 2-year period, the participants’ quality of life was assessed
using the HUI. Means and standard deviations of the HUI scores for each of the
subgroups are presented in Table 2.

To test for group differences in quality of life over the 3 time points (Baseline,
Year 1, Year 2), repeated measures ANOVA was performed. Possible covariates (the
presence of Parkinsonism, cognitive impairments, and ﬁmctional impairments) were
entered as well. The results indicated that the main effect of neuropsychiatric subgroups
was signiﬁcant (F (2, 94) = 17.05, p = 0.00, 26.6% of the variance explained).

Additionally, the linear Time effect was signiﬁcant (E (2, 94) = 6.66, p < .05, 6.6%

24

variance explained). However, the interaction term between Time and neuropsychiatric
subgroups was not signiﬁcant (see Figure 2).

Post hoc Tukey tests indicated that mean scores on HUI differed signiﬁcantly
between the Minimally Symptomatic group and the Highly Symptomatic group (p =
0.00), with the Highly Symptomatic group having a lower quality of life. Additionally,
the Highly Symptomatic group and the Predominantly Apathetic group showed
signiﬁcant differences in mean scores (p <.05), with the Predominantly Apathetic group
having a higher quality of life.

Emotional Impairment. The degree of functional impairment was assessed in the
sample over a 2-year period, using the Dependence Scale. Means and standard deviations
of the scores on this scale are presented in Table 3.

Repeated measures ANOVAs were performed to test the cluster differences in
functional impairment at three time points (Baseline, Year 1, Year 2). The covariates, the
presence of Parkinsonism, and cognitive impairments, were included. The results
indicated that the main effect of neuropsychiatric subgroups, was signiﬁcant (F (2, 93) =
9.22, p = 0.00, 16.4% explained variance). However, linear and quadratic Time effects
and Group x Time were not signiﬁcant (see Figure 3).

To further differentiate the ﬁndings, post hoc Tukey tests were performed. The
ﬁndings showed signiﬁcant mean differences between the Minimally Symptomatic group
and the Highly Symptomatic group (p < .05), with the Minimally Symptomatic group
having a lesser degree of functional impairment.

Nursing home placement. During the follow-up interviews, information was

collected on whether the patient moved into long-term care or died during follow-up. The

25

date of such events was recorded and used to calculate time to events during the 3 years
of follow-up. Cox proportional hazard models were used to perform survival analysis of
nursing home placement and mortality outcomes. The use of survival analysis allowed
for inclusion of continuous and/or dichotomous variables in the models, thus allowing for
baseline adjustment of covariates. For these analyses, differences in the time to nursing
home placement or death were compared in neuropsychiatric subgroups, after adjusting
for important baseline differences in the subgroups, such as presence of Parkinsonism,
cognitive impairment, marital status, age, and comorbid medical conditions. The choice
of covariates to be included in the model was based on previous reports, observed
correlations, and signiﬁcant group differences in scores in the sample, and our aim of
ﬁnding the most parsimonious model.

At the 3-year time point, 11 of the 32 (34%) in the Minimally Symptomatic
subgroup, 27 of the 49 (55%) in the Highly Symptomatic subgroup, and 11 out of 27
(41%) in the Predominantly Apathetic subgroup had been placed into nursing homes. On
average, the number of days to nursing home placement was 918.28 days for the
Minimally Symptomatic group, 694.79 days for the Highly Symptomatic group, and
802.57 days for the Predominantly Apathetic group. Figure 4 illustrates the observed
survival curves in days to nursing home placement in the three subgroups.

The Cox proportional hazard model showed that the overall model, which
included the 3 cluster solutions based on the presence of neuropsychiatric symptoms, and
the covariate of presence of Parkinsonism, predicted time to nursing home placement

(2 = 0.00). Of the covariates that were considered, only presence of Parkinsonism (p <

.05) was signiﬁcant. It should be noted that adding other covariates of functional

26

impairment, age, martial status and severity of co-morbid conditions, did not produce an
appreciable difference in the overall ﬁt of the model. Thus, after adjusting for the
covariate presence of Parkinsonism, the subgroups, based on the presence of
neuropsychiatric disturbances, signiﬁcantly predicted nursing home placement over a 3-
year period (9 < .05). However, post-hoc pair-wise comparisons did not yield signiﬁcant
results (See Table 4).

Survival. At the 3-year time point, 5 of 37 (14%) in the Minimally Symptomatic
group, 20 of 53 (38%) in the Highly Symptomatic group, and 12 of 31 (39%) in the
Predominantly Apathetic group had died. The mean numbers of days alive for each group
were as followed: 1051.40 days for the Minimally Symptomatic group, 905.70 days for
the Highly Symptomatic group, and 966.01 days for the Predominantly Apathetic group.
Figure 5 illustrates the observed survival curves for time to death in the 3 subgroups.

Using the Cox proportional hazard model, the overall model, which includes the 3
clusters and the presence of Parkinsonism as a covariate, signiﬁcantly predicted death (9
= 0.00) over a 3-year period. In addition, the presence of Parkinsonism (p = 0.00) was a
signiﬁcant predictor of death. Inclusion of additional covariates did not contribute to a
signiﬁcant change in the model. Thus, with the presence of Parkinsonism included, the
clusters based on the presence of neuropsychiatric symptoms remained a signiﬁcant
predictor of death over the 3-year period (2 <.05). Additionally, post-hoe pair-wise
comparisons between the subgroups indicated that the Highly Symptomatic group was
3.6 times more likely to die than the Minimally Symptomatic group. Moreover, the
Predominantly Apathetic group was 3.2 times more likely to die than the Minimally

Symptomatic group. See Table 4.

27

SeveritLand frequency of neuropsychiatric symptoms

Based on the severity and frequency of neuropsychiatric symptoms, cluster
analysis suggested 2-cluster and 4-cluster solutions. In the 2-cluster solution, individuals
in Cluster 1 (_N_= 69) endorsed minimal symptoms, whereas individuals in Cluster 2 (N =
53) endorsed a high level of symptoms. In the 4-cluster solution, Cluster 1 (N = 42) was
composed of individuals who were Minimally Symptomatic. In Cluster 2 (N = 39),
individuals with Affective/Apathetic symptoms, such as depression, anxiety, and apathy,
were present. Cluster 3 (N = 27) included individuals who were Predominantly Apathetic.
Individuals in Cluster 4 (N = 14) were Highly Symptomatic with Psychotic Features.
Characteristics of the subgroups for 4-c1uster solution are presented in Table 5. Graphical
presentation of mean proﬁles for the clusters is shown in Figure 6.

One major advantage of the 4-cluster solution over the 2-cluster model in this
portion of the analyses is that the 4-cluster solution provides a ﬁner differentiation of the
symptom proﬁles. In other words, the 4-cluster solution provides information on the roles
of the Affective/Apathetic group (Cluster 2) and the Predominantly Apathetic group
(Cluster 3), in addition to the low and high symptom groups seen with the 2-cluster
solutions. Thus, a decision was made to omit the 2-cluster solution from further
discussion.

Also, it is noted that the emergence of cluster solutions with the use of Pearson’s
correlation as a base measure to examine severity and frequency of symptoms provides
support for the validity of the clusters. Speciﬁcally, given that the severity and frequency
of symptoms is a continuous variable, elevation in proﬁles, rather than the overall pattern,

in theory would be the natural basis for clustering. However, the use of Euclidean

28

distance to account for elevations in proﬁles produced clusters that were less robust than
the use of Pearson’s correlation. This suggests that the clusters were robust across
different base measure, and hence, indicates a strong case for validity of the clusters.

Caregiver Distress. At baseline, the Minimally Symptomatic group had a total
mean score of 11.90 (S_D = 9.32) on the Caregiver Burden Scale. The Affective/Apathetic
group had a total mean score of 23.59 (SD = 10.65). The Predominantly Apathetic group
had a total mean score of 16.41 (S_D = 8.65). The Highly Symptomatic group had a total
mean score of28.50 (S_D = 15.31).

ANCOVA was used to examine the group differences in Caregiver Burden scores.
As with the presence of symptoms, the covariates of co-morbid medical conditions, the
presence of Parkinsonism, age, and the severity of cognitive and functional impairments
were initially included. Again, only functional impairment as a covariate was a
signiﬁcant predictor. Thus, in the ﬁnal model, only ﬁmctional impairment was included
as a covariate. After controlling for the covariate, ﬁmctional impairment, the four
subgroups differed signiﬁcantly on the total score of caregiver distress (E = 9.332; g =
0.00; 19% explained variance).

Post hoc comparisons indicated that the Minimally Symptomatic group differed
signiﬁcantly from the Affective/Apathetic group (p = 0.00), in that their caregivers
experience less distress. Similarly, the caregivers of the Minimally Symptomatic group
experience less distress than the caregivers of the Highly Symptomatic group (p_= 0.00).
In addition, the caregivers of the Affective/Apathetic subgroup endorsed higher distress

than the Predominantly Apathetic subgroup (p < .05). Lastly, the Predominantly

29

Apathetic subgroup showed lower level of caregiver distress than the Highly
Symptomatic subgroup (p = 0.00).

Quality of Life. Mean and standard deviations of HUI scores for the four
subgroups are presented in Table 6.

Using repeated measures AN OVA, the cluster differences in mean scores was
tested at three time points (Baseline, Year 1, Year 2). Potentially confounding variables,
the presence of Parkinsonism, cognitive impairments, and functional impairments, were
adjusted. The results indicated a signiﬁcant main effect of neuropsychiatric subgroups (E
(3, 93) = 13.51, p = 0.00, 30.4% explained variance). Also, the linear Time effect was
signiﬁcant (F (2, 93) = 6.66, p < .05, 6.6% explained variance). The interaction between
Time and the subgroups was not signiﬁcant (See Figure 7).

Post hoc Tukey tests indicated 3 signiﬁcant group mean differences. The mean
scores between the Minimally Symptomatic group and the Affective/Apathetic groups
differed signiﬁcantly (p = 0.00), with the Minimally Symptomatic group having a higher
score on the HUI. The mean scores between the Minimally Symptomatic group and the
Highly Symptomatic with Psychotic Features group differed signiﬁcantly (p = 0.00), with
the Highly Symptomatic with Psychotic Features group having a lower score on the HUI.
Lastly, the Affective/Apathetic group and the Predominantly Apathetic group showed
signiﬁcant differences in mean scores (9 < .05), with the Predominantly Apathetic group
having a higher quality of life than the Affective/Apathetic group.

Functional Impairment. The four subgroups’ means and standard deviations on

the Dependence Scale are presented in Table 7.

30

Repeated measures ANOVA was performed to test the group differences on the
Dependence Scale over a 2-year period. The covariates, the severity of cognitive
impairments, the presence of Parkinsonism, and the severity of co-morbid medical
conditions, were adjusted. The results showed the main effect neuropsychiatric subgroups
to be signiﬁcant (F (3, 93) = 5.19 2 =00, 14.3% explained variance). In addition, linear
Time effect (E (2, 93) = 5.00, p < .05, 5.1% explained variances), and linear Group x
Time effect (_E (3, 93) = 2.93, p < .05, 5.2% explained variances) were signiﬁcant (see
Figure 8).

Post hoc Tukey tests indicated a signiﬁcant group mean differences between the
Minimally Symptomatic group and the Affective/Apathetic subgroup (p = 0.00), with the
Affective/Apathetic group having a greater degree of functional impairment.

NursiggHome Placement. In Table 8, the number of individuals placed in nursing
homes after a 3-year period and the average time to nursing home placement are
presented for each cluster. Survival curves for the 4-cluster solution are presented in
Figure 9.

In the model, the covariate, the presence of Parkinsonism, was included. An
inclusion of severity of co-morbid medical conditions, marital status, and severity of
cognitive and functional impairments did not make an appreciable difference to the
overall model. Findings from the Cox proportional hazard model showed that the overall
model, which included the 4 clusters, along with the presence of Parkinsonism, was a
signiﬁcant predictor of nursing home placement (9 = 0.00). In addition, the presence of
Parkinsonism (p = 0.00) was a signiﬁcant covariate. Similarly, the differential outcome

of nursing home placement as predicted by the 4-clusters solution was signiﬁcant (p

31

<.05). Post-hoe pair-wise comparisons between the clusters did not yield signiﬁcant
results. See Table 4.

Survival. Table 8 shows the number of individuals in each subgroup that died and
the mean times to death. The survival curves are presented in Figure 10. The overall Cox
proportional hazard model, which included the 4-cluster solution and the covariate
presence of Parkinsonism, was signiﬁcant (p = 0.00) in predicting mortality over a 3-year
period. Inclusion of other covariates, age, cognitive and functional impairments, did not
change the model. The presence of Parkinsonism was a signiﬁcant covariate (p = 0.00).
Moreover, cluster membership was a signiﬁcant predictor of death (p < .01 ). Speciﬁcally,
a post-hoc pair-wise comparison indicated that the Highly Symptomatic group was 9.4
times more likely to die over a 3-year period than the Minimally Symptomatic group (p =

0.00). See Table 4.

32

Discussion

Overall, the current ﬁndings appear to provide evidence that AD patients with
neuropsychiatric symptoms can be meaningfully grouped based on the presence of
symptoms, as well as the severity and frequency of symptoms. A decision was made to
present the ﬁndings on the presence and the severity of symptoms in a parallel fashion,
rather than attempt to aggregate the ﬁndings, as aggregation could potentially lead to a
loss of critical information. In addition, this approach allowed the investigator to use the
established method of classifying based on the presence of symptoms, and concurrently
explore the potential beneﬁts of incorporating severity and frequency of symptoms. As
mentioned previously, a precedent for such an approach has been set (e.g. Tubman et al.,
1991)

The ﬁrst major ﬁnding of the study was that based on the presence of
neuropsychiatric symptoms, patients with AD can be categorized into three distinct
neuropsychiatric proﬁles: Minimally Symptomatic, Highly Symptomatic, and
Predominantly Apathetic. Using the same methodology, a categorization of AD patients
based on the severity and frequency of symptoms yielded two possible ways of grouping.
With the ﬁrst possibility, patients can be parsirnoniously identiﬁed as minimally or
highly symptomatic. However, with this strategy, limited information can be gained
beyond the general impact of neuropsychiatric symptoms on the individual. With the
second possible way of grouping, on the other hand, a ﬁner differentiation can be
achieved as 4 distinct symptom proﬁles were identiﬁed: Minimally Symptomatic,

Affective/Apathetic, Predominantly Apathetic, and Highly Symptomatic with Psychotic

33

Features. Here, in addition to the impact of high and low symptoms, we have information
on the Affective/Apathetic and the Predominantly Apathetic subgroups.

In reviewing the various proﬁles that emerged based on the presence and the
severity of symptoms, it was of note that the Minimally Symptomatic subgroup was
found consistently across the different methods of clustering. This ﬁnding was reﬂective
of the clinical picture in that a segment of the AD population has been noted to be
relatively unaffected by neuropsychiatric disturbances (Cummings, 2003; Lyketsos eta1.,
2001b; Moran et al., 2004).

In the current sample, this ﬁnding was of particular interest in that being
minimally symptomatic did not appear to be a function of severity of cognitive
impairments, as cognitive functioning was comparable across subgroups. Recently,
similar ﬁndings have been reported by Weiner and colleagues (2005). Thus, the severity
of AD pathology alone, as measured by cognitive decline and global degeneration of the
brain structures, does not necessarily account for the presence of neuropsychiatric
symptoms. Rather, it hints at a differential regional involvement in the process of
neuronal degeneration as a potential contributor to differential symptom development in
AD patients. Therefore, there is support for the view that the brain regions affected by
AD pathology can differ between individuals in such a way that some individuals have a
greater vulnerability to be affected disproportionately in one region rather than another
(Cummings, 2003a; Cummings, 2003b; Weiner et al. 2005). Thus, it would be of interest
to further explore the biological and the enviromnental characteristics that are particular

to the minimally symptomatic segment of the AD population.

34

In addition to the Minimally Symptomatic subgroup, a second notable subgroup
we identiﬁed was the Predominantly Apathetic group. The emergence of apathy as a
distinct neuropsychiatric proﬁle is supported by previous reports (Boyle & Malloy, 2003;
Landes, Sperry, & Strauss, 2005; Lyketsos et al., 2002). Similarly, the present ﬁnding
that the Affective/Apathetic group is a distinct group is also consistent with the literature.
Speciﬁcally, previous attempts to classify neuropsychiatric symptoms had found the
“affective factor” to be one of the most consistent ﬁndings across different classiﬁcation
methods (Aalten etal., 2003; Craig et al., 2004; Frisoni et al., 1999; Hope et al., 1997;
McShane, 2000; Schreinzer et al. 2005). Additionally, this differentiation of apathy from
a combined affective and apathy symptoms has been reported previously. Speciﬁcally,
Lyketsos and co-workers (2001a) found that a subgroup of AD patients endorsed apathy
without depression, whereas another subgroup endorsed both depression and apathy.
Similar ﬁndings have been reported in other studies (Holthoff et al., 2005; Landes etal.,
2001; Starkstein et al., 2001).

Of note, the current ﬁndings showed that the use of frequency and severity of
symptoms to classify resulted in an identiﬁcation of a combined Affective/Apathy
subgroup. Therefore, using severity and frequency of symptoms as the criteria for
classiﬁcation may prove to be advantageous in that it identiﬁes the affective/apathy
proﬁle, which has unique potentials for predicting disease outcomes.

A third subgroup identiﬁed by the clustering methods is the high symptomatology
group. Speciﬁcally, the clustering based on the presence of symptoms yielded a Highly
Symptomatic group, whereas the clustering based on the severity and frequency produced

a Highly Symptomatic with Psychotic Features group. There has been a precedent for

35

isolation of psychotic symptoms as a distinct symptom proﬁle (Mirakur etal., 2004;
Firsoni et al., 1999; Lyketsos et al., 2001b; Schneider & Dagennan, 2004). Furthermore,
there have been previous reports of an association between psychotic symptoms and an
overall high symptomatology (Bassiony et al., 2000). Moreover, in previous studies, there
had been reports of psychotic symptoms and high levels of neuropsychiatric symptoms
associated with more severe dementia (Craig et al., 2005). However, current ﬁndings
were not in supportive of these prior studies, as the cognitive impairments did not differ
signiﬁcantly between groups.

As for potential causal explanations for an emergence of this subgroup, recent
studies pointed to neurobiologic explanations. First, one study found that a decrease in
serotonin level in the temporal cortex was associated with an increase in psychotic
symptoms in AD, especially among women (Garcia-Alloza et al., 2005). Additionally,
there had been suggestions of underlying genetic predisposition for the development of
psychotic symptoms with AD, and its associated neurological changes (Sweet et al.,
2005; Holmes et al., 2001). Further research is needed to replicate these ﬁndings and to
validate the region- and gene-speciﬁc neurobiologic basis for the development of certain
neuropsychiatric symptoms.

Impact of neuropsychiatric symptoms on outcome measures

In addition to the primary objective of identifying neuropsychiatric symptoms as
distinct symptom proﬁles, the aim of this research also was to test the predictive validity
of subgroups on outcome measures. The present ﬁndings indicated that the subgroups,

regardless of whether the groups were based on the presence of symptoms, or on the

36

severity and frequency of symptoms, were able to predict caregiver distress, quality of
life, functional impairment, nursing home placement, and mortality.

Caregiver Distress. The extent of caregiver distress was differentially predicted by

 

the subgroups. More speciﬁcally, based on the presence of symptoms, the caregivers of
the Highly Symptomatic subgroup endorsed a greater degree of burden than the
Minimally Symptomatic and the Predominantly Apathetic group. Based on the severity
and frequency of symptoms, however, the caregivers of the Minimally Symptomatic and
the Predominantly Apathetic groups showed less distress than the Affective/Apathetic
and the Highly Symptomatic with Psychotic Features groups. These ﬁndings suggested
that caregivers are less burdened when the AD patients have a low level of
neuropsychiatric symptoms, or when their primary symptom is apathy. In addition, this
ﬁnding illustrated how the use of severity and frequency of symptoms to classify could
provide greater differential pathways to disease outcomes, as burden experienced by
caregivers of the Affective/Apathy group resembled that of the Highly Symptomatic
group, rather than that of the Predominantly Apathetic group.

These ﬁndings were in agreement with previous reports of high level of
neuropsychiatric symptoms associated with greater caregiver distress and lower quality of
life for the caregiver (Craig et al., 2005; Shin et al., 2005). However, it appears that the
burden is partly determined by the type of symptom, rather than by the general presence
of symptoms, as had been suggested previously (Teri, 1997; Rinaldi et al., 2005). As
reviewed by Ballard and colleagues (2000), severe mood disturbances, aggression,
restlessness, and apathy had been suggested in previous studies to be the symptoms most

associated with distress. In the current study, mood disturbances were consistent with the

37

Affective/Apathy, and aggression and restlessness were consistent with the Highly
Symptomatic group. Therefore, our ﬁndings appeared consistent with symptom-speciﬁc
ﬁndings from the literature. However, the current ﬁndings were not in support of apathy
causing greater distress in caregivers. One hypothesis is that the discrepancy in ﬁndings
is due to the incorporation of functional impairment as a covariate in our model. In
particular, it has been suggested that a more comprehensive examination of caregiver
burden, in addition to the NPI score, would produce a more accurate assessment of the
relationship between caregiver distress and neuropsychiatric symptoms (Rinaldi et al.,
2005). However, when this hypothesis was tested by removing the covariateﬁmctional
impairment from the model, our ﬁndings remained the same. Further studies are needed
to test the validity of ours and previous frndings.

As for why affective and high level of symptoms, in particular, cause greater
burden in caregivers, one potential explanation is that certain types of symptoms, such as
hallucination, depression, and agitation, adversely impact the quality of the relationship
between the patient and his or her caregiver. Therefore, when the quality of the
relationship is poor, the emotional reserve the caregiver has for handling behavioral
disturbances may be diminished. In fact, there has been a report of a decrease in quality
of the relationship between AD patient and the caregiver as predictive of caregiver’s
depression and anxiety (Mahoney et al., 2005). Another potential explanation for the
ﬁndings is that affective and high symptoms may entail greater functional impairment,
placing greater practical burden on the caregiver. In addition, the caregiver may feel
anger and resentment toward the patient’s dependency (DeKosky & Orgogozo, 2001). In

the present study, consistent with previous reports, only functional impairment, as a

38

covariate, was a signiﬁcant predictor of caregiver distress. Overall, the results suggested
that caregivers of AD patients with affective and high neuropsychiatric symptoms
experience greater burden. Hence, future studies involving interventions strategies
targeted toward these particular populations of caregivers may lessen the burden.

Quality of Life. The differences in quality of life between AD patients were
predicted by the neuropsychiatric subgroups over a 2-year period. Post hoc analyses
indicated that based on the presence of symptoms, the Minimally Symptomatic and the
Predominantly Apathetic subgroups showed higher quality of life than the Highly
Symptomatic group. Based on the severity and frequency of symptoms, the Minimally
Symptomatic group showed higher quality of life than the Affective/Apathetic group and
the Highly Symptomatic with Psychotic Features group. Additionally, the
Affective/Apathetic group had a lower quality of life than the Predominantly Apathetic
group.

Although previous studies have not considered the impact of co-morbid
neuropsychiatric symptoms on quality of life, the current ﬁndings are consistent with
previous ﬁndings that affective symptoms, particularly depression, have a negative
impact on quality of life (Fassino et al., 2002; Logsdon et al., 1999; Gonzalez-Salavador
et al., 2000; Selwood et al., 2005). However, the results were not in support of previous
indications that quality of life is stable over time, as ﬁndings did show a signiﬁcant time
effect (Lyketsos et al., 2003; Selwood et al., 2005).

Potentially, this discrepancy in ﬁndings could be attributed to differences in
measures used to study quality of life. In the present study, the HUI, which is a speciﬁc

measure of quality of life based on general health status, was used. In the ﬁeld of

39

pharrnacoeconomics, the HUI is commonly used to calculate quality adjusted life years.
Although this provides one objective means of examining quality of life, the use of the
HUI discounts other factors contributing to quality of life, such as having strong social
support. Such differences in the focus of the quality of life measures likely contributed to
the differences in ﬁndings between this research and prior research regarding time
effects.

One proposed explanation for the ﬁnding of neuropsychiatric symptoms’ adverse
impact on quality of life is that the presence of symptoms causes greater distress in the
AD patient (Cummings, 2003a). Cummings (2003a) suggested that this distress, in turn,
decreases the quality of life. Therefore, the development of interventions aimed at the
symptoms, and their associated distress, may potentially increase quality of life for the
AD patient.

Functional Impairment The neuropsychiatric subgroups differentially predicted
functional impairments over a 2-year period. Post hoc analyses indicated that based on
the presence of symptoms, the Minimally Symptomatic group showed a slower rate of
decline in functional impairment than did the Highly Symptomatic group. Based on the
severity and frequency of symptoms, the Affective/Apathetic group showed a higher rate
of functional decline than the Minimally Symptomatic group.

Previously, the impact of neuropsychiatric symptoms on functional impairment
had been examined using the total NPI score or the individual symptom approach. Based
on these reports, the present ﬁndings are consistent with the general consensus of
behavioral symptoms having an adverse impact on functional impairment (Weiner et al.,

2005; Norton, Malloy, & Salloway, 2001; Tekin et al., 2001). However, our ﬁndings

40

differed from previous ﬁndings of apathy (Boyle et al., 2003) and hallucination (Mok et
al., 2004) predicting signiﬁcant decline in functioning. Potentially, these discrepancies in
ﬁndings might have been due to differences in the approach used; that is, whether
individual symptoms examined alone versus incorporation of co-occurring symptoms.

In addition to the main effects of neuropsychiatric subgroups, the study found
cognitive deﬁcits and presence of Parkinsonism to be signiﬁcant covariates. These
ﬁndings were also consistent with previous reports (Mok et al., 2004; Boyle et al., 2003;
Lyketsos et al. 2005). Therefore, interventions aimed at reducing functional impairment
should consider neuropsychiatric symptoms, cognitive impairments, and presence of
Parkinsonism.

Lastly, based on the severity and frequency of symptoms, post-hoe analyses
indicated a signiﬁcant time by group interaction between functional impairment and
neuropsychiatric subgroups. Speciﬁcally, in the ﬁrst year of the study, the Highly
Symptomatic with psychotic feature group showed a slight decrease in functional
impairment in relation to the Affective/Apathetic subgroup. However, in the second year
of the study, the Highly Symptomatic group showed a signiﬁcant increase in ﬁmctional
impairment in relation to the Affective/Apathetic subgroup. Potentially, this ﬁnding can
be explained by the fact that the majority of the patients in the Highly Symptomatic
group were likely on antipsychotic medications and that by the second year of the study
signiﬁcant side effects associated with the longer-term use of the medications contributed
to the decline in functioning.

Nursing Home Placement Both clustering approaches showed neuropsychiatric

subgroups to be predictive of differential times to nursing home placement over a 3-year

41

period. However, post-hoe pair-wise comparisons were not signiﬁcant. The null ﬁndings
of the pair-wise comparison tests can likely be explained by two factors. First, although
our full model indicated overall signiﬁcant results for the subgroups, the ﬁndings were
less robust in that only 2 out of the 3 signiﬁcance tests typically used in survival analysis
showed signiﬁcant ﬁndings. Thus, ﬁndings from this portion of the analyses should be
interpreted with caution. Second, due to the nature of the study, the pair-wise
comparisons entailed lower power, as samples were smaller in post hoc analyses than in
the original model. Therefore, the likelihood of the tests detecting group differences was
smaller.

Thus, although post hoc tests did not indicate speciﬁc group differences,
observations of the means and the graphs suggested that individuals with high symptom
proﬁles were at a higher risk for nursing home placement. In contrast, individuals with
minimal symptoms consistently were at a lower risk for nursing home placement. Hence,
the high symptom proﬁle might be a risk factor for nursing home placement. These
ﬁndings are consistent with previous reports (Balestreri et al., 2001; Colerick & George,
1986; Gilley et al., 2004; Kopetz et al., 2000; Yaffe et al., 2002).

One explanation for the ﬁnding of individuals with high symptom proﬁles having
a higher risk for nursing home placement potentially is that their caregivers have
difﬁculty coping with the unsettling nature of neuropsychiatric symptoms. In particular, it
has been reported that caregivers ﬁnd certain symptoms, such as psychotic symptoms,
especially difﬁculty to handle (Moh et al., 2005). Behavioral modiﬁcation programs to
manage the high level of symptoms may result in reduced caregiver distress and delay the

time to nursing home placement (Gilley et al., 2004 Lichtenberg et al., 2005). Also,

42

psychoeducational programs and support for caregivers may reduce the need for nursing
home placement. Such implementation of intervention strategies early in the symptom
development could lead to a reduction in the overall cost of care (Butler, 1995).

m. Both clustering methods predicted differential rates of death over a 3-
year period. Post hoc pair-wise comparisons indicated that based on the presence of
symptoms, the Highly Symptomatic group was 3.6 times more likely to die than the
Minimally Symptomatic group over a 3-year period. Similarly, the Predominantly
Apathetic group was 3.2 time more likely to die than the Minimally Symptomatic group
over a 3-year period. Based on the severity and frequency of symptoms, the Highly
Symptomatic group was 9.6 times more likely to die than the Minimally Symptomatic
group over a 3-year period. Thus, it suggested that an overall high neuropsychiatric
symptom proﬁle increases the risk of death over a 3-year period.

To our knowledge, there has been one other study to consider the effects of high
neuropsychiatric symptom proﬁle on mortality in AD, and the ﬁndings are consistent
with present ﬁndings (Weiner et al., 2005). Also, there are two additional studies that
examined the impact of individual neuropsychiatric symptoms on mortality. In one study,
an increase risk of mortality was found when depression was present (Ganguli, Dodge, &
Mulsant, 2002). In the second study, tactile hallucination was associated with greater risk
of mortality (Suh et al., 2005). Given the limited research in this domain, further studies
are needed to replicate the ﬁndings.

A potential explanation for the increased risk of mortality associated with high
symptom proﬁle is that in recent studies, there were indications that certain atypical

antipsychotic medications used for the treatment of psychotic symptoms in dementia may

43

increase the risk of mortality (Schneider, Dagerrnan, & Insel, 2005). In addition, there
have been previous reports of an increased risk of cerebrovascular incident with
Risperidone (Bordaty et al., 2003). Therefore, it is conceivable that an increase risk of
mortality observed in the study for individuals with high symptom proﬁle could partially
be attributed to the use of antipsychotic drugs in these patients. In future studies, it would
be of importance to consider drug use when examining the relationships between
neuropsychiatric symptoms and mortality in AD.

Study’s Implications:

Findings from the study have both research and clinical implications. First, in
terms of research, the results of the current study provide empirical support for the
categorization of neuropsychiatric disturbances using cluster analysis. Hence, if the
method is validated in replication studies, its use in future studies might potentially aid
with ﬁnding greater consistency in outcomes. Also, it offers a tool for which
neurobiologic origins of neuropsychiatric symptoms can be explored. For instance, rather
than looking speciﬁcally at the neurological basis for depression in AD, it allows the
researcher to examine a proﬁle of co-occurring symptoms associated with depression. By
taking such an approach, consistency in ﬁndings with regard to neurological changes in
AD can potentially be achieved. By the same token, the method can be applied to
exploration of possible subtypes in AD. There have been suggestions of subtypes of AD
that show unique neuropathology, disease characteristics, including neuropsychiatric
proﬁles, and disease outcomes (Aalten et al., 2003; Lyketsos et al., 2001b; Cook et al.,
2003). An empirically-validated approach to classifying neuropsychiatric disturbances

may aid in an effort to address this possibility. From our ﬁndings, the clustering based on

44

the severity and frequency of symptoms appears suitable for research purposes, as it
identiﬁed an additional meaningful subtype to that of clustering based on the presence of
symptoms.

There are also a number of clinical implications from the study. First, the results
contribute to recent efforts in the ﬁeld to establish an empirically validated classiﬁcation
system for neuropsychiatric symptoms associated with dementia. Prior to this effort, an
initial attempt at classiﬁcation was carried out by a panel of experts at a consensus
conference (Finkel, 1996). However, recent developments in the ﬁeld had called for
establishing a system with empirical validation (Aalten et al., 2003; Lyketsos et al.,
2001b). Thus, based on ﬁndings from a latent class analysis study (Lyketsos et al.,
2001b), Lyketsos and colleagues (2001 a) proposed criteria for two syndromes:
Alzheimer-associated affective disorder and Alzheimer-associated psychotic disorder. It
is noted that the criteria set forth for Alzheimer-associated psychotic disorder is
consistent with our ﬁndings regarding the Highly Symptomatic group.

On the other hand, although the criteria proposed for Alzheimer-associated
affective disorder have features resembling our Predominantly Affective group and the
Affective/Apathy group, there is a signiﬁcant divergence. Speciﬁcally, the criteria for
affective disorder do not include apathy, although the group identiﬁed in the original
latent class analysis study showed apathy to be highly prevalent. Thus, based on current
ﬁndings, a proposed revision to the criteria put forth by Lyketsos and colleagues (2001a)
would be to differentiate two Alzheimer-associated disorders with affective symptoms. In
one, an emphasis would be given to the present of both affective symptoms, such as

depression, and apathy, in addition to other ‘less prominent’ symptoms such as those

45

proposed by Lyketsos et al. (2001a). In the second, the presence of signiﬁcant apathy
would be the main criterion. However, it is noted that having apathy as the main
diagnostic criteria does not exclude the possibility of other symptoms being present to a
lesser degree.

Second, the ﬁndings identiﬁed at-risk proﬁles for poor patient and caregiver
outcomes. Interventions targeted at symptoms associated with these proﬁles may prove
critical in altering the negative outcomes. Such interventions at the level of multiple-
syrnptom that co-occur have been suggested as an effective approach in treating
neuropsychiatric symptoms (Tractenberg et al., 2003), as more options for interventions
are becoming available in recent years. For instance, there are now a number of
behavioral management strategies for use with neuropsychiatric disturbances
(Lichtenberg et al., 2005; Politis et al., 2004). In addition, although opinions on the use of
pharmacological treatment vary, especially after recent ﬁndings on its potential risks, an
individualized consideration of risk and beneﬁts has been encouraged (Sink, Holden, &
Yaffe, 2005a; Sink, Holden, & Yaffe, 2005b; Rabin & Lyketsos, 2005). In assessing the
need for interventions, our study showed NPI to be a useful measure in effectively
identifying patients at-risk. Therefore, its use in clinical and research settings was
supported.

Also in terms of clinical implications, the ﬁndings on the role of apathy were of
particular interest. In terms of its impact on nursing home placement, the predominantly
apathetic group fell between the low and the high symptom proﬁles. This made intuitive
sense in that the caregivers of the AD patients endorsing apathy would ﬁnd their

apathetic behavior more tolerable and easy to care for than the individuals with high

46

symptom proﬁle. This was supported by ﬁndings on caregiver distress. Therefore, with a
lower caregiver distress, there would be less of a need for placing the patient in a long-
term care setting. An intriguing result, however, was that even though caregivers may
ﬁnd apathy more tolerable, the presence of apathy appeared just as detrimental as the
high symptom proﬁle when it came to predicting mortality. In the 3-cluster solution, the
Predominantly Apathetic and the Highly Symptomatic group had similarly low rates of
survival. In the 4-cluster solution, the Predominantly Apathetic group had a similar rate
of survival as the Affective/Apathetic group. Hence, it suggested that although the
presence of apathy in AD patients appears less destructive outwardly, it still has a
detrimental impact on the patient’s survival, and should not be overlooked as a target for
treatment.
Study Limitations

There were a number of limitations in the study. First, it was possible that the
presence and the severity of neuropsychiatric symptoms changed as the dementia
progressed. Given that the symptoms were assessed at the onset of the study, the potential
confounding inﬂuence of changes in neuropsychiatric symptoms over the 3-year period
could not be evaluated. However, of the available literature on the topic, one study
indicated that neuropsychiatric symptoms remained comparable across the study period
of a year and a half (Steinberg et al., 2004). Second, the NPI, used to assess
neuropsychiatric symptoms in the study, was based on informant reports. Hence, it was
conceivable that an informant bias in the reporting of the symptoms might have existed.
However, when assessing the presence of symptoms in AD patients with signiﬁcant

cognitive deﬁcits, the use of an informant may be unavoidable. Lastly, although the

47

initial sample size of 122 AD patients was respectable, the sample size became smaller as
subgroups were identiﬁed. This was especially true for the 4-cluster solution, which in
one of its groups, contained a sub-sample of only 14 patients. The small sample size is a
strength of the study in that the cluster proﬁles were predictive of outcomes. This
suggests a robust relationship between the clusters and the outcome variables.

Directions for future studies:

Given the promising indications of neuropsychiatric symptom proﬁles having
differential disease outcomes, replication studies are needed to determine whether similar
subgroups emerge in different samples. In addition, further explorations are needed to test
the extent of this predictive validity. For instance, it would be useful to test the predictive
validity in other outcome measures such as rate of cognitive decline, or differences in
medication response as suggested by Lyketsos and colleagues (2001a). Along the same
line, future studies could potentially test the usefulness of the methodology employed for
classifying neuropsychiatric disturbances in AD to other dementias, such as dementia
with Lewy bodies. Such application of the method to other dementias may provide an
empirical validation for the clinical use of neuropsychiatric symptoms in differential
diagnosis of dementias. Lastly, and perhaps most importantly, if the ﬁndings of
neuropsychiatric symptom proﬁles having differential disease outcomes prove to be
consistent across other outcome measures and dementias, an exploration into the potential
neurobiological underpinnings of the proﬁles appears warranted. For example, it is
possible that the presence of certain underlying genetic characteristics or an involvement
of particular brain structures in the disease process predispose an individual to have a

certain neuropsychiatric proﬁle (Cummings, 2003; Lam et al., 2004).

48

Summgy

Overall, the ﬁndings from the study suggested that cluster analysis was a suitable
approach for examining neuropsychiatric symptoms in AD as proﬁles of symptoms that
co-occur. Speciﬁcally, in the current study, cluster analysis identiﬁed meaningful, well-
deﬁned subgroups of AD patients that can predict caregiver and patient outcomes.
Additionally, ﬁndings from the study showed that such subgroups can be identiﬁed based
on the presence of symptoms, and on the severity and frequency of symptoms. However,
when the two clustering approaches were compared, it appeared that clustering based on
the severity and frequency of symptoms provided more information in that it identiﬁed a
combined Affective/Apathetic group. In certain outcomes, this identiﬁcation was
advantageous in that there were distinct outcomes for this subgroup.

In addition to the subgrouping of neuropsychiatric symptoms in AD patients, the
study identiﬁed at-risk proﬁles for clinical outcomes. Therefore, if future replication
studies ﬁnd the groups to be stable across samples, it warrants an investigation into
targeted intervention strategies and the neurobiologic origins of the subgroups. By
exploring such potentials, the ﬁeld could move a step closer toward the development of
more effective and targeted treatment options for neuropsychiatic symptoms, and thereby

may be able to alter the outcomes of patients with AD.

49

APPENDIX

TABLES AND FIGURES

50

Table l

Cluster Mean Proﬁles for the Presence of Symptoms

 

 

Cluster 1 Cluster2 Cluster3
N 38 53 31
Age 75.79 76.81 76.71
MMSEscore 16.58 15.19 16.71
CIRS score 19.45 21.70 20.48
EPS score 2.84 3.17 3.65
Dependence score 2.76 3.60 3.03
Total NPI score 5.58 28.60 13.74
Delusions 0.08 0.40 0.29
Hallucinations 0.03 0.32 0.16
Agitation 0.11 0.81 0.13
Depression 0.1 8 0.64 0.42
Anxiety 0.08 0.66 0.42
Elation 0.1 1 0.21 0.00
Apathy 0.08 0.77 0.94
Disinhibition 0.1 8 0.57 0.00
Irritability 0.06 0.73 0.1 1
Aberrant Motor 0.32 0.51 0.29
Behavior

 

Note: Total NPI score = Total score on the Neuropsychiatric Inventory; MMSE score =
score on the Mini-Mental Status Exam; CIRS score = score on the Cumulative Illness
Rating Scale; EPS score = score on the Modiﬁed Colombia University Parkinson’s

Disease Rating Scale; Dependence score = score on the Dependence Scale.

51

Table 2

Health Utilities Index Mean Scores for the Presence of Symptoms

 

 

 

 

Cluster N _hiean _S._D_.

Minimal Symptom 38 0.717 0.20

Baseline High Symptom 53 0.508 0.18
Apathetic 3 1 0.644 0.19

Minimal Symptom 36 0.707 0.20

Year 1 High Symptom 48 0.523 0.20
Apathetic 30 0.591 0.22

Minimal Symptom 37 0.618 0.20

Year 2 High Symptom 39 0.431 0.19
Apathetic 25 0.549 0.22

 

52

Table 3

Dependence Scale Mean Scores for the Presence of Symptoms

 

 

 

 

Cluster N m Si);

Minimal Symptom 38 2.763 1.03

Baseline High Symptom 53 3.604 1.10
Apathetic 31 3 .032 1.05

Minimal Symptom 36 3 .000 1 .12

Year 1 High Symptom 48 3.771 1.23
Apathetic 29 3 .276 1 .1 3

Minimal Symptom 37 3.351 1.25

Year 2 High Symptom 39 3.923 1.22
Apathetic 25 3 .440 1 .04

 

53

Table 4

Pairwise Compairsons between Clusters

 

Predictor [3
Nursing home placement
Presence of symptoms:

 

Cluster 1 vs. Cluster 2 0.525
Cluster 1 vs. Cluster 3 0.137
Frequency and severity of symptoms:
Cluster 1 vs. Cluster 2 0.589
Cluster 1 vs. Cluster 3 0.388
Cluster 1 vs. Cluster 4 0.606
Probability of death
Presence of symptoms:
Cluster 1 vs. Cluster 2 1.285
Cluster 1 vs. Cluster 3 1.171
Frequency and severity of symptoms:
Cluster 1 vs. Cluster 2 0.988
Cluster 1 vs. Cluster 3 0.925
Cluster 1 vs. Cluster 4 2.240

I;

1.691
1.147

1.803
1.474
1.834

3.614
3.226

2.685
2.521
9.396

p-value

0.09
0.71

0.08
0.29
0.21

0.01
0.03

0.07
0.1 1
0.00

 

54

Table 5

Cluster Mean Proﬁles for the 4-cluster Solution based on the Severity and Frequency of

 

Mats
Cluster 1 Cluster2 Cluster3 Cluster 4
N 42 39 27 14
Age 75.60 75.69 78.22 77.86
MMSEscore 17.67 14.90 16.59 13.00
CIRS score 19.50 22.18 20.74 20.00
EPS score 2.50 3.41 3.78 3.50
Dependence score 2.76 3.51 3.20 3.64
Total NPI score 4.50 25.74 11.78 45.93
Delusions 0.07 0.31 0.26 0.79
Hallucinations 0.05 0.21 0.19 0.57
Agitation 0.19 0.67 0.19 0.86
Depression 0.26 0.72 0.22 0.64
Anxiety 0.17 0.64 0.33 0.71
Elation 0.07 0.21 0.00 0.29
Apathy 0.14 0.80 0.93 0.79
Disinhibition 0.14 0.59 0.11 0.36
Irritability 0.13 0.49 0.25 0.92
Aberrant Motor 0.21 0.54 0.15 1.00
Behavior

 

55

Table 6

Health Utilities Index Mean Scores for the Severity and Frequency of Symptoms

 

 

 

 

Cluster N M_eap S_.D_,

Minimal Symptom 42 0.737 0.17

Baseline Affective/Apathetic 39 0.485 0. l 7
Apathetic 27 0.668 0. 1 7

High Symptom 14 0.444 0.16

Minimal Symptom 40 0.705 0.19

Year 1 Affective/Apathetic 36 0.514 0.18
Apathetic 26 0.609 0.24

. High Symptom 12 0.481 0.20

Minimal Symptom 41 0.644 0.20

Year 2 Affective/Apathetic 31 0.423 0.17
Apathetic 2 1 0.5 10 0.24

High Symptom 8 0.395 0.07

 

56

Table 7

mendence Sgtle Mean Scores for the Severity and Frequency of Symptoms

 

 

 

 

Cluster N Meg ﬁg

Minimal Symptom 42 2.762 0.98

Baseline Affective/ Apathetic 39 3 .5 1 3 1 .12
Apathetic 27 3.185 1.18

High Symptom 14 3.643 1.01

Minimal Symptom 40 2.850 1.00

Year 1 Affective/Apathetic 36 3.778 1.22
Apathetic 26 3 .560 1 .26

High Symptom 12 3.750 1.14

Minimal Symptom 41 3 .220 1.19

Year 2 Affective/Apathetic 31 3.903 1.19
Apathetic 21 3 .476 1 . 1 7

High Symptom 8 4.625 0.52

 

57

Table 8
Clusters based on the Severity and Frequency of Neuropsychiatric Symptoms as

Predictors of Nursing Home Placement over a 3-year Period

 

N In nursing home % Mean # of days SQ
nursing home
4-Cluster
Cluster 1 36 11 31 932.72 324.14
Cluster 2 35 19 54 754.71 416.91
Cluster 3 23 12 52 711.13 421.98
Cluster 4 14 7 50 690.57 475.49

 

58

Table 9

Clusters based on the Severity and Frequency of Neuropsychiatric Symptoms as

Predictors of Mortality over a 3 gear Period

 

N Died % Mean # of days S_D
death
4-Cluster
Cluster 1 41 5 12 1054.73 150.11
Cluster 2 39 13 33 968.69 234.43
Cluster 3 27 10 37 957.63 246.27
Cluster 4 14 9 64 820.50 307.63

 

59

FiguLe l
Cluster Means of Neuropsychiatric Smtoms based on M Presence of Smtoms

 

 

10 I' l I I l l I l
0.9— a. _
0.8— ~ ltl —
0.7- V l —
0.6— w 2 l i l , —
0.5— _ / l —
on   ,- l l .
0.3 ' '

0.2
D Minimally Syrmtomatic

0" - o l-ighly Syrmtomatic
0.0 ' El Predominantly Apathetic

Nd‘ \0 0“ at
0"» so‘ 26:99 eésig’ﬂ‘ $333 £§§n°

of"
NPI Domains

Mean Scores

Presence of Symptoms

 

 

 

 

Images in this dissertation are presented in color.

60

FiggZ

Quality of Life ova 2-Year Period based on the Presence of Symptoms

 

0.9—

Mean 0.7”
Scores 0.6
0.5
0.4
0.3
0.2

01*

I

I

I

I

I

 

1 I I

 

 

0.0

Baseline Year 1 Year 2

Years

61

Neuropsychiatric Subgroups
—- Minimally Symptomatic
—-- Highly Symptomatic
—— Predominantly Apathetic

Figure 3

Functional Impairment over g 2-Year Period based on the Presence of Symptoms

 

 

 

 

6 l T l
5 _ _
Mean
Scores 4 F f ‘ﬂfﬂf -*
fair/:7!
3 — :Z/ r
2 _ _.
Neuropsychiatric Subgroups
1 T T -—— Minimally Symptomatic
w- Highly Symptomatic
0 ‘ 1 ‘ —— Predominantly
Baseline Year 1 Year 2 Apathet'c
Years

62

Figure4

Estimated Probability of Nursing:Home Placement based on the Presence of Symptoms

Survival Function at mean of covariates

 

1.04
09—1
08-1
0.7%

06-1

0.5—l

ml

0.3-

 

 

 

 

I
200

I
400

I I
600 800

Total days home

63

I
1 000

1200

3-cluster solution
_j'] 'Minimally symptomatic'

_ﬂ 'ngth symptomatic'

'Predominantly apathetic'

Figure5

Estimated Probability of Death Based on the Presence of Symptoms

 

 

1

I 1

 

1

 

 

1.0 r

C

.3 0.8—

‘c’

3

u.

a 0.6—

.2

z

3

‘0 0.4—
0.2—
0.0 1

0 200 400 600 800 10001200
Time (Day)

64

3-cluster solution

~——— "Minimally symptomatic”
——- ‘Highly symptomatic'
—— “Predominantly apathetic'

Figure 6

4-C1uster Means of Neuromchiatric Smatoms based on the Severity and quency of

 

Symptoms

 

1-0 i I I l I I l l
0.9 _, r -
o.e ’

0.7
0.6
0.5
0.4

Mean Scores

03 Frequency and Severity of Syrmtoms

02 " .. —— Mimiiy Womaﬁc

0 1 h x . ’ _‘ —————- Affective/Apathetic .

0’0 1 ‘ ‘ L L i ' ——--- Eigﬂy Smmmm Psychotic Featues
"or «w: a a «was as at

09 \‘F V: T \ I

if rte“? ﬂﬁdtef

l

 

 

 

 

MDomains

65

Figure 7

Quality of Life over a 2-Year Period based on the Severity and Frequency of Symptoms

 

 

 

 

1.0 I I I
0.9 — —
Mean 0‘8 I H g _ "
Srnres 0.7 r “ “XXI"“N--- ~
0.6 — -«
0.5 — /_\ a
0.4 —
0.3 — a Neuropsychiatric Subgroups
0.2 — — -~~-— Minimally Symptomatic
0 1 l- _ - Affective/Apathetic
' Predominantly Apathetic
0.0 _ ' P 4 ——- Highly Symptomatic with Psychotic
Baseline Year 1 Year 2 Features

Years

66

Figure 8

Emotional Impairment over a 2-Year Period based on the severity and Frequency of

 

 

 

 

Smptoms
6 I I I
5— _
Mean
Scores4" \__/ T
3.. _ J ’//, / / _
2” 4 Neuropsychiatric Subgroups
1_ a "— Minimally Symptomatic
_"r Affective/Apathetic
l l l Predominantly Apathetic
0 —— Highly Symptomatic with Psychotic
Baseline Year 1 Year 2 Features

Years

67

Figure 9

Estimated Probability of NursinLHome Placement based on the Severity and Frequency

 

 

 

 

 

 

 

 

 

 

of Symptoms
Survival Function at mean of covariates
4-cluster solution
1.0d [1 'Minimally
symptomatic"
0-94 _I_I “Affective!
Cumulative Apathetic”
Survival
0'8- ”Predominantly
apathetic”
0.7-
II 'Highiy
symptomatic"
0.6-
0.5-
0.4d
0.3%
l l l l l I l
0 200 400 600 800 1 000 1 200

Total days home

68

Figure 10

Estimated Probabilig of Death based on the Severity and Frequency of Syrautoms

Survival Plot

 

 

 

 

 

 

1 .0
,§ 0.8— —
2
u”.
3 0.6 r r
.2
a
3
a) 0.4 r —

4—oluster solution
0.2 - — -—-—-— ‘Minimally symptomatic”
° ~————— 'AffectivelApalhetic“
'Predominantly apathetic
0.0 ' 1 ‘ ‘ ' —— 'Highly sympotmatic'
0 200 400 600 800 1000 1200

Time (Days)

69

LIST OF REFERENCES

70

Aalten, P., de Vugt, M.E., Lousberg, R., Korten, E., Jaspers, N., Senden, B.,
Jolles, J ., Verhey, F.R.J. (2003). Behavioral problems in dementia: A factor analysis of
the Neuropsychiatric Inventory. Dementia and Geriatric Cognitive DisorderLl 5, 99-105.

Balestreri, L, Grossberg, A., & Grossberg, G. (2000). Behavioral and
psychological symptoms as a risk factor for nursing home placement. International

Journal of Psychogeriatrics, 12, (Suppll ), 59-62.

Ballard, C., Lowery, K., Powell, 1., O’Brien, J ., & James, I. (2000). Impact of
behavioral and psychological symptoms of dementia on caregivers. International
Psychogeriatrics, 12( Suppl. 1 ). 93-105.

Bassiony, M.M., Steinberg, M.S., Warren, A., Rosenblatt, A., Baker, A.S., &
Lyketsos, CG. (2000). Delusions and hallucinations in Alzheimer’s disease: prevalence
and clinical correlates. International Journﬂf Geriatric Psychiatry. 15. 99-107.

Boyle, P.A., & Malloy, RF. (2003). Treating apathy in Alzheimer’s disease.
Dementia and Geriatric Cognitive Disorders, 17, 91-99.

Brodaty, H, Ames, D., Snowdon, J. etal., (2003). A randomized placebo-
controlled trial for the treatment of aggression, agitation, and psychosis of dementia.
Journal of Clinical Psychiatry, 64, 134-143.

Butler, R. Strategies to delay dysfunction in later life. In: Dall, J .L.C., Ermini, M.,
Herrling, P., Meier-Ruge S.W., Staufenbiel, H. eds. Adaptations in Aging. London,
Academic Press.

Cook, S.E., Miyahara, S., Bacanu, S.A., Perez, M.G., Lopez, O.L., Kaufer, D.I.,
Nimoganokar, V.L., Wisniewski, S.R., DeKosky, S.T. & Sweet, RA. (2003). Psychotic

symptoms in Alzheimer’s disease: Evidence for subtypes. American Journal of Geriatric
Paychiatrm 11. 406-413.

Colerick, E.F. & George, L.K. (1986). Predictors of institutionalization among

caregivers of patients with Alzheimer’s disease. Journal of American Geriatric Society,
3_4_, 493-498.

Craig, D., Mirakhur, A., Hart, D.I., Mcllroy, S.P., & Passmore, AP. (2005). A
cross-sectional study of neuropsychiatric symptoms in 435 patients with Alzheimer’s
disease. American Journal of Geriatric Psychiatry. 13 (6). 460-468.

Cummings, J .L. (2003a). The neuropsychiag of Alzheimer’s disease and related

dementias. London, Martin Dunitz.
Cummings, J .L. (2003b). Advances in the neuropsychopharmacologic

management of behavioral alterations in Alzheimer’s disease. DementiL and Cognitive
Disorders. 17. 17-54.

71

Cummings, J .L. (1997). The Neuropsychiatric Inventory: Assessing
psychopathology in dementia patients. Neurology, 48( Suppl. 6), 810-8 16.

Cummings, J .L., Mega, M., Gray, K., Rosenberg-Thompson, S., Carusi, D.A., &
Gombein, J. (1994). The Neuropsychiatric Inventory: Comprehensive assessment of
psychopathology in dementia. Neurology, 44, 2308-2314.

DeKosky, S.T., & Orgogozo, J .M. (2001). Alzheimer disease: diagnosis, costs,
and dimensions of treatment. Alzheimer’s Disease and Associated Disorders, 15, 83-87.

Devanand, D.P., Jacobs, D.M., Tang, M.X., Del Castillo-Castaneda, C., Sano, M.,
Marder, K., Bell, K., Bylsma, F.W., Albert, M., & Stern, Y. (1997). The course of

psychopathologic features in mild to moderate Alzheimer’s disease. Archives of General
Psychiatry. 54, 257-263.

Everitt, B.S., Landau, S., & Leese, M. (2001). Cluster analysis (4th ed.). London:
Arnold.

 

F assino, S., Leombumi, P. Daga, G.A., Brustolin, A., Rovera, G.G., & Fabri, F.
(2002). Quality of life in dependent older adults living at home. Archives of Gerontology
and Geriatrics, 35, 9-20.

Feeny, D., Furlong, W., Boyle, M., & Torrance, G.W. (1995). Multiattribute
health status classiﬁcation systems: Health Utilities Index. PharmacoEconomics, 7 ( 6),
490-502.

Finkel, S. (1996). Behavioral disturbances in dementia. International
Psychogeriatrics, 8, 215-551.

Folstein, M.F., Folstein, S.E., & McHugh, RR. (1975). “Mini-Mental State”: A
practical method for grading the cognitive state of patients for the clinician. Journal of
Psychiatric Research, 12, 189-198.

Frisoni, G.B., Rozzini, L., Gozzetti, A., Binetti, G., Orazio, Z., Bianchetti, A.,
Trabucchi, M., & Cummings, J .L. (1999). Behavioral syndromes in Alzheimer’s disease:
Description and correlates. Dementia and Geriatric Cognitive Disorders. 10, 130-138.

Garcia-Alloza, M., Gil-Bea, F.J., Diez-Ariza, M., Chen., C.P.L.-H., Francis, P.T.,
Lasheras, B. & Ramirez, M.J. (2005). Cholinergic-serotonergic imbalance contributes to

cognitive and behavioral symptoms in Alzheimer’s disease. NeuropsychologiaL43L442-
449.

Gilley, D.W., Bienias, J .L., Wilson, R.S., Bennett, D.A., Beck, T.L. & Evans,

DA. (2004). Inﬂuence of behavioral symptoms on rates of institutionalization for persons
with Alzheimer’s disease. Psychological Medicine, 34, 1129-1135.

72

Gonzalez-Salvador, T., Arango, C., Lyketsos, C.G., & Barbar, AC. (1999). The
stress and psychological morbidity of the Alzheimer patient caregiver. International
Journal of Geriatric Psychiatry, 14, 701-710.

Gonzalez-Salvador, T., Lyketsos, C.G., Baker, A., Hovanec, L., Roques, C.,
Brandt, J ., & Steele, C. (2000). Quality of life in dementia patients in long-term care.
International Journg of Geriatric Psychiatry. 15, 181-189.

Hargave, R., Reed, B, & Mmgas, D. (2000). Depressive syndromes and
fmctional disability in dementia. Joumal of Geriatric Psychiatry and Neurolagygl 3, 72-
77.

Holmes, C., Smith, H., Ganderton, R., Arranz, M., Collier, D., Powell, J ., &
Lovestone, S. (2001). Psychosis and aggression in Alzheimer’s disease: the effect of
dopamine receptor gene variation. Journal of Neurology, Neurosurgeﬁry, and Psychiatry,
11,]77-779.

Holthoff, V.A., Beuthien-Baumann, B., Kalbe, E., Ludecke, S., Lenz, 01.,
Zmdorf, G., Spirling,S., Schierz, K., Winiecki, P., Sorbi, S., & Herholz, K. (2005).
Regional cerebral metabolism in early Alzheimer’s disease with clinically signiﬁcant
apathy or depression. Biological Psychiatry, 57, 412-421.

Hope, T., Keene, J., Fairburn, C., McShane, R., & Jacoby, R. (1997). Behavior
changes in dementia 2: Are there behavioural syndromes? International Journal of
Geriatric Psychiatry. 12, 1074-1078.

J orm, AF. (1991). Corss-national comparisons of the occurrence of Alzheimer’s
and vascular dementias. European Archives of Psychiatry and Clinical Neuroscience.

gﬂ, 218-222.

Kopetz, 8., Steele, C.D., Brandt, J ., Baker, A., Kronberg, M., Galik, E., Steinberg,
M., Warren, A., & Lyketsos, CG. (2000). Characteristics and outcomes of dementia

residents in an assisted living facility. Interpational Journal of Geriatric Psychiatry, 15,
586-593.

Lam, L.C.W., Tang, N.L.S., Ma, S.L, Zhan, W., & Chiu, H.F.K. (2004). 5-HT2A
T102C receptor polymorphism and neuropsychiatric symptoms in Alzheimer’s disease.
International Journal of Geriatric Psychiatry, 19, 523-526.

Landes, A.M., Sperry, S.D., & Strauss, ME. (2005). Prevalence of apathy,
dysphoria, and depression in relation to dementia severity in Alzheimer’s disease. Journal
of Neuropsychiatry and Clinical Neurosciences, 17, 342-349.

Levy, M.L., Cummings, J .L., Fairbanks, L.A., Bravis, D., Calvanik, M., & Carta,
A. (1996). Longitudinal assessment of symptoms of depression, agitation, and psychosis

73

in 181 patients with Alzheimer’s disease. American Journal of Psychiatry, 153, 1438-
1443.

Lichtenberg, R.A., Kemp-Havican, J. MacNeill, S.E., & Johnson, AS. (2005).
Pilot study of behavioral treatment in dementia care units. Gerontologist. 45, 406-410.

Linn, B.S., Linn, M.W., & Gurel, L. Cumulative Illness Rating Scale. Journal of
American Geriatric Society, 16, 622-626.

Logsdon, R.G., Gibbons, L.E., McCurry, S.M., & Teri, L. (1999). Quality of life
in Alzheimer’s disease: patients and caregiver reports. Journal of Mentﬁiealth aria

Aging, 5, 21-32.

Lyketsos, C.G., Breitner, J .C.S., & Rabins, P.V. (2001a). An evidence-based
proposal for the classiﬁcation of neuropsychiatric disturbance in Alzheimer’s disease.
International Journal of Geriatric Psychiatry. 16. 1037-1042.

Lyketsos, C.G., Gonzalez-Salvador, T., Chin, J .J ., Baker, A., Black, B., & Rabins,
P. (2003). A follow-up study of change in quality of life among persons with dementia
residing in a long-term care facility. International Journal of Geriatric Psychiam, 18,
275-281.

Lyketsos, C.G., Lopez, 0., Jones, 3., Fitzpatrick, A.L., Breitner, J ., & DeKosky,
S. (2002). Prevalence of neuropsychiatric symptoms in dementia and mild cognitive
impairment. Journal of Ameican Medical Association, 288(2), 1475-1483.

Lyketsos, C.G., Sheppard, J .M.E., Steinbergu, M., Tschanz, J .A.T., Norton, M.C.,
Steffens, D.C., & Breitner, J .C.S. (2001b). Neuropsychiatric disturbance in Alzheimer’s

disease clusters into three groups: The Cache Comty study. Interpational Journal of
Geriatric Psychiatry, 16L1043-1053.

Lyketsos, C.G., Steele, C., Baker, L., Galik, E., Kopmek, S., Steinberg, M., &
Warren, A. (1997). Major and minor depression in Alzheimer’s disease: Prevalence and
impact. Journal of Neuropsychiatry and Clinical Neurosciences. 9. 556-561.

Lyketsos, C.G., Steinberg, M., Tschanz, J .A.T., Norton, M., Steffens, D.C., &
Breitner, J CS. (2000). Mental and behavioral disturbances in dementia: Findings from

the Cache Comty Study on memory in aging. American Journal of Psychiatry, 157. 708-
714.

Lyketsos, C.G., Toone, L., Tschanz, J ., Rabins, P.V., Steinberg, M., Onyike, C.U.,
Corcoran, C., Norton, M., Zandi, P., Breitner, J.C.S., & Welsh-Bohmer, K. (2005).
Population-based study of medical comorbidity in early dementia and “Cognitive
Impairment, No Dementia (CIND)”: association with functional and cognitive
impairment: the Cache Comty Study. American Journal of Geriatric Psychiatry, 13, 656-
664.

74

Mahoney, R., Regan, C., Katona, C., & Livingston, G. (2005). Anxiety and
depression in family caregivers of people with Alzheimer’s disease: the LASER-AD
study. American Journal of Geriatric Psychiatry. 13. 795-801.

McShane, R. (2000). What are the syndromes of behavioral and psychological
symptoms of dementia? International Psychogeriatrics, 12 (Suppl. 1). 147-153.

Meyers, B. (2001). The evolving typology of neuropsychiatric complications of
Alzheimer’s disease: The use of latent trait analysis. International Joml of Geriatric
Psychiatry. 16, 1030-1032.

Miller, M.D., Paradis, C.F., Houck, P.R., Mazumbar, S., Stack, J.A., Rifai, A.H.,
Mulsant, B.C., & Reynolds, CF. (1992). Rating chronic medical illness in burden in

geropsychiatric practice and research: Application of Cumulative Illness Rating Scale.
PsychiatryResearch. 41(3), 237-248.

Mirakhur, A., Craig, D., Hart, D.I., Mcllroy, S.P., & Passmore, AP. (2004).
Behavioral and psychological symptoms in Alzheimer’s disease. International Journal of
Geriatric Psychiatry. 19.1 035-1039.

Moran, M., Lynch, C., Lynch, A., Coen, R.F., Coakley, D. & Lawlor, BA.
(2004). Syndromes of behavioral and psychological symptoms in mild Alzheimer’s
disease. International Jouml of Geriatric Psychiatryal 9, 359-364.

Murman, D.L., Kuo, S.B., Powell, M.C., & Colenda, CC. (2002). A comparison
of preference-weighted health status classiﬁcation system in Alzheimer’s disease.
Neurology. 58 (suppl.3). A186.

Murman, D.L., Kuo, S.B., Powell, M.C., & Colenda CC. (2003). The impact of
parkinsonism on costs of care in patients with AD and dementia with Lewy bodies.
Neurology, 61, 944-949.

Norton, L.E., Malloy, P.E., & Salloway, S. (2001). The impact of behavioral
symptoms on activities of daily living in patients with dementia. American Journal of
Geriatric Psychiatry, 9, 41-48.

Payne, J .L., Sheppard, J .M.E., Steinberg, M., Warren, A., Baker, A., Steele, C.,
Brandt, J ., & Lyketsos, CG. (2002). Incidence, prevalence, and outcomes of depression

in residents of a long-term care facility with dementia. International Journal of Geriaﬂg
Psychiatry. 17, 247-253.

Politis, A.M., Vozzella, S., Mayer, L.S, Onyike, C.U., Baker, A., S., & Lyketsos,
CG. (2004). A randomized controlled, clinical trial of activity therapy for apathy in
patients with dementia residing in long-term care. International Journal of Geriatric
Psychiatry. 19, 1087-1094.

75

Rabins, P.V. & Lyketsos, CG. (2005). Antipsychotic drugs in dementia: what
should be made of the risks? Journal of American Medical Association, 294, 1963-1965.

Richards, M., Marder K, Bell, K., Dooneief, G., Mayeux, R., & Stern, Y. (1991).
Interrater reliability of extrapyramidal signs in a group assessed for dementia. Archives
of Neurology. 48, 1147-1149.

Rinaldi, ., Spazzafumo, L., Mastriforti, R., Mattioli, P., Marvardi, M., Polidori,
M.C., Cherubini, A., Abate, G., Barotrelli, L., Bonaiuto, S., Capurso, A., Cucinotta, D.,
Gallucci, M., Giordano, M., Martorelli, M., Masaraki, G., Nieddu, A., Pettenati, C.,
Putzu, P., Tarmnaro, V.A., Tomassini, P.F., Vergani, C., Senin, U., & Mecocci, P.
(2005). Predictors of high level of burden and distress in caregivers of demented patients:

results of an Italian multicenter study. International Journal of Geriatric Psychiatry, 20,
168-174.

Schneider, LS. (2001). Assessing outcomes in Alzheimer disease. Alzheimer
Disease and Associated Disorder, 15 (Suppl.1), S8-Sl8.

Schneider, L.S. & Dagerman, KS. (2004). Psychosis of Alzheimer’s disease:
clinical characteristics and history. Journal of Psychiatric Research, 38L105-111.

Schneider, L.S., Dagerman, K.S., & Insel, P. (2005). Risk of death with atypical
antipsychotic drug treatment for dementia: meta-analysis of randomized placebo-
controlled trials. Journal of American Medical Association, 294, 1934-1943.

Schreinzer, D., Ballaban, T., Brannath, W., Lang, T., Hilger, E., Fasching, P., &
Fischer, P. (2005). Components of behavioral pathology in dementia. International
Jourrlal of Geriatric Psychiatry, 20. 137-145.

Selwood, A., Thorgrimsen, L., & Orelle, M. (2005). Quality of life in dementia—
a one-year follow-up study. International Journg of Geriatric Psychiatry, 20, 232-237.

Shin, S.II, Carter, M., Masterrnan, D., Fairbanks, L., & Cmrmings, J .L. (2005).
Neuropsychiatric symptoms and quality of life in Alzheimer’s disease. American Journal
of Geriatric Psychiatry, 13, 469-474.

 

Sink, K.M., Holden, K.F., & Yaffe, K. (2005a). Drugs minimally effective for
neuropsychiatric symptoms in dementia. Journal of Famin Practice, 54, 407.

Sink, K.M., Holden, K.F., & Yaffe, K. (2005b). Pharmacological treatment of
neuropsychiatric symptoms of dementia: a review of the evidence. Journal of the
American Medical Association, 293, 596-608.

76

Starkstein, S.E., Mizrahi, R., & Garau, L. (2005). Speciﬁcity of symptoms of
depression in Alzheimer disease: a longitudinal analysis. American Journal of Geriatric
Psychiatry. 13, 802-807.

Steinberg, M., Sheppard, J .M., Tschanz, J .A.,T., Norton, M.C., Steffens, D.C.,
Brietner, J .C.S., & Lyketsos, CG. (2003). The incidence of mental and behavioral
disturbances in dementia: The Cache Comty Study. Journal of Neuropsychiatry and
Clinical Neurosciences. 15. 340-345.

Steinberg, M., Tschanz, J .A., Corcoran, C., Steffens, D.C., Norton, M.C.,
Lyketsos, C.G., & Breitner, J CS. (2004). The persistence of neuropsychiatric symptoms
in dementia: the Cache Comty Study. International Journal of Geriatric Psychiatry, 19,
19-26.

Stern, Y., Albert, S.M., Sano, M., Richards, M., Miller, L., Folstein, M., Albert,
M., Bylsma, F.W., & Laﬂeche, G. (1994). Assessing patient dependence in Alzheimer’s
disease. Journal of Gerontology: Medical Sciences. 49(5). M216-M222.

Sweet, R.A., Devlin, B., Pollock, B.G., Sukonick,D.L., Kastango, K.B., Bacanu,
S-A., Chowdari, K.V., DeKosky, S.T., & Ferrell, RE. (2005). Catechol-O-

methyltransferase haplotypes are associated with psychosis in Alzheimer’s disease.
Molecular Psychiatry, 10, 1026-1036.

 

Suh, G.H., Yeon, B.K., Shah, A., & Lee, J .Y. (2005). Mortality in Alzheimer’s
disease: a comparative prospective Korean study in the community and nursing homes.
International Journal of Geriatric Psychiatry, 20, 26-34.

Tekin, S., Fairbanks, L.A., O’Connor, 8., Rosenberg, S., & Cummings, J .L.
(2001). Activities of daily living in Alzheimer’s disease: Neuropsychiatric, cognitive and
medical illness inﬂuences. American Journal of Geriatric Psychiatry, 9, 81-86.

Teri, L. (1997). Behavior and caregiver burden: Behavioral problems in patients
with Alzheimer disease and its association with caregiver distress. Alzheimer disease and
associated disorders. 11 (Suppl.4). 835-838.

Thomas, P., Ingrand, P., Lalloue, F., Hazif-Thomas, C., Billon, R., Vieban, F., &
Clement, J .P. (2004). Reasons of informal caregivers for institutionalizing dementia

patients previously living at home: The Pixel study. International Journal of Geriatric
Psychiatry. 19. 127-135.

Tractenberg, R.E., Weiner, M.F., Patterson, M.B., Teri, L., & Thal, L] (2003).

Comorbidity 0f psychopathological in community-dwelling persons with Alzheimer’s
disease. Journal of Geriatric Psychiatry and Neurology, 16, 94-99.

77

Tubman, J .G., Vicary, J .R., von Eye, A., & Lerner, J .V. (1991). Qualitative
changes in relationships between substance use and adjustment during adolescence.
Journal of Substance Abuse, 3, 405-414.

Vitaliano, P.P., Russo, J ., Yomg, H.M., Becker, J., & Maiuro, RD. (1991). The
Screen for Caregiver Burden, The Gerontologist. 31(1),76-83.

von Eye, A., & Bergman, LR. (2003). Research strategies in developmental
psychopathology: Dimensional identity and the person-oriented approach. Development
and Psychopathology. 15, 553-580.

von Eye, A., Mm, E. Y., Indurkhya, A. (2004). Typifying developmental
trajectories: A decision making perspective. Psychology Science, 46. 65-98.

Weiner, M.F. Hynan, L.S., Bret, M.E. & White, C. (2005). Early behavioral
symptoms and course of Alzheimer’s disease. Acta Psychiatrica Scandinavica. 111, 367-
371.

Yaffe, K., Fox, P., Newcomer, R., Sands, L., Lindquist, K., Dane, K., &

Covinsky, KB. (2002). Patient and caregiver characteristics and nursing home placement
in patients with dementia. Journal of the American Medical Association, 287, 2090-2097.

78

 

llllllllllllllllill

A

    

 
 

iii