University of Southern California Dissertations and Theses

Accuray of subjective cognitive complaints in a longitudinal context: the effect of depression and dementia status

(USC Thesis Other)

Accuray of subjective cognitive complaints in a longitudinal context: the effect of depression and dementia status

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

ACCURACY OF SUBJECTIVE COGNITIVE COMPLAINTS IN A LONGITUDINAL
CONTEXT: THE EFFECT OF DEPRESSION AND DEMENTIA STATUS

by
Jennifer B. Dave

_____________________________________________________________________

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(PSYCHOLOGY)

August 2009

Copyright 2009 Jennifer B. Dave

ii
Acknowledgments
A number of people were instrumental in the completion of this manuscript and
my time in graduate school. First, I would like to thank my graduate advisor, Bob
Knight, who has provided feedback, guidance, ideas, support, and encouragement when I
needed it. He and Carol McCleary assisted with brainstorming with me about insight and
subjective memory complaints and possible studies, and they contributed greatly to my
thinking about this topic. I also thank Carol McCleary for making neuropsychology
interesting enough for me to pursue it clinically and in research. Carol Prescott and
Archana Jajodia provided countless hours of statistical advice and assistance and I am
incredibly grateful for that. I also appreciate the time that my other two committee
members, John Walsh and David Walsh, spent in reading and commenting on my
dissertation and proposal. My lab mates, Cecilia Poon, Jenn Kellough, and Phil Sayegh,
were nice enough to listen to my too-long first run-through of my dissertation talk and
helped me cut it by one-third. I want to thank Poorni Otilingam, who also heard my talk
and brought up excellent thoughtful questions that helped prepare me for the actual
defense. She also has been a shoulder to lean on throughout grad school and has helped
foster my foodie-ness. That is truly something to appreciate. Thank you to Linda Ercoli,
who was completely understanding of my need to essentially take off half of the month of
April to complete my dissertation. I thank her and Jennifer Dunkin for providing helpful
feedback on my INS poster. On another personal note, I thank my parents and my sisters,
who have endured my limited availability and quick phone calls during the busy times in
grad school. My parents have also carried some of my anxiety on defense days (Mom)

iii
and of course, provided me with awesome genes, which has led to my spektakyulur
braynn. Probably most importantly, I thank with my whole heart Ashley Borders, who
has 1) offered support, 2) let me borrow her frontal lobes in the organization of my
paper(s), 3) listened to my talk multiple times, 4) been patient and kind, and 5) had good
taste.

iv
Table of Contents
Acknowledgements ii
List of Tables v
Abstract vii
Chapter One: Introduction 1
Chapter Two: Methods 22
Chapter Three: Results
Figure 1. Diagnosis moderates the
effect of depression on future change
in memory.

35
46
Chapter Four: Discussion 57
References 74
Appendix A 82
Appendix B 83

v
List of Tables
Table 1. Number of participants with complete data by hypothesis. 24
Table 2. Frequencies of complaints.

27
Table 3. Demographic statistics for sample. 33
Table 4. Percent dichotomous complaints by diagnostic group.

35
Table 5. Correlation matrix for key variables for Alzheimer’s Disease
participants only.

38
Table 6. Correlation matrix for key variables for normal controls only.

39
Table 7. Mean neuropsychological difference scores by diagnostic
group.

40
Table 8. Memory change and diagnosis predicting memory complaints.

41
Table 9. Means and standard deviations for future memory change
(time 2-time1) by dichotomous memory complaints (time 1) and
diagnosis.

41
Table 10. Past memory change and depression predicting memory
complaints.

43
Table 11. Past memory change, diagnosis, and depression predicting
memory complaints.

43
Table 12. Memory complaints and depression predicting future
memory change.

44
Table 13. Memory complaints, depression, and diagnosis predicting
future memory change (time 2-time1).

45
Table 14. Past attention change and diagnosis predicting attention
complaints.

47
Table 15. Means and standard deviations for future attention change
(time 2-time1) by dichotomous attention complaints (time 1) and
diagnosis.

48

vi
Table 16. Past attention change and depression predicting attention
complaints.

49
Table 17. Past attention change, diagnosis, and depression predicting
attention complaints.

49
Table 18. Attention complaints and depression predicting future
attention change.

51
Table 19. Attention complaints, diagnosis, and depression predicting
future attention change.

51
Table 20. Past language change and diagnosis predicting language
complaints.

52
Table 21. Means and standard deviations for future language change
(time 2-time1) by dichotomous language complaints (time 1) and
diagnosis.

52
Table 22. Past language change and depression predicting language
complaints.

53
Table 23. Past language change, diagnosis, and depression predicting
language complaints.

54
Table 24. Language complaints and depression predicting future
language change.

55
Table 25. Language complaints, diagnosis, and depression predicting
future language change.

56

vii
Abstract
The purpose of this study was to determine whether subjective complaints about
change in memory, attention, and language accurately reflect past decline or predict
future cognitive decline on neuropsychological tests in older adults with normal cognition
or with Alzheimer’s disease. Further, the effect of dementia status and depressive
symptoms on these relationships was assessed. Using the USC Alzheimer’s Disease
Research Center longitudinal dataset, subjective cognitive complaints, cognitive
performance, and depressive symptoms were assessed in 132 participants. Overall,
neither diagnostic group’s complaints accurately reflected past decline for any cognitive
domain. Rather, depressive symptoms predicted endorsement of complaints for all
domains. In addition, dementia status moderated the effect of depression on future
memory decline. Finally, language complaints predicted future language decline in the
entire sample, but similar relationships were not present for attention and memory.
Implications of these findings are discussed.

1
Chapter One. Introduction
In clinical and research settings, memory complaints are often discounted by
professionals, especially in the absence of difficulties on neuropsychological
examination. Knowing when complaints and self-reported change are accurate is
important for the diagnosis of dementia, especially early diagnosis. At present, it is
unclear how much weight should be given to cognitive complaints. Ignoring seemingly
innocuous complaints could result in a missed diagnosis of dementia or other treatable
conditions such as depression. On the other hand, paying them unnecessary attention
could lead to a burden on the health care system. In addition, an absence of complaints
when people are in the preclinical stages of memory loss can lead to a failure to diagnose
and intervene. In people who already have been diagnosed with dementia, complaints
provide clues of a person’s awareness of their disease status, which can have implications
for safety, treatment, and treatment adherence (Pachana & Petriwskyj, 2006). Moreover,
there has been a recent call for considering the perspective of the person with dementia
(Downs, 2005; Droes, 2007). Thus, determining when complaints are accurate is
important for a number of clinical reasons.
Estimates of memory complaints in population-based samples of older adults over
the age of 65 range from 22 to 56% (Jonker, Geerlings, & Schmand, 2000). However,
given estimates of actual cognitive decline in community samples (as low as 7.5% for
cognitive impairment without dementia, Di Carlo, Baldereschi, Amaducci, Maggi,
Gigoletto, Scarlato, et al., 2000), it is unlikely that all of those individuals are indeed
experiencing clinically significant memory problems. In addition, some of the people not

2
reporting complaints may in fact be experiencing memory decline. Determining when
and whether these complaints are accurate has proved a daunting task for researchers.
Taken as a whole, the literature on the accuracy of memory complaints, as
measured by the association between complaints and actual memory performance, has
produced mixed results. Here, accuracy will be described as a correct association
between cognitive complaints and cognitive performance/decline. In the following
review, I will describe the cross-sectional and longitudinal literature on complaint
accuracy and awareness in older adults. First, I discuss the different questions asked by
two distinct literatures on subjective memory complaints in normal older adults and
awareness/insight of cognitive difficulties in dementia. Next, I discuss the different
methodologies used by these studies. I will then review studies that examine whether
past decline predicts future complaints, as well as research on complaints predicting
future decline or dementia. In addition, I will address the potentially moderating roles of
dementia status and depressive symptoms on the relationship between memory
complaints and performance. Finally, I will briefly examine the relative accuracy of
complaints in cognitive domains other than memory.
The Complaints and Awareness Literatures
Overall, memory complaints are most often studied in cognitively-normal older
adults. Comparatively speaking, memory complaints are less studied in older adults with
dementia, and specifically Alzheimer’s disease. However, the phenomenon of
fluctuating insight into difficulties has been described in a separate field of study which I
call the awareness literature. Though the awareness literature has methodological and

3
conceptual differences with the complaints literature, the overlap bears relevance to the
present study.
The differences between the memory complaints and awareness literatures are
summarized here. One important difference is that awareness studies tend to have a more
impaired sample than complaints studies and include people who have been diagnosed
with dementia and are often in more advanced stages of the disease (e.g., Derouesne,
Dealberto, Boyer, Lubin, Sauron, Peitte, Kohler, et al., 1999). Indeed, most complaints
studies only examine cognitively normal participants, with fewer samples including
people with subclinical cognitive impairment (but not dementia) (for review, see Reid
and MacLullich, 2006). Very few complaints studies include people with dementia, and
those that do tend to include people in the earlier stages of the disease (e.g., Wong, Lam,
Lui, Chiu, Chan, & Tam, 2006). On the other hand, awareness study samples are
composed primarily of people with AD (for review, see Aalten, Van Valen, Clare,
Kenny, & Verhey, 2005). Because their sample only has AD patients, the range of
dementia severity is large and overall tends to be worse. For example, one awareness
study had a mean MMSE score of 16 in their AD participants (e.g., Zanetti, Valotti,
Frisoni, Geroldi, Bianchetti, Pasqualetti, & Trabucchi, 1999), and yet another included
patients with mild, moderate, and severe dementia (Migliorelli, Testa, Sabe, Petracca,
Petracchi, Leiguarda, et al., 1995). On the other hand, a number of complaints studies
with non-cognitively-normal participants report mean MMSE scores in the range of 22 to
26 (Jonker et al., 2000). These latter scores thus reflect less impaired samples, and
perhaps samples with a restricted range of cognitive impairment. Such differences in

4
MMSE score levels and ranges increase the difficulty of combining results from the two
literatures.
The two research areas also measure their variables of interest differently.
Specifically, complaints are often measured by asking a simple question about deficits or
with a questionnaire aimed at presence or frequency of cognitive problems (e.g.,
Schofield, Marder, Dooneief, Jacobs, Sano, & Stern, 1997; Gagnon, Dartigues, Mazaux,
Dequae, Letenneur, Giroire, et al., 1994). Awareness, on the other hand, is measured in
one of three ways: 1) a caregiver and patient complete questionnaires about the patient’s
abilities and the discrepancy between the two is calculated (e.g., Starkstein et al., 1996);
2) the clinician estimates the patient’s awareness after an interview and testing session, or
after separate interviews with the caregiver and patient (e.g., Arkin & Mahendra, 2001);
and 3) the patient estimates their ability to perform a certain cognitive task, then performs
the task, and the discrepancy is quantified (e.g., Dalla Barba, Parlato, Iavarone, & Bollert,
1995). Studies using this final method will be described below because they are most
similar to the complaints literature. An additional difference between the two literatures
is that awareness studies often focus on the absence of complaints when there are indeed
problems, and the memory complaints literature often focuses on the presence of
complaints when there are no problems, or when the problems have not yet become
detectable. This difference exists because of the samples chosen, as people with AD
already have problems and are expected to decline further, whereas cognitively normal
individuals do not yet have cognitive impairment and thus the emphasis is on the early

5
detection of cognitive problems. Despite these differences, the two literatures both
attempt to determine one’s subjective experience of cognitive change.
Unlike most other studies, the present study includes both normal older adults and
older adults with a diagnosis of Alzheimer’s disease using the same methodology to
estimate accuracy of complaints and the predictive power of complaints. Because of the
combined sample, I will describe below the findings of both the memory complaints and
awareness literatures with regard to dementia status.
Cross-sectional and Longitudinal Methodology
One important distinction in this research field is the difference between cross-
sectional studies and longitudinal studies. Cross-sectional studies examine how people
estimate their current impairment and correlate these complaints with current cognitive
performance (e.g., Jungwirth, Fischer, Weissgram, Kirchmeyr, Bauer, & Tragl, 2004;
Lam, Lui, Tam, Chiu, 2005). Longitudinal studies examine the association of current
complaints with either past or future cognitive decline.
The former types of longitudinal studies examine whether people’s complaints are
associated with cognitive decline from a previous time point (Dufuoil, Fuhrer, &
Alperovich, 2005; Jorm, Christensen, Korten, Jacomb, & Henderson, 2001). Estimating
the relation of complaints to decline from previous cognition levels is an important
method of determining accuracy of complaints. This is especially the case when
complaints questions are asked with reference to a specific time point (i.e., “Have you
noticed a change since last year?”, Jorm et al., 2001). This method of questioning is
perhaps the most conducive to accurately measuring perceptions of cognitive change

6
because it references two concrete time points that are measured. Additionally, it is a
question that could be asked clinically and hence, may be relevant during the diagnostic
process. Unfortunately, longitudinally examining decline from a previous time is rare
because memory complaints are most often measured at the initial time point, and as
such, there is no previous measurement of cognitive performance to which one can
compare current performance. Cross-sectional studies instead assume that decline is
present when performance is below expected levels. Thus, these two types of studies are
essentially examining a similar phenomenon: current complaints about present
performance levels compared to a previous state. The previous state is either unmeasured
in the cross-sectional studies and based on population or study performance norms, or
measured in the longitudinal studies and based on one’s own previous performance.
The third type of study, longitudinal designs examining future cognitive decline,
have examined the ability of complaints at one time point to predict either future decline
on specific neuropsychological tests or screening instruments (St. John & Montgomery,
2002; Wang, van Belle, Crane, Kukull, Bowen, McCormick, et al., 2004), or
alternatively, to predict incident dementia (Schmand, Jonker, Hooijer, & Lindeboom,
1996). This type of design is undertaken because of the possibility that
neuropsychological tests are not sensitive indicators of poor cognition at a single time
point. Specifically, some researchers have hypothesized that scores on cognitive tests
may not correspond to complaints because people with complaints are able to detect
cognitive changes before the tests can (Reisberg, Gordon, McCarthy, Ferris, & deLeon,
1985). Thus, in these types of studies, decline is usually the dependent variable and

7
complaints are often the predictor variable. Often, participants without cognitive
impairment at the first time point are included. Sometimes, samples also include
participants without dementia but with cognitive impairment.
Complaints and Past Cognitive Decline
In this section, I will discuss both cross-sectional studies of complaints (with
implicit, unmeasured change in cognition) as well as longitudinal studies examining
complaints and measured past cognitive decline. Within this section, I will separately
discuss findings in people who have normal cognition and in people who have cognitive
impairment or dementia.
Cognitively normal older adults.
Although there is some inconsistency in the field, research generally shows that
cognitively normal, non-demented individuals are generally able to accurately identify
their cognitive problems (for review, see Reid and MacLullich, 2006). Cross-sectionally,
a number of community-based samples have found significant associations between
complaints and performance in normal samples (Gagnon, et al., 1994; Bassett & Folstein,
1993; Jonker, Launer, Hooijer, & Lindeboom, 1996; Lam et al., 2005; but see Jungwirth
et al., 2004). Specifically, more memory complaints are associated with worse
performance on a range of measures. However, other studies have failed to find such
associations (Jungwirth et al., 2004; Tsai, Green, Benke, Silliman, & Farrer, 2006;
O’Connor, Pollitt, Roth, Brook, and Reiss; 1990). The mixed results could be due to
inconsistency in diagnostic group between studies, as many researchers rely on brief
cognitive screens to determine whether someone is cognitively normal or impaired, rather

8
than using a diagnosis made by consensus of clinicians. Thus, it is possible that some of
the failure to find an association between memory complaints and performance could be
due to incorrect diagnostic grouping. Therefore, with some exceptions, normal
individuals are relatively accurate at identifying their memory problems in cross-sectional
studies.
Very few longitudinal studies have examined how memory complaints reflect
cognitive decline from a previous point in cognitively normal samples. Three studies that
included cognitively normal older adults found that people’s complaints about past
change can be accurate (Johannson, Allen-Burge, & Zarit, 1997; Dufuoil et al., 2005; and
Jorm et al., 2001). One study found such a relationship only after controlling for
depression, age, gender and education (Wang, Wang, Fuh, Teng, Liu, Lin, et al., 2000).
Thus, it seems that, overall, people who have normal cognition are generally able to
accurately report on current and past decline.
Cognitively impaired or demented older adults.
Relatively few cross-sectional complaints studies examine people with cognitive
impairment or dementia. The studies that do tend to include people who have subclinical
impairment or mild dementia. Cross-sectional studies examining accuracy in people
specifically with early dementia or Mild Cognitive Impairment, a type of subclinical
cognitive impairment that is worse than that found in normal aging but does not interfere
with daily functioning, (MCI; Petersen, Smith, & Waring, 1999) will be discussed here.
Cross-sectional studies in the memory complaints literature that examine older
adults with MCI, questionable dementia, or early dementia are generally inconsistent

9
regarding accuracy of complaints. Two studies found accurate reporting by people with
cognitive impairment. Lam et al. (2005) looked at normal, MCI, and AD patients. In the
entire sample, there were significant correlations of memory complaints and performance
on an immediate recall memory test. They did not look directly at the effect of the
different diagnoses. Even after excluding AD participants from their analyses, the
correlations with immediate recall remained about the same. O’Connor et al. (1990)
found that in people with mild dementia (MMSE = 19.4), memory complaints were
associated with poorer cognitive performance for people in the dementia groups (r = .27).
However, their normal group did not have a significant complaint-performance
correlation. The absence of a relationship could be because this study’s normal group
included people who would typically be considered impaired according to their MMSE
scores. Hence, this study provides very mixed results about the accuracy of people with
varying levels of cognitive impairment.
Two other studies have found differing results. Wong and colleagues (2006)
noted that in their sample of normal, questionable dementia (CDR 0.5), and mild
dementia (CDR 1) participants, memory complaints were not associated with
performance on memory tasks (regardless of whether groups were analyzed separately or
together). Another small study also reported no correlation between complaints and
performance in their dementia group (n = 10; Feehan, Knight, and Partridge, 1991); this
failure to find significance could be a power issue. Further, one research group reported
that their Age-Associated Cognitive Decline (AACD) sample had complaints that were
not associated with actual performance (Kliegel, Zimprich, & Eschen, 2005). Thus,

10
cross-sectional studies with cognitively-impaired patients report mixed findings regarding
memory complaint accuracy. It is possible that some of these inconsistencies could
reflect differences in diagnoses (e.g., MCI versus AACD) or cutoffs for cognitive
impairment.
Because very few studies have examined accuracy of memory complaints in
samples with individuals with diagnosed dementia specifically, I will discuss the
literature on awareness. As noted earlier, this literature is conceptually similar to the
complaints literature but primarily includes samples with diagnoses of AD. Most
awareness studies include people who are more impaired than those in the complaints
literature (see review by Aalten et al, 2005). Many performance-prediction awareness
studies calculate an awareness score without giving the patients’ report separately
(Migliorelli, et al., 1995; Howorth & Saper, 2003; Barrett, Eslinger, Ballentine, &
Heilman, 2005), thus limiting their relevance to the present review. The few studies
reporting actual patient estimations of cognitive problems have found that AD
participants are poor estimators (Dalla Barba et al., 1995; Graham, Kunika, Doody, &
Snow, 2005; Duke, Seltzer, Seltzer, & Vasterling, 2002). The patients tended to rate their
abilities similarly to normal controls. However, AD patients had worse memory
performance, resulting in a poorer association between memory estimation and
performance. Interestingly, although none of the aforementioned studies reported the
number of people who underestimated their memory performance (as would be reported
in the memory complaints literature), Graham et al. (2005) mentioned that one-third of
normal controls, over half of people with mild-to-moderate dementia, and 93% of

11
individuals with moderate-to-severe dementia overestimated their memory performance.
Overall, these cross-sectional awareness studies seem to indicate poor accuracy of
cognitive perceptions in people with dementia.
Longitudinal studies of awareness using methods similar to those in the
complaints literature are essentially non-existent, and longitudinal complaints studies
generally do not include people with more severe dementia. One possible reason for the
scarcity of longitudinal studies is that the progressive nature of the disease could result in
high dropout rates. Thus, I now discuss the literature on cognitive complaints and future
decline or incident dementia.
Complaints and Future Cognitive Decline
Cognitively normal older adults.
Longitudinal studies of future decline indicate that normal older adults are
relatively good predictors of their own decline (St. John & Montgomery, 2002; Wang et
al., 2004). In Reid and MacLullich’s review (2006), ten studies with subsamples of
baseline cognitively normal participants found a significant association between current
memory complaints and future decline or dementia. Five other studies found no such
association with decline. Reid and MacLullich suggest that these discrepancies among
studies may be due to a number of factors. For example, the methodological differences
among the studies, including the various methods for measuring memory complaints
[e.g., a single complaints question used by Dik et al. (2001) and Geerlings et al. (1999)
versus the Memory Functioning Questionnaire used by Smith et al. (1996) and Taylor et
al. (1992)] could create variability in what is actually measured. In addition, mood

12
variables such as depressive symptoms could result in a greater number of complaints
and thus confound the relationship of complaints and future decline. This important issue
will be addressed below. Finally, some studies have included smaller samples (Flicker,
Ferris, & Reisberg, 1993, and Taylor et al., 1992, both with total sample sizes under 55),
which could result in a failure to produce significant findings. Thus, after considering
these variables, it is likely that complaints in baseline normal people predict their future
decline.
Cognitively impaired or demented older adults.
Researchers have also attempted to longitudinally measure the predictive power
of memory complaints for future cognitive decline or dementia in people with subclinical
impairment without dementia at baseline. Again, there is inconsistent evidence of a
relationship between memory complaints and future performance in people with baseline
impairment without dementia. Schofield et al. (1997) found that people with baseline
impairment and complaints had a greater decline in memory at one-year follow up than
did both people who complained and had normal cognition and people who did not
complain. Crowe and colleagues (2006) examined 55 MCI patients. Their single
question assessing memory complaints predicted future decline on MMSE over 2 years.
Interestingly, questionnaires with multiple items of memory complaints were not
significantly correlated with MMSE decline in the same sample. By contrast, another
study found that in the group with cognitive impairment at baseline, complaints did not
significantly predict future decline (Geerlings et al., 1999). Another study first found that
people with complaints at Time 1 but not Time 2 had a greater incidence of AD at Time 2

13
(Kim, Stewart, Kim, Yang, Shin, & Yoon, 2006). However, after excluding the people
with baseline cognitive impairment, they found that this significant association
disappeared. This might suggest that people who are in the early stages of impairment at
one point are able to accurately identify their problems. One possible reason for the
discrepancies between findings may be due to the different methods of identifying
impairment and measuring complaints. Overall, longitudinal studies of people with
baseline impairment indicate some discrepancies, but with a possible trend toward
suggesting that memory complaints accurately predict future decline. However, similar
methodology has not been studied in people with diagnosed dementia.
Summary So Far
In summary, given both the complaints and awareness literatures, it seems that
level of cognitive impairment affects the accuracy of memory complaints. Overall,
people with normal cognition (and perhaps people in the very early stages of cognitive
impairment) appear to have complaints that are more highly associated with both past and
future cognitive performance and/or decline. People with baseline cognitive impairment
seem to have inconsistent or less strong associations of complaints and past or future
performance. Because this sample includes people with diagnosed dementia, and since
the awareness literature generally suggests that people with dementia are less accurate, I
propose that dementia status moderates the link between complaints and performance.
Theory supports this proposition, as it has been suggested that as frontal lobe
functions decrease, a person loses some degree of awareness of his or her cognitive state
(for review, see McGlynn and Kaszniak, 1991). In AD, frontal lobe functions start to

14
decline in the early-to-middle stages of the disease, with progressive decline into the later
stages (Braak & Braak, 1991). The degree to which awareness is associated with or is the
same construct as complaints is unknown. Therefore, it is possible that people who are
worse cognitively will have worse accuracy. However, it is also likely that dementia
status does not account for all of the relationship between complaints and performance.
In the next section, I will discuss the findings with regard to depressive symptoms as a
moderator.
Depression and Its Impact
In the memory complaints literature, greater depressive symptoms are often
thought to affect either complaints or the association between complaints and
performance. However, many of the above studies have not adequately addressed the
effect of depressive symptoms on the association between complaints and performance.
There is robust evidence that depression and memory complaints are positively
correlated (Wong et al., 2006; Geerlings et al., 1999; Johansson et al., 1997; Jorm et al.,
2001). People who have more depressive symptoms generally have more complaints.
Some studies deal with depression by statistically adjusting for it in their analyses
(Blazer, Hays, Fillenbaum, & Gold, 1997; Geerlings et al., 1999; Schofield et al., 1997;
Dik et al., 2001; Graham et al., 2005). In both normal and impaired samples, these
studies find similar associations between complaints and cognitive performance before
and after controlling for depression, suggesting that accuracy of complaints does not only
reflect depression. An interesting question is whether depression would moderate the
association between complaints and performance, since controlling for depression does

15
not provide adequate information regarding its moderating effect. Depression might
moderate this relationship as it can produce negative cognitions which highlight negative
events or filter out positive information (Beck, 1963). Very few studies have looked at
the actual moderating effect of depression on these relationships.
In cognitively normal older adults, depression appears to affect the association
between memory complaints and performance. In a cross-sectional study, a group of
cognitively normal, depressed older adults complained more than a cognitively normal,
non-depressed group, even though they did not differ on measures of cognition (Feehan
et al., 1991). Thus, the normal, non-depressed group was more “accurate” and depression
was a moderator between complaints and performance. One awareness study examined
the effect of depression on awareness in cognitively normal people (Dalla Barba et al.,
1995). They found that the cognitively normal depressed group had worse awareness
than non-depressed normal controls. These results suggest that depression does indeed
correspond with decreased accuracy in people without cognitive impairment.
In addition, two longitudinal studies found that only people with less depression
or no depression had an association between memory complaints and future dementia
(Kim et al., 2006; Schofield et al., 1997). Because baseline cognitive impairment levels
were not considered in these two studies, it is unclear whether subsamples of more or less
impaired people would have differences in association. Thus, greater depression at the
time of complaints measurement was related to less accuracy.
None of the above studies included people with diagnosed AD at baseline.
However, awareness studies (using dissimilar methods of awareness measurement) have

16
reported on the correlation of depressive symptoms and awareness in AD patients (see
Aalten et al., 2005). Greater depressive symptoms are frequently associated with greater
awareness (Smith, Henderson, McCleary, Murdock, & Buckwalter, 2000; Starkstein et
al., 1996, Harwood, Sultzer, & Wheatley, 2000). Thus, it is likely that for people with
dementia, depression is associated with greater subjective complaints.
As noted earlier, one possible theory behind these relationships is that depression
is associated with the presence of negative cognitions. These negative cognitions can
result in an increased focus on negative aspects of one’s life or thinking abilities (Beck,
1963). Depressed individuals also are less able to inhibit negative information (Lau,
Christensen, Hawley, Gemar, & Segal, 2007). For both cognitively normal and demented
older adults, greater depressive symptoms could lead to an increase in complaints.
However, for cognitively normal older adults with greater depressive symptoms, this
could result in a decrease in accuracy if cognitive deficits are minimal (i.e., an
overestimation of deficits). For people with AD, greater depressive symptoms could lead
to what appears to be more accurate perception, as people would be more likely to
complain and have deficits, simply because of the nature of the disease. Due to the
relative dearth of literature on these moderating factors in the same study, more work is
needed in this area.
Thus, in the present study, I will examine the effects of both depression and
dementia status on the accuracy of complaints. It seems that depression in cognitively
normal individuals produces greater complaints and a worse relationship of complaints
and performance or decline. By contrast, depression in AD individuals seems to produce

17
greater complaints but a better relationship between complaints and cognitive
performance.
Other Cognitive Domains
The previous sections review accuracy of memory complaints and performance.
Relatively few studies have examined complaints of cognitive domains other than
memory. The few studies that do ask about other domains include questions about
“executive functions” such as planning, organizing, and concentrating (“Lately, I have
more difficulties in planning a journey or other undertakings,” “Lately I find it difficult to
concentrate on a task”; Zimprich, Martin, & Kliegel, 2003) and language (“Do you ever
have difficulty in finding particular words?” Schmand et al., 1996). However, these
studies aggregate the complaints into one total complaint score instead of analyzing them
by domain (Newson & Kemps, 2006; Kliegel et al., 2005; Gagnon et al., 1994; Derousne
et al.; 1993, Rabin, Roth, Isquith, Wishart, Nutter-Upham, Pare, et al., 2006; Barrett et
al., 2005). As a result of this aggregation, there has not been a separate analysis of the
association of specific complaints with their related past, current, or future
neuropsychological performance. It is possible that older adults have more accurate
perceptions of certain cognitive changes than others, but this question has not previously
been asked empirically. As such, there has been a recent call to examine complaints from
different cognitive domains separately (Reid & MacLullich, 2006).
There is very little reported in any of the cognitive complaints studies with regard
to domain-specific complaints. For example, Schmand et al. (1996) merely gave
frequencies of domain-specific complaints, reporting that in their non-demented sample,

18
memory problems and word-finding problems were both endorsed in almost 40% of their
sample, whereas approximately 8% endorsed concentration problems. Because their
complaints scale was combined into a total score, domain-specific comparisons to their
respective tests are not available. Rabin et al. (2006) examined executive function
complaints in normal and MCI participants. They found that MCI patients were less
accurate at predicting their future executive abilities because they overestimated
problems compared to performance. In fact, none of the executive complaints in either
group matched up with executive performance. No other studies reported any domain-
specific findings.
Although it is unknown how people will respond to separate self-report questions
about their cognitive abilities, some information about normal aging is known. Generally
speaking, compared to younger adults, normal cognitive aging involves decreased
performance on tests of visual memory, word fluency, and visuospatial skills, but
cognitive domains are relatively preserved otherwise (Lezak, Howieson, & Loring,
2004). However, these decreases in performance are not large enough to be in the
clinically-impaired range.
In AD, the early part of the disease process is characterized by cell loss in the
medial temporal lobes, including the hippocampus (Braak & Braak, 1991). The primary
function typically associated with the hippocampus and temporal lobes is memory, and as
such, memory difficulty is one of the earliest symptoms of AD. The typical progression
of decline on neuropsychological tests in people with AD begins with decreases in
memory, category fluency, and confrontation naming (for review, see Salmon & Bondi,

19
2009). As the disease progresses, the prefrontal and parietal areas are affected (Braak &
Braak, 1991). Patients experience declines in executive functions (such as attention,
complex mental tracking, divided attention, self-monitoring, behavioral inhibition) and
visuospatial abilities. In the present study, only the executive abilities of
attention/concentration were examined.
Because this question about domain-specific complaints and performance has
never been tested before, it is unknown what results are likely to be found. Thus, I
hypothesize that individuals will show a consistent pattern of accuracy. In other words,
accuracy in the memory, attention, and language domains will be similar. This study is
an important first step in examining these exploratory hypotheses.
The current study
The current study aimed to determine how dementia status and depression affect
the accuracy of complaints when compared to past and future cognitive performance.
The sample came from a longitudinal dataset of cognitively normal and cognitively
impaired (i.e., Alzheimer’s disease) older adults from the USC Alzheimer’s Disease
Research Center. The study differs from previous work in a few important ways.
Unlike most previous studies examining cognitive complaints, this study asked
specifically about retrospective perception of change from a specific time point. Many
other studies ask questions such as “Do you have problems with your memory?” (Dik et
al., 2001) or “Overall, do you feel you can remember things as well as you used to?”
(Jorm et al., 1997). Some of the discrepancies in prior studies could be due to how the
complaints questions have been asked. Participants could be responding based on

20
different interpretations of the question, including comparing themselves either to how
they used to perform or to others their own age. Prior studies generally have not asked
specifically and thus, participants within each study could be responding based on their
different interpretation of the question. In the present investigation, having a specific
time point of reference will likely restrict all participants to respond to the same question,
leaving little room for individual interpretation.
In addition, the complaints measure used in this study asked about subjective
change in multiple cognitive domains instead of only asking about memory complaints.
No literature to date has examined domain-specific complaints as they relate to past or
future performance on corresponding cognitive tests. Indeed, Reid and MacLullich
(2006) called for the future analysis of non-memory complaints. It is unknown whether
people are better estimators of their abilities in certain cognitive areas.
Further, the present study examined differences in clinically diagnosed normal
controls and participants with Alzheimer’s disease. Many other studies rely on brief
cognitive screens or assume community residents are normal in determining who fits into
their normal group (e.g., Dufuoil et al., 2005; Dik et al., 2001). Doing so could
unknowingly include people with preclinical dysfunction. All participants in this study
have been thoroughly tested and diagnosed by consensus of multiple clinicians, thus
increasing the reliability of diagnostic categories.
Using difference scores to calculate cognitive change from year to year, as well as
both logistic and linear regression, I tested the following hypotheses:
Hypothesis 1: Dementia status will moderate the association between complaints

21
and past change in memory. Cognitively normal participants will be relatively accurate.
AD participants will be less accurate.
Hypothesis 2: Dementia status will moderate the association between complaints
and future decline in memory. AD participants will be less accurate than cognitively
normal participants.
Hypothesis 3: People with higher levels of depressive symptoms will be more
likely to endorse complaints. Higher levels of depressive symptoms will result in worse
accuracy of complaints for cognitively normal individuals and better accuracy of
complaints for people with dementia, for both past decline and future decline.
Hypothesis 4: Though evidence is sparse, it is expected that accuracy will be
consistent across memory, attention, and language domains.

22
Chapter Two. Methods
Participants and Procedure
Participants included individuals enrolled in the longitudinal Alzheimer’s Disease
Research Center (ADRC) study at the University of Southern California. The study has
data from 1984 to the present and has followed people with normal cognitive function
(NC) and with Alzheimer’s disease (AD) and other types of cognitive impairment over
time. After the consent process and an initial intake interview, participants completed a
standard neuropsychological assessment with multiple cognitive tests (see below), as well
as a neurological examination. The cognitive assessments were performed either in the
ADRC office at USC or in the participants’ homes. Doctoral-level clinical psychology
graduate students and clinical psychologists carried out these typically two-hour
assessments. Participants were tested approximately every year. However, some
younger participants with normal cognition were administered the battery at longer
delays (between two and five years). Exceptions to these rules were made on a case-by-
case basis. Mean delay between time 0 and time 1 testings was 13.42 months for the AD
group and 15.96 months for the NC group. Mean delay between time 1 and time 2
testing was 12.89 months for the AD group and 16.38 months for the NC group.
Participants were included in the present study if they 1) were asked the
subjective complaints questions (used only in the neuropsychological battery from 1996
to 2005; described below), 2) had at least 2 neuropsychological data time points (one at
the time of complaint measurement, and one before and/or after the time of complaint
measurement), and 3) had either normal cognition or AD during their participation in the

23
study. Participants were excluded if they had diagnoses of stroke, age-associated
memory impairment, vascular dementia, possible AD (i.e., dementia possibly due to AD,
but possibly resulting from other etiologies), Parkinson’s disease, and frontotemporal
dementia. In the ADRC study, probable AD diagnoses were determined according to the
NINCDS-ADRDA diagnostic criteria (McKhann, Drachman, Folstein, Katzman, Price, &
Stadlan, 1984). Diagnoses required agreement at a consensus meeting that included the
evaluating neurologists, psychologists, and nurses. The diagnosis included for this study
was the last diagnosis assigned.
From the complete ADRC dataset (up to 2006), 1194 individuals had been
enrolled in the study. The sample available for this study was initially diminished
because 840 participants were enrolled before 1996 (the first year the complaints question
was asked, and thus a necessary time for the current study). Of these 840, only 104 had
time points up to or after 1996. With these 104 as well as the 354 individuals enrolled
between 1996 and 2005 (combined n = 458), 252 had at least two time points within the
time frame of the study. Of the 252 who had two time points, 103 did not have adequate
testing measures (only one measurement on all three cognitive tests) or did not have
adequate complaints measurement (e.g., complaints questions were not administered for
any of the three domains).
Thus, a total of 149 participants fit the above-mentioned study criteria. Two age
outliers were removed (< 50 years and > 100 years) and 15 MCI participants were
removed (see Results section below for further details). Thus, a total of 132 participants
had either Alzheimer’s disease (AD) or normal cognition (NC). Of these, 111 individuals

24
had at least two measurements of language, 107 had two measurements of memory, and
99 had two measurements of attention performance (at the time of complaints
measurement and either before or after complaints were measured). The number of
participants used in each analysis is described in Table 1. Participants were excluded
from the analyses for the particular hypothesis if they did not have complete data for
either complaints, neuropsychological measure, or depression.
Table 1. Number of participants with complete data by hypothesis.
Memory Attention Language
Past change predicts complaints 102 95 102
Complaints predict future change 97 98 94
Depression and past change
predicts complaints
98 93 97
Depression and complaints
predict future change
95 97 93

Choice of Time Points
For participants with three time points, Time 1 was chosen as the first time point
in which complaints were measured and in which neuropsychological data was available
for at least one time point before (T0), at (T1), and/or after (T2) the time of complaint
measurement. In determining which time point would be T1, I attempted to maximize
available data. Thus, if there was a time point with valid data (complaints and
neuropsychological data) for two of the three domains, but there was more complete data
one or two years later, I chose that later time point as T1. Participants with only two time
points were included for the past decline analyses (Hypothesis 1) if they had
neuropsychological data before the time of complaint measurement. Similarly,

25
participants with only neuropsychological data after complaint measurement were
included for the future decline analyses (Hypothesis 2).
Measures
Demographic characteristics. Demographic characteristics such as age, highest
level of education, gender, and ethnicity were recorded at the intake interview and will be
described below. Because both age and education can affect one’s cognitive
performance, these variables were included as covariates in the analyses.
Complaints questionnaire. Beginning in 1996, all participants were asked
whether they noticed any changes since their previous testing in general and then in the
domains of a) attention/concentration, b) language (word-finding, speech, pronunciation
of multi-syllabic words), and c) memory (See Appendix A). Responses to the general
question were re-categorized into the other three domains when appropriate. Rather than
making a composite measure of cognitive complaints, each of these questions was treated
as a separate variable for this study. Geerlings et al. (1999) demonstrated the validity of a
single question in the measurement of complaints. The open-ended answers were coded
(see “Analysis” section below for coding guidelines). Because responses to these
questions were not originally entered into the longitudinal ADRC database, data were
retrieved from stored paper files at Rancho Los Amigos National Rehabilitation Center in
Downey, California.
Data coding. To analyze the complaints variables, I devised a coding scheme and
then coded the responses. Codes were devised after a thorough review of the data. Seven
main codes were created: full denial of problems (i.e., absence of complaints), defensive

26
denial, partial denial/endorsement, endorsement of “age-appropriate” problem, full
endorsement, “I don’t know” responses, and non sequiturs (See Appendix B for coding
manual and examples). Responses for each of the three domains (attention, language,
and memory) were coded for each valid entry at each time point. Responses were
independently coded by two coders, both with doctoral-level clinical psychology training.
In the event of disagreements, the codes of the first author were used. Reliability was
calculated for all responses and was found to be very high (Kappas = .95 for attention,
.98 for language, and .96 for memory).
Frequencies of complaints by diagnostic group were examined. As noted above, I
initially devised seven complaints categories. The frequencies of complaints for the
combined sample of NC and AD participants are noted in Table 2. Certain codes were
infrequently used, such as “defensive denial” and “endorsement of problems as ‘normal
for age.’” Specifically, they each were coded less than 5% of the time for memory,
attention, and language complaints. The “partial endorsement” category was also coded
less frequently (i.e., from 5-14% of complaints). Thus, I condensed codes based on the
conceptual similarities of the groupings. I combined “defensive denial” responses into
the “denial” category. I also made an “endorsement” category by combining “normal for
age,” “partial endorsement,” and full endorsement. “I don’t know” and non sequitur
responses were entered as missing because they did not stand on their own as a
qualitatively and conceptually distinct category. The frequencies of people endorsing
complaints are similar to what other studies have reported (for review, see Jonker et al.,
2000).

27
Table 2. Frequencies of complaints.
Memory
T1
Attention
T1
Language
T1
Full denial/non-endorsement 62 88 71
Defensive denial/deflected 5 4 2
Partial denial/partial endorsement 16 11 18
Endorsement of problems “normal for age” 6 0 3
Full endorsement 33 26 27
“I don’t know” 1 0 0
Non sequitur 1 1 1

Dementia status/severity. The initial study proposal sought to evaluate the effect
of dementia severity (as measured by the caregiver-rated Clinical Dementia Rating scale -
CDR) on the relationships between complaints and change in test scores. After reviewing
the data, one concern was the relatively few participants in certain dementia severity
groups. Specifically, for time 1, there were 69 participants in the CDR 0 (no dementia)
group, 23 participants in the CDR 0.5 (very mild) dementia severity group, 36 in the
CDR 1.0 (mild dementia) group, 12 participants in the CDR 2 (moderate dementia)
group, and 4 in the CDR 3 (severe dementia) group (these numbers include the 15 MCI
participants). Even combining conceptually similar groups (e.g., CDR 2 and CDR3)
resulted in relatively small group numbers for many groups. As a result of these small
numbers, I decided to analyze the data by clinical diagnostic category (i.e., AD and NC).
As noted above, diagnosis of AD required consensus by neurologists, psychologists, and
other clinicians.
Neuropsychological measures. The following tests are organized into broad
cognitive domains based on categorization noted in Lezak et al. (2004).

28
Memory
Word list recall is the only type of memory task that has been used consistently
during the entire ADRC longitudinal study. However, two versions of the word list recall
task have been employed. From 1987 to 2001 the CERAD word list was used. From
2001 to the present, the CVLT-II word list was administered. Because of the longitudinal
nature of the study, the switch from CERAD to CVLT-II word list is problematic. The
tests have a number of differences (e.g., number of words, length of time between
immediate and delayed recall). Because of these differences and because of the relatively
small number of participants who switched from CERAD to CVLT-II or who have
CVLT-II scores only, CERAD score was used exclusively as the memory measure. Of
note, over 60% of the total ADRC sample has two or more CERAD time points.
CERAD (Consortium to Establish a Registry for Alzheimer’s Disease) word list
memory (Morris, Heyman, Mohs, Hughes, van Belle, Fillenbaum, et al., 1989): The
CERAD word list involves the visual presentation of one list of ten unrelated words three
times. Immediate recall is tested after each presentation. Delayed recall is tested after a
delay of approximately five to ten minutes. The total immediate recall score from the
three trials was used as the memory measure because, compared to delayed recall, it is
less susceptible to ceiling and floor effects and has a greater range of possible scores (0-
30 for immediate recall versus 0-10 for delayed recall). The CERAD word list task has a
one-month test-retest reliability of .43 for individuals with moderate dementia and .64 for
normal controls (Morris et al., 1989). It has been shown to load onto a memory factor in
a factor analysis (Morris et al., 1989).

29
Language
Category Fluency/Animal Naming (Morris et al., 1989): Category Fluency
measures language abilities, specifically semantic word generation. Individuals are
asked to produce in one minute as many words as possible belonging to a certain
semantic category. Common categories include animals (used in the present study), fruits
and vegetables, or items in a supermarket. This task usually is affected earlier than
phonemic (letter) fluency in people with Alzheimer’s disease (Fama, Sullivan, Shear,
Cahn-Weiner, Yesavage, Tinklenberg, et al., 1998). The Animals test has one-month
test-retest reliability coefficients of .76 for moderately demented people and .67 for
normal controls (Morris et al., 1989). Morris et al. (1989) also reported that this test
loaded on a language factor.
Attention/Concentration
Trail Making Test – Part A (Reitan, 1958): It is difficult to get a pure measure of
attention in neuropsychological tests (Lezak et al., 2004). Trails A measures simple
attention, visual tracking, and motor speed. In the ADRC battery, it was administered
from 1995 to 2006. It requires participants to draw lines connecting numbered circles in
order, as quickly as possible. The score given is time in seconds. Trails A has reliability
coefficients ranging from .60 to .90 (Lezak et al., 2004).
Depression.
The Geriatric Depression Scale (GDS; Yesavage, Brink, Rose, Lum, Huang,
Adey, et al., 1983) is a 30-item measure of depression. Responses are recorded as “yes”
or “no” to a number of symptoms (e.g., “Do you frequently feel like crying?” and “Do

30
you enjoy getting up in the morning?”). The total number of endorsed symptoms is
summed and higher total scores indicate greater depression. It was initially developed for
and normed on older adults, but is highly correlated with other depression measures
(Lezak et al., 2004). The GDS has good internal consistency and split-half reliability
(coefficients approximately .90; Brink, Yesavage, Lum, Heersema, Adey, & Rose, 1982).
Responses from people with dementia have been shown to be valid until the very severe
stages of dementia (Burke, Nitcher, Roccaforte, & Wengel, 1992). In addition,
sensitivity and specificity ratings for the self-rated GDS and the clinician-rated Hamilton
Depression Rating Scale were similar and high for dementia inpatients (Lichtenberg &
Christensen, 1992).
Plan of Analyses
Statistical Analyses
First, data was examined for nonlinearity, non-normality, and heteroscedasticity.
Skewness and kurtosis of data were also examined. Descriptive data analysis was
performed with SPSS 12.0 to describe the sample in terms of age, education, ethnicity,
and gender. In addition, I examined all other key variables (neuropsychological scores,
depression levels, frequency of complaint endorsement). Statistics are reported for each
of the three diagnostic groups (NC and AD, with MCI included for comparison). Group
differences in continuous variables were analyzed with independent t-tests and ANOVA,
and in categorical variables were analyzed with chi-square statistics. I used correlations
to determine the interrelationships between the variables and to assess problematic
multicollinearity.

31
For my analyses, I calculated the differences in neuropsychological scores by
subtracting the earlier score from the later score. I did this for all three variables
(memory: CERAD word list immediate recall; attention: Trails A; language: Animal
Naming). Using these difference variables, I used logistic regression to test hypotheses
involving performance predicting complaints (a categorical variable: endorsement versus
denial). I again used logistic regression when depression was added to the regression
predicting complaints. For my second hypothesis that complaints would predict future
cognitive change, I ran a two-way ANOVA. Finally, when I added depression in the
model predicting future cognitive change, I used multiple regression.
I considered having three diagnostic groups: normal control, AD, and MCI. The
MCI group proved problematic because it had few participants (n = 15). I also
considered folding the MCI group into the AD group; however, the two groups differed
significantly on most key variables (e.g., neuropsychological measures; see Table 3).
The MCI and normal groups did not significantly differ on most key variables, but I felt it
did not make sense conceptually to combine these two groups, as many researchers
believe that MCI may be a pre-clinical stage of dementia (Petersen et al., 1999; Morris &
Cummings, 2005). Thus, despite the fact that the MCI group is clinically interesting,
because so few individuals with mild cognitive impairment (MCI) fit the above criteria,
they were excluded. However, the data for MCI participants is included in the
descriptive statistic table.

32
All regression analyses were initially run with age and education included.
However, in none of the equations did either variable reach significance. Therefore, all
of the following analyses are presented without age and education.
Initial Efforts to Use Latent Difference Score Modeling
Although I proposed to take actual difference scores in my neuropsychological
data to measure change, I first tried using latent difference score models (McArdle &
Hamagami, 2001) to estimate the difference in neuropsychological performance between
time points. These models (and structural models more generally) are advantageous
because they attempt to remove error variance from the true difference scores, can
accommodate missing data, and allow for the multiple proposed relationships across
multiple timepoints (King, King, McArdle, Saxe, Doron-La Marca, & Orazem, 2006). I
used the MPlus program 5.0 (Muthén & Muthén, 2006) with the missing at random
assumption (Little and Rubin, 1987) and maximum likelihood estimation.
Table 3. Demographic statistics for sample.
NC (n=62) MCI (n=15) AD (n=70)
Age (T1) 75.74 (8.42)
a

(54-94)
79.27 (7.78)
ab

(63-92)
79.69 (7.55)
b

(63-99)
Gender (% Female) 63
a
60
a
63
a

Ethnicity (%
Caucasian)
100
a
93
a
86
a

Education (years) 15.42 (1.96)
a

(9-18)
15.00 (2.48)
a

(10-18)
13.96 (2.75)
b

(6-20)
MMSE T0 28.55 (1.40)
a

(23-30)
28.25 (1.42)
a

(26-30)
21.46 (4.87)
b

(10-30)
MMSE T1 28.56 (1.20)
a

(23-30)
27.60 (1.92)
a

(24-30)
19.03 (6.35)
b

(5-30)
MMSE T2 28.45 (1.64)
a

(22-30)
26.86 (2.14)
a

(22-30)
18.02 (6.18)
b

(4-29)
Word list memory T0 21.44 (3.47)
a

(15-28)
19.50 (3.99)
a

(13-27)
9.95 (5.16)
b

(0-26)

33
Table 3, Continued.
Word list memory T1 20.24 (3.44)
a

(12-28)
18.42 (4.19)
a

(13-25)
9.52 (5.20)
b

(0-20)
Word list memory T2 20.92 (3.67)
a

(11-27)
18.50 (3.61)
a

(14-24)
8.60 (5.71)
b

(0-21)
Trails A T0 36.26 (13.56)
a

(21-93)
46.00 (12.56)
a

(23-60)
77.65 (46.59)
b

(23-212)
Trails A T1 36.67 (14.16)
a

(21-108)
47.53 (22.76)
a

(25-120)
82.63 (59.82)
b

(21-280)
Trails A T2 34.52 (11.70)
a

(18-73)
46.29 (29.66)
a

(25-141)
109.06 (68.84)
b

(28-312)
Animals T0 20.71 (5.57)
a

(9-38)
15.56 (4.87)
b

(7-22)
9.98 (4.81)
c

(2-21)
Animals T1 19.91 (5.21)
a

(9-35)
15.50 (4.72)
b

(6-24)
8.13 (4.42)
c

(0-21)
Animals T2 20.66 (4.38)
a

(12-29)
14.64 (5.46)
b

(4-22)
7.58 (4.13)
c

(0-17)
GDS T0 3.38 (4.96)
a

(0-30)
3.64 (3.29)
ab

(0-9)
6.22 (4.51)
b

(0-20)
GDS T1 3.66 (3.77)
a

(0-19)
2.93 (4.10)
a

(0-14)
6.57 (4.81)
b

(0-23)
Groups that do not have the same subscript are significantly different from each other.
All significant ps < .05.

After initial analyses, many of the models involving group analyses did not run or
had errors that indicated that the results were likely not reliable. For example, I ran a
latent difference score (LDS) model of memory scores with a combined sample of AD
and normal controls. This combined sample model fit the data relatively well ( χ
2
(2, N
=132) = 3.34, CFI = .996, RMSEA = .07). Next I ran the LDSM model separated by
diagnostic group (AD vs. NC). The two-group model had errors indicating negative
residual variances or high correlations between variables. These errors resulted in
unreliable fit statistics. The same errors were generated in all other group models run
(memory, language, and attention univariate LDS models, as well as models testing
specific hypotheses). The most likely explanation for these errors is that there were too

34
few participants in each group for the 2-group latent difference score models (n ≤ 70).
Having too few participants could result in an unstable covariance matrix, and structural
equation models use covariance matrices to estimate model parameters and fit. Because
the reliability of these models was uncertain, I decided to use simple difference scores to
determine change in performance from time 0 to time 1, and likewise from time 1 to time
2.

35
Chapter Three. Results
Descriptive Statistics
As noted earlier, 132 total participants were included in the final analyses: 62 with
normal cognition (NC) and 70 with Alzheimer’s disease (AD). The two diagnostic
groups differed on age and education, as the NC group was younger and more educated.
The range of depression scores was zero symptoms endorsed to 23 symptoms
endorsed. Eighty-five percent of the sample scored within the normal, non-depressed
range (0 to 9 symptoms).
An analysis of the two final codes by domain and diagnosis (AD vs. NC) is in
Table 4. Overall, people with AD were more likely than normal controls to endorse
memory complaints at time 1. AD and NCs did not differ on frequency of attention and
language complaints.
Table 4. Percent dichotomous complaints by diagnostic group.
AD group NC group
Complaints Denial Endorsement Denial Endorsement χ
2
(df =2)
Memory
1
46% 54% 66% 34% 4.84*
Attention
1
68% 32% 77% 23% 1.41
Language
1
64% 36% 56% 44% 0.79
* p < .05.
Next, I examined the distributions of the neuropsychological variables (CERAD
immediate delay words recalled, Animal Naming total, and Trails A time) and their
difference scores, as well as the depression variable (GDS) at all three time points for AD
and NC participants. Most variables had acceptable skewness and kurtosis statistics, with
the exception of difference scores of Trails A at time 1 – time 0 and Trails A at time 2 -
time 1, which were significantly skewed. I attempted to log-transform the data, but they

36
remained skewed. I then Winsorized the data, bringing in 2 outliers on the time 1 to time
0 change and 2 outliers from the time 2 to time 1 change scores. Winsorizing involves
pulling in the largest observed values to the next largest value (Wilcox, 2003). Doing so
avoids removing outliers and decreasing sample size. Wilcox notes that even with 20%
of the data Winsorized in very small samples, accurate results can still be produced
(2003). With the Trials A variable, less than 5% of highest values were Winsorized.
This step decreased the skew to a less extreme level (Time 1-Time 0 Skew = 1.65, SE =
.25; Time 2-Time 1 Skew = 1.35, SE = .25). Thus, for the attention analyses, I used the
Winsorized data.
Tables 5 and 6 present correlations for neuropsychological difference scores,
depression score, age, education, and complaints (dichotomized) for AD and NC
participants separately. For the most part, change scores from time 0 to time 1 were
significantly negatively correlated with change scores from time 1 to time 2 for both the
AD and NC groups. Only the two attention change scores were not significantly
correlated in the AD group, although the correlation approached significance (r = -.23).
In addition, depression at time 0 was only significantly positively correlated with
language change score (time 1-time 0) in the NC group (r = .32). Depression at time 1
was significantly negatively correlated with memory change (time 2-time1) in the NC
group (r = -.39). Age and education were not significantly correlated with any variables
in the AD group. In the NC group, age was negatively correlated with change in memory
(time 1-time 0; r = -.45), and education was negatively correlated with attention change
(time 2-time 1; r = -.31). Finally, in both groups, memory and language complaints were

37
not significantly correlated with their respective change scores. Instead, only attention
complaints were associated with attention change (time 1-time 0) in the AD group (r =
.38). No other attention complaints and change score correlations were significant.
T-test of mean differences of change in neuropsychological performance by
diagnosis (AD vs. NC) revealed greater decline in performance in the AD group on
memory (time 2 – time 1), attention (time 1 – time 0 and time 2 – time 1), and language
(time 2 – time 1). There were no significant differences by group on memory (time 1 –
time 0) change and language (time 1 – time 0) change (see Table 7).

38
Table 5. Correlation matrix for key variables for Alzheimer’s Disease participants only.
∆Mem
1-
0

∆Mem
2-
1

∆Attn
1-
0

∆Attn
2-
1

∆Lang
1-
0

∆Lang
2-
1

GDS
T0
GDS
T1
Age Educ Mem
Cpts
Attn
Cpts
∆Mem
1-0
-
∆Mem
2-1
-.38** -
∆Attn
1-0
-.101 .07 -
∆Attn
2-1
.18 -.34* -.23 -
∆Lang
1-0
.36** -.18 -.06 .08 -
∆Lang
2-1
-.11 .35* .13 -.33 -.37* -
GDS T0 -.06 -.05 .25 .16 .03 .05 -
GDS T1 -.12 -.13 .03 .07 -.22 .08 .67*** -
Age .19 .14 .07 -.05 -.11 -.04 -.13 .01 -
Education -.26 .00 -.21 .16 -.03 -.06 -.07 .05 .08 -
Memory
complaints1
-10 .06 .05 -.19 -.06 .22 .25 .29* -.04 .01 -
Attention
complaints1
-.05 -.17 .38* -.15 .00 .05 .39** .50*** .03 -.09 .38** -
Language
complaints1
-.04 -.23 -.17 .04 .01 -.14 .31* .27* -.19 -.04 .29* .48***
*Correlation is significant at the 0.05 level (2-tailed).
**Correlation is significant at the 0.01 level (2-tailed).
***Correlation is significant at the 0.001 level (2-tailed).
Note. Complaints are dichotomous (0=denial, 1= endorsement); point biserial correlations are reported.

39
Table 6. Correlation matrix for key variables for normal controls only.
∆Mem
1-
0

∆Mem
2-
1

∆Attn
1-
0

∆Attn
2-
1

∆Lang
1-
0

∆Lang
2-
1

GDS
T0
GDS
T1
Age Educ Mem
Cpts
Attn
Cpts
∆Mem
1-0
-
∆Mem
2-1
-.43** -
∆Attn
1-0
.11 -.34* -
∆Attn
2-1
.08 .05 -.48** -
∆Lang
1-0
.17 -.06 .05 .26 -
∆Lang
2-1
.06 -.03 -.03 -.29 -.64*** -
GDS T0 .28 -.06 -.02 .04 .32* .24 -
GDS T1 -.15 -.39** .10 -.07 .13 -.19 .41** -
Age -.45** .09 .16 .1 .15 -.26 -.14 .09 -
Education .22 -.16 .06 -.31* -.20 .10 -.02 .10 -.23 -
Memory
complaints1
.19 -.20 .20 -.08 .37** -.25 -.02 .20 -.04 .06 -
Attention
complaints1
-.16 -.01 -.03 -.14 .23 -.21 -.04 .32* -.01 .22 .45*** -
Language
complaints1
.03 -.22 .15 -.08 .26 -.26 .06 .20 .09 .04 .49***.36**
*Correlation is significant at the 0.05 level (2-tailed).
**Correlation is significant at the 0.01 level (2-tailed).
***Correlation is significant at the 0.001 level (2-tailed).
Note. Complaints are dichotomous (0=denial, 1= endorsement); point biserial correlations are reported.
40

Table 7. Mean neuropsychological difference scores by diagnostic group.
NC AD
Change memory (T1-T0) -1.19 (3.46) -.72 (3.62)
Change memory (T2-T1) .45 (3.57)** -1.51 (3.17)
Change attention (T1-T0) 1.77 (12.86)* 11.07 (30.33)
Change attention (T2-T1) .26 (11.33)** 27.56 (39.93)
Change language (T1-T0) -.77 (5.20) -1.90 (3.33)
Change language (T2-T1) 1.00 (4.81)** -1.40 (3.11)
* p < .05.
** p < .01.

Memory

Hypothesis 1 (past memory change predicts memory complaints).
Logistic regression analyses in SPSS was used to test the first hypothesis (i.e.,
whether change in memory from time 0 to time 1 predicted complaints – denial vs.
endorsement). The dependent variable was memory complaints at time 1. Independent
variables were change in memory from time 0 to time 1, diagnosis, and the interaction of
diagnosis and change in memory. The difference scores were not significantly skewed. I
then calculated the interaction of diagnosis (AD vs. NC) and standardized change in
memory. The full model with diagnosis and change score and their interaction compared
to a constant-only model was not significantly different, χ
2
(3, n = 102) = 5.74, p > .05
(see Table 8).
Hypothesis 2 (memory complaints predict future memory change).
The second hypothesis tested whether memory complaints from an earlier time
point (time 1) would predict future memory decline (from time 1 to time 2; indicating
that people are noticing change before the memory tests can adequately detect it). I ran a
two-way ANOVA using the memory difference scores as the dependent variable, and
diagnostic group, complaints at time 1, and their interaction as independent variables.
41

The only significant main effect was diagnosis, F (1, 93) = 8.83, p < .01. The AD group
had greater decline on memory neuropsychological measures (M = -1.51, SD = 3.22) than
the NC group (M = .71, SD = 3.47; see Table 9). The main effect of complaints and the
interaction of complaints and diagnosis were not significant (Complaints: F (1, 93) = .55;
Interaction: F (1, 93) = 1.67).
Table 8. Memory change and diagnosis predicting memory complaints.
95% Confidence
Interval for Odds
Ratio
Variables B Wald Test
(z-ratio)

Odds Ratio

Upper

Lower
Intercept -.10 .12
Change (time1 – time0) -.47 1.74 .63 .311 1.26
Diagnosis (0 = NC) .73 3.08 2.07 .92 4.65
Change X diagnosis -.47 2.20 1.99 .80 4.92
Endorsement as reference group.
Final model: χ
2
(3, n = 102) = 5.74, p = .13.
All ps > .05.
Table 9. Means and standard deviations for future memory change (time 2-time1) by
dichotomous memory complaints (time 1) and diagnosis.
NC AD Total
Denial 1.24 (3.32) -1.73 (2.80) -.04 (3.42)
Endorsement -.18 (3.64) -1.35 (3.56) -.91 (3.60)
Total .71 (3.47) -1.51 (3.22)
Diagnosis F (1, 93) = 8.83, p < .01.
Hypothesis 3-1 (depression and past memory decline predicting memory
complaints).
To test hypothesis three, I looked first at the relationship between depressive
symptoms and complaints in the whole sample. I examined symptoms at the time of
complaints measurement because conceptually, it seems that current depressive
symptomatology could affect one’s present estimation of cognition. It is also possible
42

that prior depressive symptoms could affect one’s cognition and complaints, as
depression has been considered a prodrome of dementia (Ownby, Crocco, Acevedo,
John, & Loewenstein, 2006). Because of this, I considered depression at time 0 in
supplementary analyses included at the end of each domain’s section.
The mean levels of depressive symptoms were 6.70 (SD = 4.92) for people who
endorsed memory complaints and 3.92 (SD = 4.04) for people who denied complaints. A
one-factor between-subjects ANOVA indicated that this was a significant difference, F
(2, 116) = 11.31, p = .001. People who endorsed memory complaints had more
depressive symptoms than people who denied memory complaints.
Next, I ran a logistic regression in which memory complaints was the dependent
variable. The independent variables included depressive symptoms, change in memory
from time 0 to time 1, and their interaction. The overall model was significant, χ
2
(3, n =
98) = 9.03, p < .05 (see Table 10). Parameter estimates indicated that only depressive
symptoms were significant predictors of endorsement of complaints. Specifically, for
every one unit increase in depressive symptoms, the log-odds of endorsing complaints
relative to denying complaints would be expected to increase by .60 units when memory
change score and the interaction of change score and depressive symptoms also entered
into the equation. In odds ratio terms, with increases of one unit of depression, an
individual would be 1.45 times more likely to endorse complaints, p < .01. Memory
change as a main effect and the interaction between depressive symptoms and memory
change were not significant parameters.

43

Table 10. Past memory change and depression predicting memory complaints.
95% Confidence
Interval for Odds Ratio
Variables B Wald Test
(z-ratio)
Odds
Ratio

Upper

Lower
Intercept .30 1.92
Change (time1 – time0) -.16 .44 .86 .54 1.36
Depression** -.60 7.12 .55 .35 .85
Change X depression .04 .03 1.04 .67 1.60
Endorsement as reference group.
Final model: χ
2
(3, n = 98) = 9.03, p < .05.
** p < .01.
I next tested a larger regression equation with the diagnosis, depressive
symptoms, and past memory change variables predicting memory complaints. The
equation was not significant, χ
2
(7, n = 98) = 13.54, p > .05 (see Table 11).
Table 11. Past memory change, diagnosis, and depression predicting memory complaints.
95% Confidence
Interval for Odds
Ratio
Variables B Wald Test
(z-ratio)
Odds
Ratio

Upper

Lower
Intercept .43 1.39
Change (time1 – time0) -.65 2.45 .52 .23 1.18
Diagnosis (NC) -.32 .45 .73 .28 1.86
Depression -.56 1.67 .57 .25 1.33
Depression X diagnosis -.14 .06 .87 .30 2.51
Change X depression -.13 .11 .87 .40 1.94
Change X diagnosis .94 3.01 2.57 .89 7.43
Change X depression X diagnosis .14 .07 1.15 .42 3.11
Endorsement as reference group.
Final model: χ
2
(7, n = 98) = 13.54, p > .05.
All ps > .05.
Effect of prior depressive symptoms (time 0) on complaints (time 1).
To explore whether past depressive symptoms predict future memory complaints
(rather than current depression correlating with current complaints), I ran the previous
analyses with depressive symptoms at time 0 instead of at time 1. First, there was no
44

significant difference in mean depression score between the individuals with complaints
(M = 5.95, SD = 5.04) and those without (M = 4.10, SD = 4.81), F (1, 96) = 3.35, p = .07.
Depressive symptoms at time 0, change in memory from time 0 to time 1, and
their interaction did not significantly predict type of complaints at time 1 ( χ
2
(3, n = 92) =
6.93, p > .05). Further, a model that included the three-way interaction (depression time
0 X change in memory time 1-time 0 X diagnosis) was also not significant ( χ
2
(7, n = 92)
= 10.90, p > .05).
Hypothesis 3-2 (depression and memory complaints predicting future decline).
To explore the moderating effect of depressive symptoms on the association
between complaints at time 1 and future decline (time 1 to time 2), I used multiple
regression analyses. Thus, the complaints variable at time 1, depression, and the
interaction of depression and complaints were entered as independent variables. Memory
change from time 1 to time 2 was the dependent variable. The overall model was
significant, F (3, 91) = 4.28, p < .01, R
2
= .12 (see Table 12). Depression was the only
significant path, with greater depressive symptoms leading to less future memory change
(B = -.39, p < .01). Neither memory complaints nor the interaction of depression and
complaints were predictors of future memory change. However, these results are
qualified by the results that arose with the addition of diagnosis.
Table 12. Memory complaints and depression predicting future memory change.
B t Sig
(Constant) -.52 .60
Depression -.39** -2.931 .004
Complaints (0 = Denial) -.07 -.65 .52
Depression X Complaints .10 .72 .47
Model: F (3, 91) = 4.28, p < .01. R
2
= .12.
** p < .01.
45

When I added dummy-coded diagnosis (0 = NC, 1 = AD) and interactions of
diagnosis with the other independent variables, the regression equation was again
significant, F (7, 87) = 3.39, p < .01, R
2
= .21 (see Table 13). Again, depression was a
significant predictor of change, with greater depressive symptoms associated with less
change, B = -.56, p < .01. However, there was a significant interaction of depressive
symptoms and diagnosis (B = .44, p < .05). The simple regression equations of diagnosis
on memory change at high, average, and low values of depressive symptoms are shown
in Figure 1 (Aiken & West, 1991). Post-hoc tests of the simple slopes revealed that in the
low depressive symptoms group, the NC participants had significantly less decline than
AD participants. In fact, the NC people with few depressive symptoms actually
improved on the memory scores, whereas the AD group did not change. For the mean
level of depressive symptoms, both the normal and AD groups declined only minimally.
Finally, in the high depressive symptoms group, normal controls had a significantly
higher amount of decline than people with AD group, who also declined but to a smaller
degree.
Table 13. Memory complaints, depression, and diagnosis predicting future memory
change (time 2-time1).
B t Sig.
(Constant) .12 .90
Depression -.56** -3.20 .002
Diagnosis (0 = NC) -.17 -1.52 .13
Complaints (0 = Denial) .00 -.03 .97
Depression X Diagnosis .44* 2.11 .04
Depression X Complaints .55 .87 .38
Diagnosis X Complaints -.08 -.64 .52
Depression X Diagnosis X Complaints -.63 -.99 .33
Model: F (7, 87) = 3.39, p < .01. R
2
= .21.
* p < .05, ** p < .01.
46

Time 0 depression predicting future cognitive change was not separately examined
because it would essentially be redundant with time 1 depression predicting future
change, which has already been analyzed above.
Attention
My final aim of this study was to examine the earlier hypotheses (#1-3) with
different cognitive domains, specifically attention complaints and language complaints
and their respective neuropsychological test scores. Thus, I repeated the above analyses
for attention and language.
Figure 1. Diagnosis moderates the effect of depression on future change in memory.
T-test of simple slopes: AD t (91) = .059, n.s.; NC t (91) = -3.20, p < .05.

Diagnosis moderating the effect of depression
on future change in memory (T2-T1)
-2
-1.5
-1
-0.5
0
0.5
1
1.5
2
2.5
-1 0 1
Standardized depression scores
Change in memory (T2 - T1)
NC
AD
47

Hypothesis 1 (past attention change predicts attention complaints).
The analyses for past attention change predicting attention complaints were
carried out using logistic regression. A model with change in attention, diagnosis, and
their interaction predicting attention complaints approached significance, χ
2
(3, n = 95) =
7.73, p = .052 (see Table 14). However, no parameters were significant.
Table 14. Past attention change and diagnosis predicting attention complaints.
95% Confidence
Interval for Odds Ratio
Variables B Wald Test
(z-ratio)
Odds
Ratio

Upper

Lower
Constant -1.21* 11.52 .30
Change (time1 – time0) -.15 .06 .86 .26 2.92
Diagnosis (NC) .34 .44 1.40 .52 3.77
Change X diagnosis .81 1.38 2.24 .58 8.63
Endorsement as reference group.
Final model: χ
2
(3, n = 95) = 7.73, p = .052.
* p < .05.
Hypothesis 2 (attention complaints predict future attention change).
To test whether attention complaints from an earlier time point (time 1) would
predict future attention decline, I ran a two-way ANOVA using the attention difference
scores (attention time 2 score minus attention time 1 score) as the dependent variable and
diagnostic group and complaints as independent variables (see Table 15). The main
effect of diagnosis was significant, such that AD participants had greater decline in
attention, F (1, 94) = 19.88, p < .001 (M-AD = 29.20, SD = 38.81; M-NC = .26, SD =
11.33). The main effect of complaints was nonsignificant, F (1, 94) = 1.74, p > .05, as
was the interaction between complaints and diagnosis, F (1, 94) = 0.53, p > .05.

48

Table 15. Means and standard deviations for future attention change (time 2-time1) by
dichotomous attention complaints (time 1) and diagnosis.
NC AD Total
Denial 1.15 (11.73) 33.23 (39.52) 14.70 (31.22)
Endorsement -2.54 (9.86) 20.57 (37.53) 9.44 (29.79)
Total .26 (11.33)
a
29.20 (38.81)
b

Different subscripts indicate difference at p < .001.
Diagnosis F (1, 94) = 19.88, p < .001.

Hypothesis 3-1 (depression and past attention decline predicting attention
complaints).
To begin, a one-factor between-subjects ANOVA indicated significant differences
in mean time 1 depression scores, with greater depression scores in people who endorsed
complaints (M

= 8.34, SD = 5.00) than in those who denied (M = 3.91, SD = 3.76), F (1,
122) = 28.80, p < .001.
Next, I ran a logistic regression to examine the moderating effect of depression on
the relationship between change in performance and complaints. As before, attention
complaints was the dependent variable, and depression, change in performance, and their
interaction were independent variables. The overall model was significant, χ
2
(3, n = 93)
= 15.02, p < .01 (see Table 16). Parameter estimates indicated that only depression was a
significant predictor of complaints. Specifically, for every one unit increase in
depression, the log-odds of endorsing complaints relative to denying complaints would be
expected to increase by .76 units when attention change score and the interaction of
change score and depression were also entered into the equation (B = .76, p < .01). In
odds ratio terms, with increases of one unit of depression, one would be 1.53 times more
likely to endorse complaints, p < .01. Attention change score as a main effect and the
interaction with depression were not significant parameters.
49

A model of diagnosis, depression, change in attention scores, the three pairs of
two-way interaction terms, as well as a three-way interaction of depression at time 1 X
diagnosis X change in attention scores (time 0 to time 1) predicting attention complaints
was significant, χ
2
(7, 93) = 17.27, p < .05 (see Table 17). However, no paths were
significant (all ps > .15).
Table 16. Past attention change and depression predicting attention complaints.
95% Confidence
Interval for Odds Ratio
Variables B Wald Test
(z-ratio)
Odds
Ratio

Upper

Lower
Intercept 1.01* 15.47
Change (time1 – time0) -.44 3.13 .65 .40 1.05
Depression -.76* 9.23 .47 .29 .76
Change X depression .01 .001 1.01 .53 1.90
Endorsement as reference group.
Final model: χ
2
(3, n = 93) = 15.02, p < .01.
* p < .05.

Table 17. Past attention change, diagnosis, and depression predicting attention
complaints.
95% Confidence
Interval for Odds
Ratio
Variables B Wald Test
(z-ratio)
Odds
Ratio
Upper Lower
Intercept 1.02*7.25
Change (time1 – time0) .21 .11 1.24 .35 4.41
Diagnosis (0 = NC) .22 .14 1.24 .40 3.90
Depression -.65 1.91 .52 .21 1.32
Depression X diagnosis -.21 .13 .81 .25 2.58
Change X depression .42 .21 1.52 .25 9.16
Change X diagnosis -.86 1.43 .42 .10 1.74
Change X depression X diagnosis -.47 .22 .63 .09 4.38
Endorsement as reference group.
Final model: χ
2
(7, n = 93) = 17.28, p = .016.
* p < .05.

50

Effect of depression (time 0) on complaints (time 1).
I again ran the same analyses with depression at time 0. A simple ANOVA
revealed that the mean depression for people who endorsed attention complaints was
higher (M = 6.40, SD = 5.34) than depression for people who denied complaints (M =
4.18, SD = 4.60), F (1, 100) = 4.48, p < .05. When I included performance and the
interaction between depression and performance, the overall model was significant, χ
2
(2,
n = 89) = 10.42, p < .05. However, none of the parameter estimates were significant.
Finally, the overall model including diagnosis and its interactions was not significant, χ
2

(7, n = 89) = 11.27, p > .05.
Hypothesis 3-2 (depression and attention complaints predicting future attention
decline).
To explore the relationship of depression on the association between attention
complaints at time 1 and future decline (time 1 to time 2), I used standard multiple
regression. Complaints (time 1), depression (time 1) and their interaction were entered as
independent variables. Attention change from time 1 to time 2 was the dependent
variable (see Table 18). The overall model was not significant, F (3, n = 97) = 2.11, p >
.05, R
2
= .06.
The model including diagnosis and the interactions with diagnosis added was
significant, F (7, n = 97) = 4.38, p < .001, R
2
= .26; see Table 19. The only significant
path again was diagnosis (B = .51, p < .001). The AD participants had greater decline
than NC participants. Thus, for attention, only diagnosis was a significant predictor of
51

decline when depression, complaints, and resulting interactions were included in the
model.
Table 18. Attention complaints and depression predicting future attention change.
B t Sig
(Constant)*** 3.95 .000
Depression .12 .89 .38
Complaints (0 = Denial) -.17 -1.57 .12
Depression X Complaints .17 1.22 .23
Model: F (3, N = 97) = 2.11, p > .05. R
2
= .06.
*** p < .001.
Table 19. Attention complaints, diagnosis, and depression predicting future attention
change.
B t Sig.
(Constant) .13 .90
Depression -.03 -.18 .85
Diagnosis (0 = NC) .51*** 4.35 .000
Complaints (0 = Denial) -.05 -.37 .71
Depression X Diagnosis -.05 -.25 .80
Depression X Complaints .06 .25 .81
Diagnosis X Complaints -.21 -1.29 .20
Depression X Dx X Complaints .21 .73 .47
F (7, 97) = 4.38, p < .001 (R
2
= .26).
*** p < .001.
Language
To continue the final study question regarding different cognitive domains and
their relationship with complaints, I examined the language domain.
Hypothesis 1 (past language change predicts language complaints).
The logistic regression equation of language performance, diagnosis, and their
interaction predicting language complaints was not significant, χ
2
(3, n = 102) = 3.57, p >
.05; (see Table 20).

52

Table 20. Past language change and diagnosis predicting language complaints.
95% Confidence
Interval for Odds Ratio
Variables B Wald Test
(z-ratio)
Odds
Ratio

Upper

Lower
Intercept .66* 5.14
Change (time1 – time0) -.49 2.93 .62 .35 1.07
Diagnosis (NL) -.16 .14 .86 .37 1.96
Change X diagnosis .45 .90 1.58 .62 4.03
Endorsement as reference group.
Final model: χ
2
(3, n = 102) = 3.57, p > .05; * p < .05.

Hypothesis 2 (language complaints predict future language change).
To test whether language complaints from an earlier time point (time 1) predict
future language decline, I ran a two-way ANOVA using the attention difference scores
(Animal Naming time 2 score minus Animal Naming time 1 score) as the dependent
variable and diagnostic group and type of complaint at time 1 as independent variables.
There was a significant main effect of diagnosis on language change, F (1, 90) = 8.37, p
< .01; see Table 21. The AD group had greater decline on memory neuropsychological
measures (M = -1.40, SD = 3.20) than the NC group (M = 1.14, SD = 4.83). In addition,
there was a significant main effect of language complaints on future language decline,
such that endorsement (M = -1.15, SD = 4.25) was associated with greater decline than
denial of complaints (M = .32, SD = 4.14), F (1, 90) = 4.06, p < .05. The interaction of
diagnosis and complaints was not significant, F (1, 90) = .78, p > .05.
Table 21. Means and standard deviations for future language change (time 2-time1) by
dichotomous language complaints (time 1) and diagnosis.
NC AD Total
Denial 2.15 (4.27) -1.09 (3.49) .32 (4.14)
Endorsement -.33 (5.32) -2.06 (2.44) -1.15 (4.25)
Total 1.14 (4.83) -1.40 (3.20)
Diagnosis F (1, 90) = 8.37, p < .01.
Complaints F (1, 90) = 4.06, p < .05.
53

Hypothesis 3-1 (depression and past decline predicting language complaints).
To begin, the mean levels of time 1 depressive symptoms was 6.17 (SD = 4.82)
for people who endorsed language complaints and 4.32 (SD = 4.22) for people who
denied language complaints. These means were significantly different, F (1, 114) = 4.80,
p < .05. Thus, people who endorsed language complaints reported more depression.
Next, I ran a logistic regression to examine the effect of depression, change in
performance (time 1 – time 0), and their interaction (all entered as independent variables)
on language complaints (the dependent variable). The overall model was significant, χ
2

(3, n = 97) = 10.65, p < .05; see Table 22. Greater depression was associated with 1.42
greater odds of endorsing language complaints, p < .05. Memory change as a main effect
and the interaction were not significant parameters.
Table 22. Past language change and depression predicting language complaints.
95% Confidence
Interval for Odds
Ratio
Variables B Wald Test
(z-ratio)
Odds
Ratio

Upper

Lower
Intercept .51 5.20
Change (time1 – time0) -.36 2.12 .69 .43 1.13
Depression -.55* 5.25 .58 .36 .92
Change X depression -.53 3.54 .59 .34 1.02
Endorsement as reference group.
Final model: χ
2
(3, N = 97) = 10.65, p < .05.
* p < .05.
A model of diagnosis, depression, change in language scores, the three pairs of
two-way interaction terms, as well as a three-way interaction of depression at time 1 X
diagnosis X change in language scores (time 0 to time 1) predicting language complaints
was significant, χ
2
(7, n = 97) = 15.41, p < .05; see Table 23. However, none of the
54

parameters were significant. Thus, it seems that this combination of variables in the
equation contributed to some portion of the overall variance in predicting complaints.
Table 23. Past language change, diagnosis, and depression predicting language
complaints.
95% Confidence
Interval for Odds
Ratio
Variables B Wald Test
(z-ratio)
Odds
Ratio

Upper

Lower
Intercept .17 .20
Change (time1 – time0) -.80 3.33 .45 .19 1.06
Diagnosis (0 = NC) .62 1.44 1.86 .67 5.13
Depression -.84 3.25 .43 .17 1.08
Depression X diagnosis .26 .20 1.29 .43 3.92
Change X depression -.81 2.77 .45 .17 1.15
Change X diagnosis 1.21 3.19 3.36 .89 12.68
Change X depression X diagnosis -.01 .00 .99 .26 3.78
Endorsement as reference group.
Final model: χ
2
(7, N = 97) = 15.41, p < .05.
All ps > .05.
Effect of depression (time 0) on complaints (time 1).
I again ran the same analyses with depression at time 0. A simple ANOVA
revealed no significant differences in mean depression between people who denied
language complaints (M = 4.16, SD = 4.83) and people who endorsed complaints (M =
5.52, SD = 4.58), F (1, 93) = 1.75, p > .05. When I included performance and the
interaction between depression and performance, the overall model was not significant, χ
2

(3, n = 94) = 6.94, p > .05. The overall model including diagnosis and its interactions
was also not significant, χ
2
(7, 94) = 12.29, p > .05.

55

Hypothesis 3-2 (language complaints and depression predicting future language decline).
To explore the relationship of depression on the association between language
complaints at time 1 and future decline, analyses using multiple regression were
employed. Complaints was entered as an independent variable, along with depression
and their interaction. Language change from time 1 to time 2 was the dependent variable.
The overall model was not significant, F (3, 89) = 1.34, p > .05, R
2
= .04, see Table 24.
When diagnosis and its interactions were added, the regression equation was
significant, F (7, 85) = 2.49, p < .05, R
2
= .17, see Table 25. Diagnosis was the only
significant predictor of language change, such that the AD group had greater decline, B =
-.32, p = .01.
Finally, all logistic regression analyses were run separately with prior
neuropsychological score included as a covariate (in addition to prior change score) to
determine whether the level of prior score affected complaints at time 1. Doing so did
not significantly change any of the above findings for memory, attention, or language.
Table 24. Language complaints and depression predicting future language change.
B t Sig.
(Constant) .45 .65
Depression -.38 -.74 .46
Complaints (0 = Denial) -.14 -1.31 .19
Depression X Complaints -.05 -.39 .70
Model: F (3, 89) = 1.34, p > .05.
R
2
= .04.
All ps > .05.

56

Table 25. Language complaints, diagnosis, and depression predicting future language
change.
B t Sig.
(Constant) 1.91 .06
Depression -.15 -.83 .41
Diagnosis (0 = NC) -.32* -2.62 .01
Complaints (0 = Denial) -.19 -1.54 .13
Depression X Diagnosis .29 1.48 .14
Depression X Complaints -.44 -1.06 .29
Diagnosis X Complaints .02 .16 .88
Depression X Diagnosis X Complaints .29 .67 .50
Model: F (7, 85) = 2.49, p < .05. R
2
= .17.
* p < .05.
57

Chapter Four. Discussion
In the present study, I examined the relationships between depression, complaints,
and cognitive performance in different cognitive domains over time in a group of older
adults with normal cognition and with Alzheimer’s disease. This study expanded on
previous efforts in that it aimed to both determine whether complaints accurately describe
cognitive change that has occurred and to examine the ability of complaints to predict
future decline in performance in a single sample of individuals. In addition, this study
sought to determine the relative difference in “accuracy” between older adults with
normal cognition and individuals with Alzheimer’s disease. The current study is also the
first of its kind to differentiate complaints by cognitive domain and relate those
complaints to cognitive performance of that specific domain. Finally, as previous studies
have found a relationship between complaints and depressed mood (Derouesne et al.,
1999; O’Connor et al., 1990), I also sought to determine whether levels of depressive
symptoms were the sole determiner of complaints, or whether and how they affected the
relationship between complaints and future or past change in cognitive performance.
Several consistent and interesting results emerged.
I will first discuss the relationships between past change in performance and
complaints for all three cognitive domains, including the effects of both diagnosis and
depression on these relationships. I will then discuss results for my second hypothesis
that complaints predict future change in performance, considering results with and
without diagnosis and depression.

58

Past Change in Performance Predicting Cognitive Complaints
First, in all three cognitive domains, complaints overall were not significantly
associated with prior change in performance. In addition, the relationship between past
performance and complaints was not moderated by diagnostic group. In other words,
participants in neither the normal or AD groups accurately complained about their actual
cognitive decline in the past year. These findings partially support my hypothesis. I
expected that people with dementia would show a poor association between complaints
and past cognitive decline, whereas cognitively normal individuals would be more
accurate.
As expected, past decline did not predict complaints in the AD group. Although
this is the first known study to examine measured past decline and complaints in a sample
with diagnosed AD participants, the cross-sectional awareness literature corroborates
these findings. In general, the awareness literature shows that people with AD are not
accurate at estimating their cognitive level, likely because the disease process affects
their ability to monitor their performance and update their self-perceptions (Graham et
al., 2005). These deficits in awareness could be due to frontal lobe dysfunction, which is
affected in the early to middle stages of AD and which is thought to be associated with
self-concept (Lopez, Becker, Somsak, Dew, & DeKosky, 1994; Mangone, Hier, Gorelick,
Ganellen, Langenberg, Boarman, et al., 1991). The frontal lobes are also responsible for
self-monitoring (Alvarez & Emory, 2006). Deficits in self-monitoring could result in
individuals failing to integrate changes into their self-concept. This could lead to a
59

failure to endorse problems when problems are actually present. In addition, the relative
inaccuracy might be due to a pure memory failure on the part on the AD participants.
People with dementia could have particular difficulty accurately remembering and
reporting on their cognitive change.
In contrast with much of the complaints literature for normal older adults
(Johannson et al., 1997; Dufuoil et al., 2005; and Jorm et al., 2001), cognitively normal
participants did not show a strong association between complaints and past decline.
There are several possible reasons for these unexpected results. Methodologically, there
could be a sampling bias in that individuals enrolling in the ADRC study could have a
higher level of concern for cognition at baseline. In contrast, the three studies that found
a general association between complaints and past decline had large epidemiological
samples that did not specifically recruit volunteers concerned about their memory (Jorm
et al., 2001; Dufuoil et al., 2005; Johannson et al., 1997). The type of sample in this
study is likely more representative of typical patients who present to physician’s offices.
Further, the normal control group in this study had a small range in difference scores, as
their decline from time 0 to time 1 was limited. This could have resulted in the absence
of a relationship between past decline and complaints at time 1.
Alternatively, these results could reflect that some people without dementia are
actually not aware of their cognitive change. Some insights can be gleaned from the
literature on emotional awareness, which describes different processes that are important
in awareness of emotional states. Some of these processes include the perception and
experience of emotion (Lane, 1998). If either of these processes is “faulty,” people could
60

fail to be aware of their internal states. Specifically, people could fail to appropriately
endorse complaints because they do not accurately perceive deficits. They also may not
appraise their level of cognition as different from before or from what would be expected.
Although an analogous model of cognitive awareness has not yet been tested, Reid &
MacLullich (2006) similarly suggested that the absence of memory complaints in people
who have cognitive change could result from an inability to remember or an
unwillingness to endorse complaints. Various other factors such as mood, premorbid
personality characteristics, and embarrassment could make individuals reluctant to
acknowledge that their cognition has changed.
Complaints might be inaccurate if they reflect negative affect rather than
cognitive change. Various studies have found associations between complaints and
anxiety (Jorm et al., 2001), depression (Jungwirth et al., 2004), and neuroticism (Kliegel,
Zimprich, & Eschen, 2005). Anxiety could produce a misperception of internal states.
Such a misperception has been hypothesized to drive panic disorder, in which one
misinterprets non-threatening bodily states as harmful (Clark, 1986). Thus, with
cognitive complaints, anxiety could produce a misperception or misinterpretation of
cognitive abilities, in which the individual complains about cognitive problems without
necessarily being accurate.
Although anxiety was not measured in this sample, I explored the impact of
concurrent and prior depressive symptoms on complaints and their relationship with past
cognitive change. First, mean depressive symptoms were within the normal, non-
depressed range. Among all domains, greater concurrent depressive symptoms were
61

associated with endorsement of complaints. In fact, when depressive symptoms, past
change in performance, and their interaction were entered into the logistic regression,
only depressive symptoms were associated with greater odds of endorsing complaints for
all three domains. This is consistent with both memory complaints and awareness studies
examining depression (Wong et al., 2006; Geerlings et al., 1999; Johansson et al., 1997;
Jorm et al., 2001; Starkstein et al., 1996). As suggested by previous studies, the
association between depressive symptoms and complaints could reflect the negative
cognitions that often accompany depression, such as thinking that one is performing
worse than one actual is or having expectations that one cannot meet and then feeling
self-critical (Beck, 1963). Alternatively, depressive symptoms could be associated with
complaints because some of the symptoms of depression include difficulties with
attention and concentration (APA, 2000). Along the same lines, the depression measure
used, the Geriatric Depression Scale, has a number of items that reflect cognitive
symptoms of depression, including poor concentration and feeling as if one’s mind is not
as clear as it used to be. Thus, the association between depressive symptoms and
complaints could be due to the overlap in items. Unfortunately, these items could not be
removed from the analyses because only the summed GDS score was available in the
dataset. In sum, current depressive symptoms, rather than past cognitive change, could
be fueling complaints.
To further explore the relationship between depression and complaints, I also
examined the effect of prior depressive symptoms on current complaints and their
relationship with cognitive decline. When I examined depressive symptoms at time 0,
62

some differences from analyses with depressive symptoms at time 1 emerged. To begin,
mean symptoms were only significantly greater in people who endorsed attention
complaints compared to individuals who denied attention complaints. There were no
mean differences in depressive symptoms at time 0 in individuals who endorsed versus
denied memory and language complaints. This could reflect the fact that people were
asked to evaluate their cognition from a prior time point in which they had higher
depressive symptoms. Because problems with attention and concentration are associated
with depression, it would make sense that people with greater earlier depressive
symptoms would recall and endorse attention and concentration problems. However, the
association between time 0 depressive symptoms and attention complaints disappeared
when other study variables were also entered into a regression equation. Thus, the
association between time 0 depressive symptoms and attention complaints is not
consistent. Since this is the first known study to examine the relationship between earlier
depressive symptoms and later attention complaints, these results should be replicated. In
sum, however, these results suggest that current depressive symptoms are more closely
associated with current complaints than are past depressive symptoms.
Cognitive Complaints Predicting Future Decline
For my hypothesis that complaints at one point would predict future change in
performance, a more complex story emerged. First, diagnosis was a significant predictor
of future cognitive change in all three domains. In particular, people with AD had greater
decline. This would be expected given the nature of the normal aging and dementia
processes, as cognition in people with dementia changes at a more rapid pace than in
63

people without dementia. Next, within a regression testing whether complaints,
diagnosis, and their interaction predict future decline, complaints did not predict future
memory or attention decline. However, language complaints predicted future language
change. The language finding will be discussed below, but first I address memory and
attention.
The absence of a relationship between memory and attention complaints and
future decline for both diagnostic groups is in contrast to what was hypothesized. The
relationship between attention complaints and future attention decline has not been
studied previously. However, several studies have examined memory complaints, with
very mixed results. Although some studies have found that complaints indeed predict
future decline or dementia (Crowe et al., 2005; Geerlings et al., 1999), other studies have
found that complaints do not predict future decline (Jorm et al., 1997; Wang et al., 2000).
Indeed, in one review, one-third of studies examining memory complaints in normal
older adults found no association with future decline or future dementia (Reid &
MacLullich, 2006). Moreover, there are essentially no studies of complaints and future
decline in individuals with AD, but in people with subclinical decline or mild cognitive
impairment, there has again been mixed evidence. Some studies show complaints are
indeed predictive of future decline in subclinically impaired individuals (e.g., Crowe et
al., 2006), but other studies do not (e.g., Geerlings et al., 1999). In sum, it appears that
memory complaints inconsistently predict future decline.
It is unlikely that the lack of association between memory complaints and future
decline in this study was due solely to methodological differences from previous studies.
64

The only prior study to examine memory complaints in reference to a specific prior time
point found that complaints did predict future decline and incident dementia (St. John &
Montgomery, 2002). Additionally, other studies with similar follow-up periods of one to
two years found associations of memory complaints and future cognitive decline
(Schofield et al. 1997; Tobiansky et al., 1995; Johansson et al., 1997). Thus, complaints
reference point and length of delay are likely not limiting the current results.
Instead, given the mixed findings in the literature, it seems probable that other
variables may moderate the relationship between memory complaints and future memory
decline. For example, levels of baseline cognitive impairment could vary widely even
within diagnostic groups. As such, individuals with greater impairment in the dementia
group may have more difficulties accurately reporting on memory decline than demented
individuals with less impairment. Future studies with adequate sample sizes should
examine the moderating effect of dementia severity. Along the same lines, some studies
do not differentiate the participants’ types of dementia. Different types of dementia have
dissimilar presentations, and awareness or the ability to introspect could be affected
differentially. For example, an individual with vascular dementia could have strokes in
areas of the brain that do not affect awareness. However, another individual might have
frontotemporal dementia, which is known for its profound impact on insight (Neary,
Snowden, & Gustafson, 1998). Further, the progression of different types of dementia
varies. Individuals with vascular dementia could have additional cerebrovascular events
in the year after complaints were assessed. Thus, for the dementias that progress in a
stepwise manner, we might not expect current complaints to predict future performance.
65

Future studies should examine whether the typical progression of various dementias
moderate the relation between complaints and future decline. Finally, various life events
could occur to impact cognition between time 1 and time 2 of the study, thus resulting in
worse performance than might be anticipated. For example, major stressful events such
as falls, delirium, and widowhood might impact one’s functioning. Therefore, an absence
of complaints at time 1 might be poorly associated with future decline if the decline was
sudden and unexpected.
With regard to language, endorsement of language complaints was associated
with later decline. This suggests that language complaints may accurately foretell
language change. Accuracy of language complaints on future language decline has never
before been studied and presents an interesting finding. One explanation for the early
(i.e., before tests can detect deficits) identification of language change is that language
difficulties may be more easily noticed by patients and others around them. Speech and
language are external behaviors that may require less internal reflection. Further, there
are potential social implications for having language problems such as word-finding
difficulties and comprehension difficulties, including feelings of embarrassment and
inability to fluently communicate. Thus, language problems may be easier to reflect
upon because they are more noticeable and more salient in everyday functioning.
This significant finding must be tempered to some extent because when
depressive symptoms, complaints, and their interaction were entered into a regression
equation predicting future language decline, language complaints was no longer a
significant predictor of decline. In addition, amount of depressive symptoms also was
66

not a significant predictor of language change. It is likely that language complaints
contributed to future decline to some degree, and because depressive symptoms and
complaints were correlated, they likely had a combined contribution to the model and
shared the overall variance of the equation. Thus, in the current sample, language
complaints appear to have a tenuous relationship with future language decline that would
need to be replicated.
One last interesting result involved the effect of depression and its interactions on
future memory change. Depressive symptoms at time 1 predicted future change in
memory. However, this main effect was qualified by an interaction of depressive
symptoms and diagnosis. Only in cognitively normal individuals did depressive
symptoms drastically affect future memory change. Specifically, a lack of depressive
symptoms was associated with actual improvement in memory, whereas high depressive
symptoms were related to greater decline in memory than even AD participants
experienced. Further, AD individuals with little to no depressive symptoms exhibited
minimal future decline in memory scores. One possible reason for these findings is that
cognitively normal people who do not endorse depressive symptoms are in general good
cognitive health and may experience practice effects from year-to-year administration of
tests. In addition, individuals with AD but without depression also did not show much
decline. It therefore appears that not having depressive symptoms may be protective
against memory decline for both cognitively normal and demented individuals.
The finding that cognitively normal individuals who endorsed high depressive
symptoms showed worse memory decline than AD depressed individuals is also
67

interesting. Researchers have found that major depression might affect hippocampal
functioning, perhaps because of elevated cortisol levels (Bremner, Narayan, Anderson,
Staib, Miller, & Charney, 2000; Von Gunten, Fox, Cipolotti, & Ron, 2000). This reduced
hippocampal functioning could in turn result in worse memory performance.
Hippocampal functioning in people with AD is already affected by the disease. Thus,
there may be comparatively less of a detrimental effect of depression on cognition in
people with AD. Future studies should examine these and other reasons why cognitively
normal depressed individuals have worse memory decline than depressed people with
AD.
Limitations and Implications
There are a number of limitations to this study. To begin, the final sample
included a relatively limited number of participants. Although 132 people were included
in the analyses, not all individuals had data for all time points and for all cognitive
domains. Thus, there are varying levels of participants included in any given analyses,
but most analyses examined approximately 100 participants. Ideally, more individuals
would have fit the inclusion criteria and would have had valid data. On a related note,
some of the regression equations had several predictors and interactions, which may have
stretched the limits of power for the model. This limitation of power could have resulted
in Type II error, or false negatives, for some of the models. With a larger sample, more
significant findings may have emerged.
In addition, it is possible that the change scores were not accurate assessments of
“true” change. Some researchers have discussed the pitfalls of using simple difference
68

scores, particularly across more than two time points (King, King, McArdle, Saxe,
Doron-Lamarca, & Orazem, 2006). Subtracting scores from the first to the last time
points could overlook non-linear change. In addition, others have proposed that
difference scores may be less than reliable because they could include random
measurement error (Cronbach and Furby, 1970). However, the notion of unreliability has
been challenged (Rogosa & Willett, 1983; Allison, 1990). To circumvent the possible
issues of difference scores, I attempted to use latent difference score models, but my
sample size prevented meaningful results. Further, cognitive change over one year may
not be enough to detect significant change. In the normal control group in particular,
cognitive change over one year may be restricted in range, thus limiting the ability to find
significant results. Change in performance over a greater number of years (e.g., 5-10)
might have a greater range and therefore be more likely to correlate with complaints.
I originally hypothesized that dementia severity would be a possible moderator of
the association between complaints and decline. Examining dementia severity is of
interest because people within dementia groups could vary widely in cognition levels and
perhaps in their ability to accurately detect cognitive change. However, these analyses
were not performed because when the final dataset was available, it became clear that an
analysis by dementia severity was unrealistic due to limited sample sizes of CDR severity
groups. However, future studies could determine whether there are dementia severity
subgroups that are relatively more or less accurate and whether one group is driving the
overall absence of associations between complaints and cognitive change.
69

One additional limitation was that information regarding participant medications
was not available in the ADRC dataset. Thus, I was unable to examine the possible
ameliorating or exacerbating relationship of antidepressants or cognitive-enhancing
medications on depression, complaints, or cognition.
Further, one important area for future research would be to determine the degree
of overestimation versus underestimation of cognitive complaints. Participants as a
whole generally were found to be inaccurate, but they could have been inaccurate in
different ways. Typically, people with Alzheimer’s disease are thought to underestimate
their difficulties, whereas individuals with normal cognition are thought to overestimate
their problems. Unfortunately, the complaints measure in the present study did not lend
itself to such analyses.
Finally, using one neuropsychological measure to represent each cognitive
domain could be limiting. Ideally, a memory measure would include all aspects of the
construct, such as learning, retrieval, and recognition of information. Individuals who
subjectively experience memory difficulties could have deficits on one, two, or all three
of these subtests. Thus, using only one measure of memory in the present study could
have failed to detect other types of memory problems. Future studies would benefit from
making a composite measure or latent variable of each of the cognitive domains using
multiple tests and subscores from the tests.
Despite these limitations, this study makes a number of important contributions
and strengthened the complaints literature in various ways. To begin, the complaints
measurement in this study assessed complaints differently than many other studies. First,
70

my complaints measure asked about change from a specific prior time point, ensuring
that all participants were evaluating the same concept of change with the same time
reference. In previous studies, participants were often asked a rather open-ended
question about current memory problems experienced (e.g., Jorm et al., 2001). Asking
individuals to compare their abilities to those from a specific time point may force them
to answer self-referentially, rather than compared to how they think other people are
functioning. I assumed that self-referential comparisons would yield more accurate
estimates of cognitive change. However, the data here do not necessarily support this
hypothesis. As noted earlier, perhaps it is unrealistic to ask AD patients to make a
comparison of their cognition from a previous time point, as doing so relies so heavily on
their memory abilities which are impaired. Dementia patients may be able to make
moment-by-moment judgments more accurately than year-by-year judgments. Older
adults with dementia (even with severe levels) have been shown to be reliable reporters
of their depressive symptoms in the past week (Burke et al., 1992). Thus, perhaps
accuracy of complaints in AD individuals could be improved by asking about cognitive
problems in the past week. Future studies should test this hypothesis. In addition, it may
also prove a difficult task for normal older adults to think about their relative
performance compared to a specific prior time point. Given misperceptions about what is
normal for one’s age and what is abnormal (with some people overestimating their
difficulties and other underestimating), cognitively intact individuals could be poor
estimators because they have limited knowledge of what is normal for aging, and they
may discount change sometimes or inflate it other times. Thus, an interesting future
71

research question could be whether knowledge of typical changes in aging are associated
with greater accuracy.
Further, using the same measure in a combined normal and dementia sample also
drew together the complaints and awareness literatures. By examining the same
phenomenon in older adults with normal cognition and with Alzheimer’s disease, we can
compare their levels of accuracy, and doing so may shed light on reasons for complaint
inaccuracy. For example, poor awareness in people with AD has been hypothesized to
stem from frontal lobe dysfunction. A similar hypothesis has not been proposed in the
cognitively normal groups. However, because the cognitively normal group and the AD
group had similar associations between complaints and cognitive change, future research
may want to examine the effect of frontal lobe deficits on poor complaints accuracy in
both demented and non-demented samples. In addition, the previously-discussed
interaction of diagnosis and depression on memory complaints demonstrates the value of
combining samples.
This is also the first study to separate complaints and performance by domain, and
thus to suggest that older adults may be differentially aware of their performance in
different cognitive domains. Prior studies have asked primarily about memory. In
contrast, this study also asked about perceptions of change regarding language and
attention. The finding that language complaints predicted future language change will be
important to further explore in future studies. Word-finding difficulties are common in
older adults, but asking about perceived change from a discrete time point (rather than
compared to earlier in life) may be uncommon. It is possible that most people
72

acknowledge language changes from earlier in life, but when asked about recent changes,
only individuals who actually changed endorsed problems. Changes to language may be
more obvious to the patient and individuals around the patient, as communication
abilities are easily perceived compared to attention and memory problems, which are
more internal. In research studies, asking only about memory complaints could result in
individuals appearing less aware than they actually are. Thus, if these language findings
are replicable, both clinicians and researchers would benefit from asking older adults
about perceived language decline.
Finally, the finding that depression rather than cognitive decline predicted
complaints in each domain has a number of implications. To begin, given that
complaints about past change were not accurate, questions continue to arise about the
clinical utility of asking individuals about their cognition. Although it certainly seems
humane and respectful to consider the patient’s perception, the information may prove
misleading. These findings also suggest the importance of screening for depression if a
patient presents with complaints. Prior research has suggested that depression in older
adults presents differently than in younger adults. For example, older adults are thought
to evidence fatigue and loss of interest, irritability, less sadness, and more health
complaints (Murrell, Himmelfarb, & Wright, 1983; Newmann, Engel & Jensen, 1991).
Cognitive complaints may be another key indicator of depression in older adults. In
addition, clinicians may want to be particularly vigilant about detecting and treating
depression given that lower depressive symptoms were associated with better future
memory performance. Thus, depression in both cognitively normal and AD older adults
73

appears to affect both subjective perceptions of cognition as well as objective future
decline in memory.

74

References

Aalten, P., Van Valen, E., Clare, L., Kenny, G., & Verhey, F. (2005). Awareness in
dementia: A review of clinical correlates. Aging & Mental Health, 9, 414-422.

Aiken, L. S. & West, S. G. (1991). Multiple regression: Testing and interpreting
interactions. Thousand Oaks, CA: Sage.

Allison, P. D. (1990). Change scores as dependent variables in regression analysis. In C.
Clogg (Ed.), Sociological Methodology (pp. 93-114) Oxford: Basil Blackwell.

American Psychiatric Association. (2000). Diagnostic and statistical manual of mental
disorders: DSM-IV-TR. Washington, DC: Author.

Alvarez, J. A., & Emory, E. (2006). Executive function and the frontal lobes: A meta-
Analytic review. Neuropsychology Review, 16, 17-42.

Arkin, S. & Mahendra, N. (2001). Insight in Alzheimer's patients: Results of a
longitudinal study using three assessment methods. American Journal of Alzheimers
Disease and Other Dementias, 16, 211-24.

Barrett, A. M., Eslinger, P. J., Ballentine, N. H. & Heilman, K. M. (2005). Unawareness
of cognitive deficit (cognitive anosognosia) in probable AD and control subjects.
Neurology, 64, 693-699.

Bassett, S. S.& Folstein, M. F. (1993). Memory complaint, memory performance, and
psychiatric diagnosis: A community study. Journal of Geriatric Psychiatry and
Neurology, 6, 105–111.

Beck, A. T. (1963). Thinking and depression: I. Idiosyncratic content and cognitive
distortions. Archives of General Psychiatry, 9, 324-333.

Blazer, D. G., Hays, J. C., Fillenbaum, G. G., Gold, D. T. (1997). Memory complaint as
a predictor of cognitive decline: A comparison of African American and White
elders. Journal of Aging and Health, 9, 171-184.

Brakk, H. & Brakk, E. (1991). Neuropathological staging of Alzheimer-related changes.
Acta Neuropathologica, 82, 239-259.

Bremner, J. D., Narayan, M., Anderson, E. R., Staib, L. H., Miller, H. L., & Charney, D.
S. (2000). Hippocampal volume reduction in major depression. American
Journal of Psychiatry, 157, 115-117.

75

Brink, T. L., Yesavage, J. A., Lum, O., Heersema, P., Adey, M. B., & Rose, T. L.
(1982). Screening tests for geriatric depression. Clinical Gerontologist, 1, 37-44.

Burke, W. J., Nitcher, R. L., Roccaforte, W. H., & Wengel, S. P. (1992). A prospective
evaluation of the Geriatric Depression Scale in an outpatient geriatric assessment
center. Journal of the American Geriatrics Society, 40, 1227-1230.

Clark, D. M. (1986). A cognitive approach to panic. Behavioural Research and Therapy,
24, 461-470.

Cronbach, L. J. & Furby, L. (1970). How should we measure "change" - or should we?
Psychological Bulletin, 74, 68-80.

Crowe, M., Andel, R., Wadley, V., Cook, S., Unverzagt, F., Marsiske, M., et al. (2006).
Subjective cognitive function and decline among older adults with
psychometrically defined amnestic MCI. International Journal of Geriatric
Psychiatry, 21, 1187-1192.

Dalla Barba, G., Parlato, V., Iavarone, A., & Bollert, F. (1995). Anosognosia, intrusions,
and “frontal” functions in Alzheimer’s disease and depression.
Neuropsychologia, 33, 247-259.

Derousne, C. Dealberto, M. J., Boyer, P., Lubin, S., Sauron, B., Peitte, F., Kohler, F., et
al (1993). Empirical evaluation of the Cognitive Difficulties Scale for
assessment of memory complaints in general practice: A study of 1628
cognitively normal subjects aged 45-75 years. International Journal of Geriatric
Psychiatry, 8, 599-607.

Di Carlo, A., Baldereschi, M., Amaducci, L., Maggi, S., Grigoletto, F., Scarlato, G., et al.
(2002). Cognitive impairment without dementia in older people: Prevalence,
vascular risk factors, impact on disability. Journal of the American Geriatrics
Society, 48, 775-782.

Dik, M. G., Jonker, C., Comijs, H. C., Bouter, L.M., Twisk, J. W. R., van Kamp, G. J., et
al. (2001). Memory complaints and APOE-4 accelerate cognitive decline in
cognitively normal elderly. Neurology, 57, 2217–2222.

Downs, M. (2005). Editorial: In the eye of the beholder. Aging and Mental Health, 9,
381-383.

Droes, R. M. (2007). Insight in coping with dementia: Listening to the voice of those
who suffer from it. Aging and Mental Health, 11, 115-119.
76

Dufouil, C., Fuhrer, R., & Alperovitch, A. (2005). Subjective cognitive complaints and
cognitive decline: Consequence or predictor? The Epidemiology of Vascular
Aging Study. Journal of the American Geriatrics Society, 53, 616-621.

Duke, L. M., Seltzer, B., Seltzer, J. E., & Vasterling, J. J. (2002). Cognitive components
of deficit awareness in Alzheimer’s disease. Neuropsychology, 16(3), 359-369.

Fama, R., Sullivan, E. V., Shear, P. K., Cahn-Weiner, D. A., Yesavage, J.
A.,Tinklenberg, J. R., et al. (1998). Fluency performance patterns in Alzheimer's
disease and Parkinson's disease. Clinical Neuropsychologist, 12, 487-499.

Feehan, M., Knight, R. G., & Partridge, F. M. (1991). Cognitive complaint and test
performance in elderly patients suffering depression or dementia. International
Journal of Geriatric Psychiatry, 6, 287-293.

Flicker, C., Ferris, S. H., & Reisberg, B. (1993). A longitudinal study of cognitive
function in elderly persons with subjective memory complaints. Journal of the
American Geriatrics Society, 41, 1029–1032.

Gagnon, M., Dartigues, J. F., Mazaux, J. M., Dequae, L., Letenneur, L., Giroire, J. M., et
al. (1994). Self-reported memory complaints and memory performance in
elderly French community residents: Results of the PAQUID Research Program.
Neuroepidemiology, 13, 145–154.

Geerlings, M. I., Jonker, C., Bouter, L. M., Ader, H. J., & Schmand, B. (1999).
Association between memory complaints and incident Alzheimer’s disease in
elderly people with normal baseline cognition. American Journal of Psychiatry,
156, 531-537.

Graham, D. P., Kunika, M. E., Doody, R., & Snow, A. L. (2005). Self-reported
awareness of performance in dementia. Cognitive Brain Research, 25, 144-152.

Harwood, D. G., Sultzer, D. L., & Wheatley, M.V. (2000). Impaired insight in
Alzheimer disease: Association with cognitive deficits, psychiatric symptoms, and
behavioural disturbances. Neuropsychiatry, Neuropsychology, and Behavioral
Neurology, 13, 83-88.

Howorth, P. & Saper, J. (2003). The dimensions of insight in people with dementia.
Aging & Mental Health, 7, 113–122.

Johansson, B., Allen-Burge, R., & Zarit, S. H. (1997). Self reports on memory
functioning in a longitudinal study of the oldest old: relation to current,
prospective, and retrospective performance. Journal of Gerontology B;
Psychological Science and Social Science, 52, 139-146.
77

Jonker, C., Geerlings, M. I., & Schmand, B. (2000). Are memory complaints predictive
for dementia? A review of clinical and population-based studies. International
Journal of Geriatric Psychiatry, 15, 983-991.

Jonker, C., Launer, L. J., Hooijer, C., & Lindeboom, J. (1996). Memory Complaints and
Memory Impairment in Older Individuals. Journal of the American Geriatrics
Society, 44, 44-49.

Jorm, A. F., Christensen, H., Korten, A. E., Jacomb, P. A., & Henderson, S. (2001).
Memory complaints as a precursor of memory impairment in older people: A
longitudinal analysis over 7-8 years. Psychological Medicine, 31, 441-449.

Jungwirth, S., Fischer, P., Weissgram, S., Kirchmeyr, W., Bauer, P., & Tragl, K-H.
(2004). Subjective memory complaints and objective memory impairment in the
Vienna-transdanube aging community. Journal of the American Geriatrics
Society, 52, 263-268.

Kim, J., Stewart, R., Kim, S., Yang, S., Shin, I., & Yoon, J. (2006). A Prospective Study
of Changes in Subjective Memory Complaints and Onset of Dementia in South
Korea. American Journal of Geriatric Psychiatry, 14, 949-956.

King, L. A., King, D. W., McArdle, J. J., Saxe, G. N., Doron-Lamarca, S., & Orazem, R.
J. (2006). Latent difference score approach to longitudinal trauma research.
Journal of Traumatic Stress, 19, 771-785.

Kliegel, M. K., Zimprich, D., & Eschen, A. (2005). What do subjective cognitive
complaints in persons with aging-associated cognitive decline reflect?
International Psychogeriatrics, 17, 499–512.

Lam, L. C. W., Lui, V. W. C., Tam, C. W. C., & Chiu, H. F. K. (2005). Subjective
memory complaints in Chinese subjects with mild cognitive impairment and early
Alzheimer’s disease. International Journal of Geriatric Psychiatry, 20, 876–882.

Lane, R. D. (1998). Neural correlates of levels of emotional awareness: Evidence of an
interaction between emotion and attention in the anterior cingulate cortex.
Journal of Cognitive Neuroscience, 10, 525-535.

Lau, M. A., Christensen, B. K., Hawley, L. L., Gemar, M. S., & Segal, Z. V. (2007).
Inhibitory deficits for negative information in persons with major depressive
disorder. Psychological Medicine, 37, 1249-59.

Lezak, M. D., Howieson, D. B., & Loring, D. W. (2004). Neuropsychological
assessment (4
th
ed.). New York: Oxford.
78

Lichtenberg, P. A. & Christensen, B. (1992). Extended normative data for the Logical
Memory subtests of the Wechsler Memory Scale – Revised: Responses from a
sample of cognitively intact elderly medical patients. Psychological Reports, 71,
745-746.

Little, R.J.A. and Rubin, D.B. (1987). Statistical Analysis with Missing Data. New York:
John Wiley.

Lopez, O. L., Becker, J. T., Somsak, D., Dew, M. A., & DeKosky, S. T. (1994).
Awareness of cognitive deficits and anosognosia in probable Alzheimer's disease.
European Neurology, 34, 277-282.

Mangone, C. A., Hier, D. B., Gorelick, P. B., Ganellen, R. J., Langenberg, P., Boarman,
R., et al. (1991). Impaired insight in Alzheimer's disease. Journal of Geriatric
Psychiatry and Neurology, 4, 189-93.

McArdle, J. J., & Hamagami, F. (2001). Linear dynamic analyses of incomplete
longitudinal data. In L. Collins & A. Sayer (Eds.), Methods for the analysis of
change (pp. 137–176). Washington, DC: American Psychological Association.

McGlynn, S. M. & Kaszniak, A. W. (1991). Unawarenss of deficits in dementia and
schizophrenia. In G. P. Prigatano & D. L. Schacter (Eds.), Awareness of deficit
after brain injury: Clinical and theoretical issues. New York: Oxford University
Press.

McKhann, G., Drachman, D., Folstein, M., Katzman, R., Price, D., & Stadlan E. M.
(1984). Clinical diagnosis of Alzheimer’s disease. Report of the NINCDS-
ADRDA work group. Neurology, 34, 939-944.

Migliorelli, R., Testa, A., Sabe, L., Petracca, G., Petracchi, M., Leiguarda, R., et al.
(1995). Anosognosia in Alzheimer’s disease: A study of associated factors. The
Journal of Neuropsychiatry and Clinical Neurosciences, 7, 338-344.

Morris, J.C. & Cummings, J. (2005). Mild cognitive impairment (MCI) represents early-
stage Alzheimer's disease. Journal of Alzheimer's Disease, 7, 235-239.

Morris, J. C., Heyman, A., Mohs, R. C., Hughes, J. P., van Belle, G., Fillenbaum, G., et
al. (1989). The Consortium to Establish a Registry for Alzheimer's Disease
(CERAD). Part I. Clinical and neuropsychological assesment of Alzheimer's
disease. Neurology, 39, 1159-1165.

79

Murrell, S. A., Himmelfarb, S., & Wright, K. (1983). Prevalence of depression and its
correlates in older adults. American Journal of Epidemiology, 117, 173-185.

Muthen, L.K., & Muthen, B.O. (2006). Mplus user’s guide. Los Angeles, CA: Muthen
& Muthen.

Neary D., Snowden, J. S., Gustafson, L., Passant, U., Stuss, D., Black, S., et al. (1998).
Frontotemporal lobar degeneration: a consensus on clinical diagnostic criteria.
Neurology, 51, 1546–1554.

Newmann J. P., Engel R. J., Jensen J. E. (1991). Age differences in depressive symptom
experiences. Journal of Gerontology Psychological Sciences, 46A, 224-235.

Newson, R. S. and Kemps, E. B. (2006). The nature of subjective cognitive complaints
of older adults. International Journal of Aging and Human Development, 63,
139-151.

O’Connor, D. W., Pollitt, P. A., Roth, M., Brook, P. B., & Reiss, B. B. (1990). Memory
complaints and impairment in normal, depressed, and demented elderly persons
identified in a community survey. Archives of General Psychiatry, 47, 224–227.

Ownby, R. L., Crocco, E., Acevedo, A., John, V., & Loewenstein, D. (2006). Depression
and risk for Alzheimer disease: Systematic review, meta-analysis, and
metaregression analysis. Archives of General Psychiatry, 63, 530-538.

Pachana N. A., and Petriwskyj, A. M. (2006). Assessment of insight and self-awareness
in older drivers. Clinical Gerontologist, 30, 23-38.

Petersen, R. C., Smith, G. E., & Waring, S. C. (1999). Mild cognitive impairment:
Clinical characterization and outcome. Archives of Neurology, 56, 303-308.

Rabin, L. A., Roth, R. M., Isquith, P. K., Wishart, H. A., Nutter-Upham, K. E., Pare, N.,
et al. (2006). Self- and informant reports of executive function on the BRIEF-A
in MCI and older adults with cognitive complaints. Archives of Clinical
Neuropsychology, 21, 721-732.

Reid, L. M. & MacLullich, A. M. J. (2006). Subjective Memory Complaints and
Cognitive Impairment in Older People. Dementia and Geriatric Cognitive
Disorders, 22, 471–485.

Reisberg, B., Gordon, B., McCarthy, M., Ferris, S. H. & deLeon, M. J. (1985). Insight
and denial accompanying progressive cognitive decline in normal aging and
Alzheimer’s disease. In Stanley, B (Ed.), Geriatric psychiatry: Clinical, ethical,
and legal issues. Washington D.C.: APA.
80

Reitan R. M. (1958). Validity of the Trail Making test as an indicator of organic brain
damage. Perceptual Motor Skills, 8, 271-276.

Rogosa, D. R. & Willett, J. B. (1983). Demonstrating the reliability of the difference
score in the measurement of change. Journal of Educational Measurement, 20,
335-343.

Salmon, D. P. & Bondi, M. W. (2009). Neuropsychological assessment of dementia.
Annual Review of Psychology, 60, 257–282.

St John, P. & Montogomery, P. (2002). Are cognitively intact seniors with subjective
memory loss more likely to develop dementia? International Journal of Geriatric
Psychiatry, 17, 814–820.

Schmand, B., Jonker, C., Hooijer, C., & Lindeboom, J. (1996). Subjective memory
complaints may announce dementia. Neurology, 46, 121-125.

Schofield, P. W., Marder, K., Dooneief, G., Jacobs, D. M., Sano, M., & Stern, Y. (1997).
Association of subjective memory complaints with subsequent cognitive decline
in community-dwelling elderly individuals with baseline cognitive impairment.
American Journal of Psychiatry, 154, 609-615.

Smith, C. A., Henderson, V. W., McCleary, C. A., Murdock, G. A., & Buckwalter, J. G.
(2000). Anosognosia and Alzheimer’s disease: The role of depressive symptoms
in mediating impaired insight. Journal of Clinical and Experimental
Neuropsychology, 22, 437-444.

Smith, G. E., Petersen, R. C., Ivnik, R. J., Malec, J. F., & Tangalos, E. G. (1996).
Subjective memory complaints, psychological distress, and longitudinal change in
objective memory performance. Psychology of Aging, 11, 272–279.

Starkstein, S. E., Sabe, L., Chemerinski, E., Jason, L., & Leiguarda, R. (1996). Two
domains of anosognosia in Alzheimer’s disease. Journal of Neurology,
Neurosurgery, and Psychiatry, 61, 485-490.

Starkstein, S. E., Sabe, L., Chemerinski, E., Jason, L., & Leiguarda, R. (1996). Two
domains of anosognosia in Alzheimer’s disease. Journal of Neurology,
Neurosurgery, and Psychiatry, 61, 485-490.

Taylor, J. L., Miller, T. P., & Tinklenberg, J. R. (1992). Correlates of memory decline:
A 4-year longitudinal study of older adults with memory complaints. Psychology
and Aging, 7, 185-193.
81

Tobiansky, R., Blizard, R., Livingston, G., & Mann, A. (1995). The Gospel Oak Study
stage IV: the clinical relevance of subjective memory impairment in older people.
Psychological Medicing, 25, 779–786.

Tsai, D. H., Green, R. C., Benke, K. S., Silliman, R. A., & Farrer, L. A. (2006).
Predictors of subjective memory complaint in cognitively normal relatives of
patients with Alzheimer’s disease. The Journal of Neuropsychiatry and Clinical
Neurosciences, 18, 384-388.

Von Gunten, A., Fox, N. C., Cipolotti, L., & Ron, M.A. (2000). A volumetric study of
hippocampus and amygdala in depressed patients with subjective memory
problems. Journal of Neuropsychiatry and Clinical Neuroscience, 12, 493–498.

Wang, L., van Belle, G., Crane, P. K., Kukull, W. A., Bowen, J. D., McCormick, W. C.,
et al. (2004). Subjective memory deterioration and future dementia in people
aged 65 and older. Journal of the American Geriatric Society, 52, 2045–2051.

Wang, P. N., Wang, S. J., Fuh, J. L., Teng, E. L., Liu, C. Y., Lin, C. H., et al. (2000).
Subjective memory complaint in relation to cognitive performance and
depression: A longitudinal study of a rural Chinese population. Journal of the
American Geriatrics Society, 48, 295–299.

Wilcox, R. R. (2003). Applying Contemporay Statistical Techniques. San Diego:
Academic Press.

Wong, C. H. Y., Lam, L. C. W., Lui, V. W. C., Chiu, H. F. K; Chan, S. S. M., & Tam, C.
W. C. (2006). Subjective complaints and self-evaluation of memory test
performance in Questionable dementia. International Journal of Geriatric
Psychiatry, 21, 937-944.

Yesavage, J. A., Brink, T. L., Rose, T. L., Lum, O., Huang, V., Adey, M., et al. (1982-
1983). Development and validation of a geriatric depression screening scale: a
preliminary report. Journal of Psychiatric Research, 17, 37-49.

Zanetti, O., Vallotti, B., Frisoni, G. B., Geroldi, C., Bianchetti, A., Pasqualetti, P., &
Trabucchi, M. (1999). Insight in dementia: When does it occur? Evidence for a
nonlinear relationship between insight and cognitive status. Journal of
Gerontology: Psychological Sciences, 54B, 100-106.

Zimprich, D., Martin, M., & Kliegel, M. (2003). Subjective cognitive complaints,
memory performance, and depressive affect in old age: A change oriented
approach. International Journal of Aging and Human Development, 57, 339-366.

82

Appendix A
Complaints Questionnaire

PRE-TESTING PATIENT INTERVIEW

QUESTIONS FOR PATIENT/SUBJECT:
1. Have you noticed any cognitive changes since we saw you last? (follow-up testing) or
Have you notices any cognitive changes during the past few years? (initial testing)

2. More specifically, have you noticed changes in:

a. Attention/concentration –

b. Problems with language – (word finding, speech, pronunciation of multi-
syllabic words)

c. Memory – (problems with learning new information)

3. What about visuospatial skills such as:

a. Map reading –

b. constructing – (putting things together with your hands)

4. Have you been feeling unusually anxious or depressed?

5. (If yes to depressed) Have you been considering ending your life?
83

Appendix B

Coding manual
1 2 3 4 5
Full denial/non-
endorsement
Defensive
denial/deflected
Partial denial/partial
endorsement (or overall
denial/endorsement
with qualifications, e.g.,
of severity or
frequency)
Endorsement of
problems
participant
considers
“normal for
age.”
Full
endorsement
--“No”
--I’m fine.
--I don’t have any
problems.

Nothing out of the
ordinary.

I wouldn’t say I
have anything bad.

“Dash”

No change

-Respondent notes
a change unrelated
to the domain
asked about.

I don’t think so.
(NOT “I don’t
know.”)

Same.

I haven’t noticed
changes.
“I don’t do that
anymore.”

“Depends on
my interest.”

-Notes physical
change that
result in the
avoidance of
activities asked
about (e.g., “I
don’t do ‘X’
because of my
macular
degeneration.”)

“Ask my wife.”

--I think so.
--I have problems with
X, but it’s not new/I’ve
always had this
problem/same as last
year.
--Notes problems,
maybe related to
physical changes. “I
have problems, but it’s
because my vision has
decreased.”
--I have that problem,
but it’s not functionally
impairing.
--It’s worse, but not that
bad.
--“My ability is
somewhat decreased.”
--I lose a word now and
then.
--Occasionally forget
names.
--Minor probs.
--“Sort of”
--A little worse
--Sometimes.
Yes, but the
changes are
normal for a
person my age.

I’ve changed
since I was
younger, but it’s
not bad or
abnormal.

No worse than
someone else my
age.

Yes- not
unusual.
--Definitely.
--Yes.
--My
[cognitive
domain] has
gotten worse.

--I’m more
forgetful.

--I just can’t
seem to
remember
anymore.

--I have word-
finding
problems.

--For attention:
I misplace
items more

--“It’s not as
good as it once
was.”

--It’s no better
than it was.

6 7
I don’t know Non sequitur
I don’t know
I’m not sure

My sinuses are
bothering me.

I work at the school.

Abstract (if available)

Abstract The purpose of this study was to determine whether subjective complaints about change in memory, attention, and language accurately reflect past decline or predict future cognitive decline on neuropsychological tests in older adults with normal cognition or with Alzheimer’s disease. Further, the effect of dementia status and depressive symptoms on these relationships was assessed. Using the USC Alzheimer’s Disease Research Center longitudinal dataset, subjective cognitive complaints, cognitive performance, and depressive symptoms were assessed in 132 participants. Overall, neither diagnostic group’s complaints accurately reflected past decline for any cognitive domain. Rather, depressive symptoms predicted endorsement of complaints for all domains. In addition, dementia status moderated the effect of depression on future memory decline. Finally, language complaints predicted future language decline in the entire sample, but similar relationships were not present for attention and memory. Implications of these findings are discussed.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Underlying neural mechanisms of depression and dementia

PDF

Crosscultural differences in dementia diagnosis and care-seeking in Hispanic and non-Hispanic white outpatients

PDF

Affective neuropsychiatric symptoms and neural connectivity in the early stages of Alzheimer’s disease

PDF

Predicting cognitive decline and dementia in elderly twins from indicators of early life oral health

PDF

Brainstem structural integrity in the progression of Alzheimer's disease

PDF

Dementia, co-morbid depressive symptoms, and mortality in older adults: a twin study

PDF

Comparison of participant and study partner predictions of cognitive impairment in the Alzheimer's disease neuroimaging initiative 3 study

PDF

More knowledge, better plans: a study of heterogeneity in dementia prevalence and mid-life cognitive changes

Pdf

Alzheimer's disease: risk factors, value, and alternative payment models

PDF

Effects of AT-1 receptor blockers on cognitive decline and Alzheimer's disease

PDF

Spouse aggression, depression, and physical health: a multivariate longitudinal study of midlife couples

PDF

The effects of time and space on health status of older adults in China

PDF

Renin-angiotensin system antihypertensive medicines with and without blood-brain-barrier crossing potential: effects on cognitive outcomes in the elderly

PDF

Independent and interactive effects of depression genetic risk and household socioeconomic status on emotional behavior and brain development

PDF

The role of the locus coeruleus in Alzheimer’s disease and cerebrovascular function: insights from neuroimaging, neuropsychology, and biofluid markers

PDF

A series of longitudinal analyses of patient reported outcomes to further the understanding of care-management of comorbid diabetes and depression in a safety-net healthcare system

PDF

Self-perceptions of Aging in the Context of Neighborhood and Their Interplay in Late-life Cognitive Health

PDF

Experience-dependent neuroplasticity of the dorsal striatum and prefrontal cortex in the MPTP-lesioned mouse model of Parkinson’s disease

PDF

Evaluating social-cognitive measures of motivation in a longitudinal study of people completing New Year's resolutions to exercise

PDF

Exploring the effects of Korean subject marking and action verbs’ repetition frequency: how they influence the discourse and the memory representations of entities and events

Asset Metadata

Creator Dave, Jennifer B. (author)

Core Title Accuray of subjective cognitive complaints in a longitudinal context: the effect of depression and dementia status

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Psychology

Publication Date 07/18/2009

Defense Date 05/19/2009

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag aging,Alzheimer's disease,dementia,Depression,OAI-PMH Harvest,subjective complaints

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Knight, Bob G. (committee chair), McCleary, Carol (committee member), Prescott, Carol A. (committee member), Walsh, David (committee member), Walsh, John P. (committee member)

Creator Email jennifbd@usc.edu,jennifer.dave@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-m2368

Unique identifier UC1276431

Identifier etd-Dave-3005 (filename),usctheses-m40 (legacy collection record id),usctheses-c127-407218 (legacy record id),usctheses-m2368 (legacy record id)

Legacy Identifier etd-Dave-3005.pdf

Dmrecord 407218

Document Type Dissertation

Rights Dave, Jennifer B.

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Repository Name Libraries, University of Southern California

Repository Location Los Angeles, California

Repository Email cisadmin@lib.usc.edu