University of Southern California Dissertations and Theses

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

(USC Thesis Other)

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

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Running head: DEMENTIA, DEPRESSION, AND MORTALITY 1

Dementia, Co-morbid Depressive Symptoms, and Mortality in Older Adults: A Twin Study
Jung Yun Jang
University of Southern California
Second Year Project
May 2013

Author Note

This research was supported by grant R01 AG08724 from the National Institute on Aging.
DEMENTIA, DEPRESSION, AND MORTALITY 2

Table of Contents

Abstract ............................................................................................................................................3

Introduction ......................................................................................................................................4

Method .............................................................................................................................................9
Participant Population ..............................................................................................................9
Measures ................................................................................................................................10
Study Samples ........................................................................................................................13
Statistical Analyses ................................................................................................................15

Results ............................................................................................................................................17
Study 1 ...................................................................................................................................17
Study 2 ...................................................................................................................................20

Discussion ......................................................................................................................................21

References ......................................................................................................................................26

Tables and Figures .........................................................................................................................32

Appendix ........................................................................................................................................42

DEMENTIA, DEPRESSION, AND MORTALITY 3
Abstract

Dementia and cognitive impairment are known to increase the risk of mortality in older adults. It
has been suggested as well that depressive symptoms in persons with dementia further elevate
their risk of mortality. However, these associations have not been investigated in a genetically
informative sample. Using a population-based sample of twins in the Study of Dementia in
Swedish Twins (HARMONY), Study 1 of the present study examined the association between
dementia and mortality in the entire sample as well as within pairs of twins discordant for
dementia status. Study 2 tested the association between co-morbid depressive symptoms and
mortality in twins with dementia. In Study 1, Cox proportional hazards regression analysis
revealed that individuals with dementia had more than doubled risk of death, compared with
individuals with normal cognition (HR=2.38, 95% CI [2.07, 2.73]). As hypothesized,
comparisons within dementia discordant pairs also showed significantly greater risk of death for
those with dementia, indicating that genetic and early environmental influences did not account
for the increased risk of death in individuals with dementia. Indeed, median survival time
estimated since the demented twin’s age of onset was 9.8 years for twins with dementia and 14.7
years for their cognitively normal co-twins. Study 2 failed to replicate previous studies and found
no effect of co-morbid depressive symptoms on mortality in twins with dementia. The present
study is the first to look at survival of dementia patients using the co-twin design. These findings
may be informative to patients, families, clinicians, and policy makers in making decisions for
dementia care.

DEMENTIA, DEPRESSION, AND MORTALITY 4
Many studies have reported increasing incidence and prevalence of dementia in older
adults as our life expectancy has expanded substantially over the years. For example, one expert
review of published studies estimated approximately 24.3 million people diagnosed with
dementia worldwide with a projection of 4.6 million new cases each year (Ferri et al., 2005).
Alzheimer’s disease (AD), the most common type of dementia, is now the sixth leading cause of
death for people in the U.S. (Alzheimer’s Association, 2012).
As the burden of dementia on the public in terms of cost, care, and treatment is
continuously expected to be on the rise, research endeavors have focused on characterizing the
consequences of dementia and shortened lifespan in particular (Fitzpatrick, Kuller, Lopez,
Kawas, & Jagust, 2005; Waring, Doody, Pavlik, Massman, & Chan, 2005). Findings have
demonstrated that persons with dementia have their annual risk for mortality two to four times
higher than cognitively normal, age-matched older adults (Dewey & Saz, 2001; Fitzpatrick et al.,
2005). Also reported in the epidemiological literature are estimates of the duration of survival in
dementia patients. Differences exist among reports depending on a number of factors, including
the setting, type of the study, and dementia subtypes. For example, estimates of survival duration
tended to be shorter for long-term care patients and longer for outpatient clinic patients, probably
due to differences in dementia severity (Mölsä, Marttila, & Rinne, 1986). Moreover, studies
found a favorable prognosis for patients with AD, compared with vascular dementia (VaD) in
survival duration (Fitzpatrick et al., 2005; Knopman, Rocca, Cha, Edland, & Kokmen, 2003;
Mölsä et al., 1986). Particularly, survival in AD has been a subject of many population- or
community-based studies, with ranges from 3 to 9 years after diagnosis according to a recent
summary (Helzner et al., 2008).
In addition, a large body of literature has examined the association between cognitive
DEMENTIA, DEPRESSION, AND MORTALITY 5
decline and mortality in old age. One study summarized research findings to evaluate the effect
of cognitive impairment on mortality in community-living samples aged 65 and older with or
without dementia and reported a dose-response relationship between the severity of cognitive
impairment and increased mortality (Dewey & Saz, 2001). Another study conducted in the U.K.
using population-based samples aged 65 and older with global cognitive functioning measured
by Mini-Mental State Examination (MMSE) scores produced similar results (Adjusted HR =
1.24, 1.77, and 3.20; 95% CI [1.13-1.36], [1.60-1.97], and [2.82-3.63] for mild, moderate, and
severe impairment respectively, compared with intact cognitive functioning) (Neale, Brayne, &
Johnson, 2001). However, there has not been an attempt to account for the potential effects of
genetic background and shared early environmental factors when examining the risk of mortality
and duration of survival in cognitive impairment and dementia, a question that can be examined
using twin data.
It is of great interest to patients, families, and clinicians to be able to predict trajectories
in the context of cognitive impairment and dementia so as to ensure the patient’s quality of life
and effectively manage his or her functional limitations during the rest of the patient’s life.
Amongst potential factors affecting duration of survival in dementia patients, research has shown
that depressive symptoms may be detrimental to the prognosis of cognitive impairment and
dementia, including mortality (Gilley et al., 2004; Lee & Lyketsos, 2003; Okura et al., 2011).
Depressive symptoms and dementia often co-occur and appear to be closely related, leading
investigators to try to delineate the temporal association between depression and dementia in
order to understand the underlying mechanisms (Korczyn & Halperin, 2009). It is particularly
important to understand the role of depression in survival of persons with cognitive impairment
and dementia because depressive symptoms can be treated appropriately.
DEMENTIA, DEPRESSION, AND MORTALITY 6
Previous research has demonstrated that there is a significant relationship between
depression and increased mortality in older adults without regard to cognitive impairment. A
systematic review of literature has revealed a positive association between depression and non-
suicide mortality in adults aged 60 and older (Schulz, Drayer, & Rollman, 2002). Another
review of literature has used meta-analysis to combine the effect sizes of studies using
community-residing samples aged 65 and older with clinically diagnosed depression and found
an elevated risk of mortality (OR = 1.73, 95% CI [1.53, 1.93]) (Saz & Dewey, 2001). It is
important to note that the majority of studies included in these reviews did not take into
consideration the level of the participant’s cognitive functioning and other alternative
explanations.
While researchers have demonstrated that depression and cognitive impairment each
independently predicts a greater risk of mortality in older adults, fewer studies have investigated
the effect of depressive symptoms on mortality of persons with cognitive impairment and
dementia. Frequent reports of neuropsychiatric symptoms are well-documented in persons with
cognitive impairment

and dementia (Apostolova & Cummings, 2008; Lyketsos et al., 2002), and
depression is highly prevalent in both groups. Findings from studies of depression in dementia
have shown its association with a greater rate of functional decline and negative outcomes. For
instance, the presence of depression was linked to faster decline in measures of global cognitive
functioning over time (i.e., MMSE score differences), compared with dementia alone (Rapp et
al., 2011). In addition, there is evidence to support that adverse impacts of depression in
dementia include decreased ability to perform daily activities of living and higher likelihood of
nursing home placement

(Kales, Chen, Blow, Welsh, & Mellow, 2005; Lee & Lyketsos, 2003).

Furthermore, studies have shown that co-morbid depressive symptoms contribute to an
DEMENTIA, DEPRESSION, AND MORTALITY 7
increased risk of mortality in dementia patients

(Arve, Lehtonen, & Tilvis, 1998; Hoch et al.,
1993; Okura et al., 2011). Findings seem to be mixed possibly due to heterogeneity in
methodology. For example, one study examined a number of potential predictors for mortality
and found no significant association between co-occurring psychiatric symptoms (including
depression) and increased mortality in a community-based sample of AD patients (Larson et al.,
2004). However, studies using a systematic approach to the problem have consistently reported
that co-morbid depression is a significant risk factor for mortality in an older adult population
with varying degrees of cognitive impairment. In a study using a large community sample,
researchers have shown the additive effect of depression and cognitive impairment on mortality

(Mehta et al., 2003). They stratified their sample into 9 subgroups based on the combined
indices of depression and cognitive impairment and found that the group with the most severe
level of both cognitive impairment and depressive symptoms showed a threefold increased risk
for mortality than the group with the lowest levels on both measures (Adjusted HR = 3.1, 95%
CI [2.0, 4.7]). Furthermore, researchers of the Aging, Demographics, and Memory Study
(ADAMS) investigated the relationship between neuropsychiatric symptoms, risk of
institutionalization, and mortality in individuals with cognitive impairment not dementia (CIND)
and those with dementia and showed that depression had a significant association with a greater
risk of institutionalization (HR = 3.06, 95% CI [1.09, 8.59]) and mortality (HR = 1.56, 95% CI
[1.08, 2.26])

(Okura et al., 2011). Of note, this study recruited its participants from a nationally
representative sample of older adults in the U.S. with different severity of cognitive impairment.
However, no prior study thus far has used twin data to look at the apparently harmful effect of
depressive symptoms on survival of people with cognitive impairment and dementia.
With these observations, we conducted two studies. Study 1 examined the association
DEMENTIA, DEPRESSION, AND MORTALITY 8
between dementia and mortality, while Study 2 investigated the association between co-morbid
depressive symptoms in dementia and mortality, employing the co-twin control design. The
design is based on the understanding that all twins, regardless of their zygosity, share 100% of
their rearing environment (except for the few pairs reared separately), and that in principle,
monozygotic (MZ) twins share 100% of their genetic profile while dizygotic (DZ) twins share on
average 50% of segregating genes. Therefore, it allows us to make stronger causal inferences
through examining the association between a risk factor and an outcome in three steps (following
the guidelines detailed in Lichtenstein et al. (2002)). For Study 1, the first step (Step 1) included
the entire sample effectively to perform standard case-control analyses. The next step (Step 2)
tested for the association within monozygotic (MZ) and dizygotic (DZ) pairs discordant for
dementia status. This analysis provides a way to control for potential confounding effects of
unmeasured environmental factors. In the third step (Step 3), the same analysis evaluated the
effect only within the discordant MZ twins, controlling for potential confounding effects of
genetic factors. Additionally, we estimated the risk of mortality for the unaffected twins whose
co-twin was also cognitively normal and the unaffected twins whose co-twin was diagnosed with
dementia and tested whether the results are significantly different (Step 4). We hypothesized in
Step 1, significantly elevated risk for death in twins with dementia, compared with cognitively
normal twins, and in Steps 2 and 3, no significant change in this association, and no significant
difference in Step 4. If these hypotheses were supported, the result would favor the interpretation
that the dementing illness and associated factors, not genetic or early environmental confounds,
accounted for any increased mortality in dementia patients.
For Study 2, we tested whether co-morbid depressive symptoms were related to increased
mortality in individuals with dementia. It was predicted that risk of death would be significantly
DEMENTIA, DEPRESSION, AND MORTALITY 9
greater for demented twins with co-morbid depressive symptoms, compared with demented
twins without any depressive symptoms. Study 2 involved two steps, first a case-control analysis
including all participants with dementia, and second a co-twin-control analysis among twin pairs
where both members of the pair had dementia. We hypothesized that dementia with co-morbid
depression would result in reduced survival compared to dementia without co-morbid
depression.
Method
Participant Population
Participants of the current study comprised individuals included in the Study of Dementia
in Swedish Twins (HARMONY: Gatz et al., 2005). The HARMONY study population is a
subset of twins in the Swedish Twin Registry (STR), which contains data from all twins in
Sweden born since 1886 (Lichtenstein et al., 2002). The HARMONY investigators developed a
large database for dementia research, identifying all living twins in the STR aged 65 and older at
the time of telephone screening. Of the 20,269 twins, 5,733 did not undergo the cognitive
screening process due to refusal to participate, inability to trace, and death, 30 had missing data,
and 594 were ineligible for interview due to deafness or illness. A total of 14,008 twins had their
screening outcome, including 63 individuals who had been screened during a previous
longitudinal cohort study (Gatz et al., 1997).
A total of 2,139 twins were invited to the clinical phase, including: 1) All twins who
screened positive, except those whose co-twin had died before age 65, because the pair would be
uninformative for twin studies, 2) twins who were ineligible for interview because of dementia
per their informant report, 3) co-twins of the index twins diagnosed with dementia, and 4) a
small sample of normal control pairs. Upon completion of the clinical phase, diagnoses were
DEMENTIA, DEPRESSION, AND MORTALITY 10
available for 1,557 twins.
Measures
Study 1: Evaluation of dementia. HARMONY employed a two-stage process,
screening phase followed by clinical workup, to find individuals with dementia. Researchers first
contacted twins and administered a telephone cognitive screening instrument called TELE (Gatz,
Reynolds, Nikolic, Lowe, & Karel, 1995). If a twin demonstrated poor abilities on the TELE, his
or her informant was then interviewed to assess the extent to which the twin’s cognitive abilities
affected daily functioning, using the Blessed Dementia Rating Scale (BDRS; Blessed,
Tomlinson, & Roth, 1968). Twins who screened positive (i.e., cognitive impairment indicated
by the findings of the TELE and/or BDRS) and twins who were not interviewable due to
dementia were subsequently referred for complete clinical workup. Their co-twin was also
referred for clinical workup even if the co-twin’s own screening result was negative.
The clinical workup protocol was modeled after the Consortium to Establish a Registry
for Alzheimer’s Disease (CERAD: Morris et al., 1989) and consisted of physical and
neurological examination, a complete medical history based on medical record review and study
partner interview, a neuropsychological assessment, collection of blood for laboratory tests, and
referral for neuroimaging. An assessment team of a nurse and a physician conducted the
evaluation at the twin’s residence, and each team was blind to zygosity and any information
pertaining to the co-twin.
Based on the workup and review of medical records, a consensus on clinical dementia
diagnosis as well as differential diagnoses was reached through the following three-step
processes: 1) initial diagnosis made by the assessment team, 2) independent diagnosis rendered
by a diagnostic review board, consisting of a neurologist and a psychologist, and 3) resolution of
DEMENTIA, DEPRESSION, AND MORTALITY 11
any disagreements referred back to the diagnostic review board.
Clinical diagnoses of dementia were formulated according to DSM-IV criteria. The
“questionable dementia” category included individuals who did not fulfill one of the first three
DSM-IV diagnostic criteria; specifically, memory impairment, other types of cognitive
impairment, and functional disability. Differential diagnoses were made using the
NINCDS/ADRDA criteria for AD (McKhann et al., 1984), NINDS-AIREN criteria for VaD
(Román et al., 1993), Lund and Manchester criteria for frontotemporal dementia (Lund and
Manchester Groups, 1994; Neary et al., 1998), and consensus criteria for dementia with Lewy
bodies (McKeith et al., 1996).
Study 1: Cases and controls. For the purposes of Study 1, cases included individuals
with a diagnosis of dementia or questionable dementia. In order to address the appropriateness of
categorizing dementia and questionable dementia into the same group, the same analyses were
repeated excluding questionable dementia from cases, and the results were compared. Controls
included individuals with normal cognitive functioning, that is, individuals who screened
negative or had normal results from their clinical workup.
Study 2: Evaluation of depressive symptoms. The HARMONY assessment team
administered the Neuropsychiatric Inventory (NPI) as part of clinical workup battery. The NPI is
a well-validated and widely-used survey of neuropsychiatric and behavioral symptoms in
persons with cognitive impairment (Cummings, 1997). Based on a structured interview with an
informant who can report daily observations of the subject, clinicians collect information on the
presence (yes or no), severity (3-point scale), and frequency (4-point scale) of the current
neuropsychiatric symptoms in 10 domains, including depression. In the HARMONY study, the
NPI Nursing Home version was used for twins who resided in a nursing home at the time of
DEMENTIA, DEPRESSION, AND MORTALITY 12
clinical workup phase. If a twin had two reports of neuropsychiatric symptoms from two
informants (e.g., family member and nursing home staff), then the higher score was used.
Study 2: Cases and controls. For study 2, only individuals with dementia or
questionable dementia were included. Cases were those with depressive symptoms based on the
NPI report at the time of assessment. Controls were those who did not show any depressive
symptoms based on the NPI report at the time of assessment. A secondary analysis required a
greater degree of severity to be classified as a case; cases included individuals who had clinical
levels of depressive symptoms, measured by the product of the NPI Frequency and Severity
scores (F×S) of 4 or higher (Schneider et al., 2001). Controls included individuals who were not
experiencing clinical levels of depressive symptoms, based on the F×S score less than 4.
Mortality. The twin’s vital status and date of death information were obtained through
linkage of the STR to the Swedish Population Register. Individuals were censored at death or the
date of the most recent death data update prior to data analysis, which was November 10, 2010.
Covariates. The following variables were included as covariates in the analyses.
Demographic variables. Age (using the year of birth), sex, and zygosity were considered
covariates. The twin’s zygosity in HARMONY was determined based on self-report, with which
comparisons of DNA markers showed nearly 99% agreement in a random sample of 199 pairs
(Lichtenstein et al., 2002). In the case-control analyses, zygosity was coded as MZ versus DZ. A
relatively small number of twins with indeterminate zygosity (i.e., if twins disagreed on their
zygosity or only one twin reported) were excluded from the analyses. In the co-twin control
analyses, age and zygosity were controlled by design as twins are matched on these variables.
Education. Educational attainment was based on the twin’s self-report on the number of
years of education completed. The number of years of education was entered into the models as a
DEMENTIA, DEPRESSION, AND MORTALITY 13
continuous variable.
Chronic illnesses. Information on chronic illness was derived from either self- or
informant-report health status data, collected as part of the telephone screening. Chronic illnesses
included heart disease, lung disease, diabetes, and cancer, with each condition coded as a binary
variable. Heart disease category encompassed a variety of disorders, such as heart failure, heart
attack, hypertension, and lipid disorder. Lung disease category also included multiple disorders,
such as chronic bronchitis and emphysema. A twin (or an informant) may have indicated the
presence of any number of these disorders in each category. Only three individuals had both self-
and information-report data on history of chronic illnesses, and the reports were in agreement.
CIRS and CDR. As part of clinical workup, the HARMONY assessment team
completed the Cumulative Illness Rating Scale (CIRS; Linn, Linn, & Gurel, 1968) based on
medical information and the Clinical Dementia Rating (CDR; Morris, 1993) Scale based on
structured interviews with the participant and informant. In Study 2, CIRS Severity Index was
used to characterize the sample’s health status and computed as a summary score based on the
average of all the CIRS items except the psychiatric/behavioral system. The CDR provided
information on dementia severity of the Study 2 sample. Mild dementia included a CDR score of
0.5 or 1, moderate dementia a CDR score of 2, and severe dementia a CDR score of 3.
Study Samples
Study 1. The sample for the Study 1 case-control analysis (Step 1) of the current study
consisted of 13,085 individuals, including 781 cases (600 with dementia plus 181 with
questionable dementia) and 12,304 normal controls. A total of 236 twins of indeterminate
zygosity were excluded from the current study because zygosity was one of the variables
included in the analyses (See Figure 1). Three different models were tested in Study 1 case-
DEMENTIA, DEPRESSION, AND MORTALITY 14
control analyses to find the best-fitting model to account for mortality, which is detailed later.
All 13,085 individuals were included in Model 1. Of these, 16 individuals (4 cases and 12
controls) were excluded from Model 2 due to missing educational attainment data. In Model 3,
additional 535 individuals (185 cases and 350 controls) were excluded from the analysis due to
missing data on chronic illnesses covariates. Approximately 24% of cases were excluded from
Model 3 regression analysis due to lack of data on history of chronic illnesses. Preliminary
analyses indicated that older age was a significant predictor of missing data (OR = 1.20, 95% CI
[1.04, 1.10]). In order to address this problem, Model 1 was rerun using the sample included in
Model 3, and these results were compared with the original results.
For the co-twin control analyses (Steps 2 and 3), a total of 237 MZ and DZ pairs were
identified, in which one twin had either dementia or questionable dementia while the co-twin
was cognitively normal (55 MZ, 108 same-sex DZ, and 74 opposite-sex DZ pairs). A total of 48
pairs were excluded from the analyses because of missing chronic illnesses information, leaving
the final sample for the co-twin control analyses to be 189 pairs (41 MZ, 87 same-sex DZ, and
61 opposite-sex DZ pairs) in total.
The sample for comparing controls in Step 4 consisted of 12,304 individuals, including
12,067 unrelated controls and the 237 co-twin controls. Of these, a total of 362 individuals were
excluded due to missing data on covariates (12 for education and 350 for chronic illnesses
information). Thus, the final sample for the Step 4 analyses included 11,942 individuals,
consisting of 11,719 unrelated controls and 223 co-twin controls.
Study 2. The sample comprises those with dementia or questionable dementia. Of the
total of 805 individuals with dementia or questionable dementia, 24 did not have zygosity
determined, and 207 did not have the NPI data, leaving the final sample of 574 individuals for
DEMENTIA, DEPRESSION, AND MORTALITY 15
analysis. Of those missing the NPI data, 40 were screened during the previous longitudinal
cohort study, whose protocol did not include the NPI, 90 died before the assessment, 32 had the
minimum information to establish a diagnosis. Preliminary analyses showed that those without
the NPI data tended to be older (OR = 1.07, 95% CI [1.04, 1.10]). Thus, the final sample for the
Study 2 case-control analysis (Step 1) consisted of 230 cases (i.e., with co-morbid depressive
symptoms) and 344 controls (i.e., without co-morbid depressive symptoms). For the co-twin
control design, a total of 56 twin pairs had matched cognitive status (28 MZ, 23 same-sex DZ,
and 5 opposite-sex DZ). Of these, 17 pairs were discordant for depression, where one twin had
depressive symptoms and the other did not.
Statistical Analyses
The current study used the Statistical Analysis System (SAS version 9.2; SAS Institute,
Inc., Cary, North Carolina) for data analysis. The sample characteristics were presented with
respect to variables, and differences between groups were examined using analysis of variance.
Before running statistical models, as a descriptive step, bivariate correlations were presented to
look at interrelations among variables and to rule out the problem of multicolinearity.
Study 1 case-control analyses (Step 1). Crude incidence rates of death were first
estimated for cases and controls. Using the product-limit method (i.e., Kaplan-Meier method),
median survival time since the age at dementia onset was estimated for all dementias and by
dementia subtypes based on the differential diagnosis. The association between dementia and
mortality was examined, using the Cox proportional hazards regression analysis with the time to
death as the dependent variable. For individuals with dementia, the time of origin was
established using their age at dementia onset, and the time to death was then determined by
calculating their age of onset until their age at death. For controls and individuals with
DEMENTIA, DEPRESSION, AND MORTALITY 16
questionable dementia, the time to death was calculated based on their age at cognitive screening
(i.e., entry into HARMONY) until their age at death. In order to make the time to death estimates
comparable across all individuals, the mean number of years since the age of dementia onset
until the age at cognitive screening was calculated for individuals with dementia, and this value
was added to the time to death for controls and individuals with questionable dementia.
To explore the best-fitting model, three different models were compared. Model 1 (crude
model) included sex, birth year, and zygosity as covariates. Model 2 included sex, birth year,
zygosity, and years of education as covariates. Finally, Model 3 included sex, birth year,
zygosity, years of education, and history of chronic illnesses (heart disease, lung disease,
diabetes, and cancer) as covariates. All statistical models for Step 1 case-control analysis were
adjusted for underlying correlations in twin data, using a robust estimation of the standard error.
The best-fitting model was used for all subsequent analyses in Study 1.
Study 1 co-twin control analyses (Steps 2 and 3). For controls in the co-twin control
analyses, the time of origin was established using the index twin’s age at dementia onset if the
index twin had dementia, or age at cognitive screening if the index twin had questionable
dementia. Therefore, the time to death was calculated based on the number of years each twin
lived since the index twin’s age of onset or age at cognitive screening within dementia discordant
pairs. Using the product-limit method, median survival time since the age at dementia onset (age
at cognitive screening if the index twin had questionable dementia) was estimated for each twin,
and the estimates were compared. Next, stratified Cox proportional hazards regression analyses
were performed to compare cases and controls in their risk of mortality, using each pair as a
striatum. In Step 2, stratified Cox analysis was run using discordant twin sample including both
MZ and DZ pairs, controlling for shared environmental and some genetic influences. In Step 3,
DEMENTIA, DEPRESSION, AND MORTALITY 17
stratified Cox analysis was run using discordant twin sample including only MZ pairs,
controlling for shared environmental and in principle all genetic influences.
Study 1 comparing controls (Step 4). Cox proportional hazards regression estimated
relative risk of mortality, testing differences in survival between healthy twins who had a co-twin
with dementia compared to healthy twins whose co-twins were also healthy. Robust estimation
of the standard error was used to account for underlying correlations in twin data.
Study 2 case-control analyses. The same approach to determine the time to death was
used as Study 1 Step 1 case-control analysis. Cox proportional hazards regression with a robust
estimation of standard error was performed to examine the association between co-morbid
depressive symptoms in dementia and mortality. As described earlier, cases were defined in two
different methods: based on the presence of any depressive symptoms and then on clinical levels
of depressive symptoms. The crude models included sex, birth year, and zygosity as covariates.
Inclusion of other covariates, such as years of education, dementia severity, health status, was to
follow in a stepwise fashion as in Study 1 to evaluate the best-fitting model. We also planned to
conduct stratified Cox hazards regression analyses to test the association between co-morbid
depressive symptoms in dementia and mortality within demented twin pairs discordant for
depression status. However, analyses were discontinued after the first results found no evidence
of an effect due to co-morbid depression.
Results

Study 1
Step 1 case-control analyses. Table 1 presents characteristics of the participants in the
case-control analyses by groups and diagnosis at the time of HARMONY assessment. Both
samples for Model 1 (crude model) and Model 3 (full model) are described in Table 1. Cases
DEMENTIA, DEPRESSION, AND MORTALITY 18
tended to be older at the time of cognitive screening and had less education, compared with
controls. The mean age of dementia onset in this sample was 77.86 years (SD = 7.65) for
individuals with AD and 77.79 years (SD = 7.82) for individuals with VaD. The crude incidence
rates of death were estimated for cases and controls. Results indicated that the incidence rate was
269 deaths per 10,000 person-years for controls (95% CI [261, 276]) and 1,020 deaths per
10,000 person-years for cases (95% CI [949, 1,097]). For individuals diagnosed with dementia,
median survival time was estimated to be 7.9 years (95% CI [7.5, 8.5]) for all dementia types,
8.2 years (95% CI [7.6, 9.2]) for AD, and 6.7 years (95% CI [5.6, 7.7]) for VaD. Table 2 shows
Pearson’s and tetrachoric correlations between variables investigated in Step 1 case-control
analyses and reveals no evidence of multicolinearity.
Results for Models 1, 2, and 3 are shown in Table 3.1. Although hazard ratios were
similar across models, comparisons of the Cox hazards regression models suggested that the full
model (Model 3) with demographic and health covariates was the best-fitting model to predict
mortality in this sample, based on indications of model-fit statistics. Therefore, the full model
(Model 3) was used for all subsequent analyses in Steps 2, 3, and 4. Further, the same set of
comparisons was repeated, excluding individuals with questionable dementia from cases, and the
results are shown in Table 3.2. There was no notable difference in findings from these two sets
of analyses, providing support that it was appropriate to combine dementia and questionable
dementia into the same group for the purposes of the current study (from this point on, this group
is referred to as “individuals with dementia”).
In the test of Model 3, individuals with dementia had a two-fold increase in their risk of
mortality, adjusting for sex, age (using birth year), zygosity, years of education, and a variety of
chronic medical conditions (HR = 2.38 95% CI [2.07, 2.73]). Heart disease, lung disease,
DEMENTIA, DEPRESSION, AND MORTALITY 19
diabetes and cancer were all significant predictors of mortality, but dementia remained a
significant predictor of mortality while taking those chronic diseases into account. In addition,
because individuals with missing covariates data were excluded from the Model 3 analyses,
Model 1 was run again with the sample included in Model 3, and the results were compared in
Table 3.3. The results did not differ substantially.
The association between dementia and increased mortality was examined by sex and age,
and findings are summarized in Table 4. For males, individuals with dementia had a 2.54 times
higher risk of mortality than controls (HR = 2.54, 95% CI [2.08, 3.10]), whereas for females,
individuals with dementia had a 2.29 times higher mortality risk than controls (HR = 2.29, 95%
CI [1.90, 2.76]). The sample was stratified by age using 5-year increment age bands to have six
different age groups, and the relative risks comparing individuals with dementia and controls
within each age band are shown in Table 4. The effect of dementia on risk of death was highest
among those who were younger and lowest, although still significant, among those who were
older. A set of post hoc analyses further revealed that older individuals were more likely than
younger individuals to have heart disease (OR = 1.06, 95% CI [1.05, 1.06]), diabetes (OR = 1.02,
95% CI [1.01, 1.03], and cancer (OR = 1.02, 95% CI [1.01, 1.03].
Steps 2 and 3 co-twin control analyses. In the sample of MZ and DZ pairs discordant
for dementia, median survival time was estimated to be 8.2 years (95% CI [7.6, 9.0]) for cases
and 13.6 years (95% CI [12.7, 14.7]) for controls (Log-rank statistic = 64.3, p < .0001).
Including only the pairs where the index twin was diagnosed with dementia (i.e., not
questionable dementia), median survival time was estimated to be 9.8 years (95% CI [9.2, 10.2])
for twins with dementia and 14.7 years (95% CI [13.6, 15.7]) for their healthy co-twins (Log-
rank statistic = 42.6, p < .0001). Further, in pairs discordant for AD, median survival time was
DEMENTIA, DEPRESSION, AND MORTALITY 20
estimated to be 9.8 years (95% CI [8.7, 10.3]) for twins with AD and 15.1 years (95% CI [11.6,
16.9] for their healthy co-twins (Log-rank statistic = 26.3, p < .0001). There were a total of 138
pairs (39 MZ, 60 same-sex DZ, and 39 opposite-sex DZ) where both members of the twin pair
were deceased. A matched pair t-test tested that twins with dementia had a significantly shorter
lifespan since the age of onset (M = 8.10, SD = 5.01) than their cognitively normal co-twins (M
= 10.15, SD = 5.16), t(138) = -5.68, p<.0001. In Step 2 co-twin control analysis, including
discordant MZ and DZ pairs, a stratified Cox proportional hazards regression analyses found that
the risk of mortality was about four and a half times higher for twins with dementia than for their
cognitively normal co-twins (HR = 4.46, 95% CI [2.88, 6.91]), controlling for shared
environmental and some genetic influences. In Step 3 co-twin control analysis, including only
the discordant MZ pairs, results showed nearly a five-fold increase in the risk of mortality for
twins with dementia, compared with their cognitively normal co-twins (HR = 4.81, 95% CI
[1.65, 14.09]). Figure 2 illustrates and compares findings from Step 1 case-control analyses and
Step 2 and 3 co-twin control analyses.
Step 4 comparing controls. For twins who died, there was no significant difference
between unaffected twins whose co-twin had dementia and unaffected twins without demented
co-twin in their time to death, t(4,931) = .97, p>.05. Further, there was no significant difference
between the two control groups in their risk of mortality (HR = 1.01, 95% CI [0.85, 1.19]).
Study 2
Step 1 case-control analyses. Table 5 presents characteristics of the participants in the
case-control analyses by groups. Findings revealed no significant difference between cases (i.e.,
dementia patients with co-morbid depressive symptoms) and controls (i.e., dementia patients
without co-morbid depressive symptoms) in terms of age at screening and years of education.
DEMENTIA, DEPRESSION, AND MORTALITY 21
The crude incidence rates of death were estimated to be 928 deaths per 10,000 person-years for
controls (95% CI [832, 1036]) and 1,019 deaths per 10,000 person-years for cases (95% CI [892,
1,163]). Cox proportional hazards regression analyses showed that there is no significant
difference between cases and controls in their risk for mortality in the crude model, including
sex, birth year, and zygosity as covariates (HR = 1.14, 95% CI [0.95, 1.37]). Secondary analyses
defining cases as individuals with clinical depression (NPI F×S score of 4 or higher) yielded
similar results (not shown). Therefore, Steps 2 co-twin control analysis was not performed.
Discussion
The purposes of the current study were to examine the effect of dementia on longevity of
patients (Study 1) and whether depressive symptoms moderated this association (Study 2), using
a population-based study of dementia in Swedish twins, known as HARMONY. Findings of
Study 1 showed a significant association between dementia and increased risk of mortality in
both case-control and co-twin control designs. Results of the present study are first to include co-
twin controls, providing a test of the effect of dementia on mortality taking genetic and early
environmental influences into account as well as the measured covariates. The lack of
attenuation of findings compared to case-control analyses indicates that the shortened survival in
dementia is not due to unmeasured genetic or shared early environmental influences on both
dementia and survival.
In the case-control analyses, risk of death was 2.38 times greater for individuals with
dementia than individuals with normal cognition, accounting for the influences of age, sex,
zygosity, years of education, and chronic medical conditions. Results were similar for men and
for women. Relative risk of death in individuals with dementia compared with controls was
highest for younger and lowest for older individuals. This suggests that for those who developed
DEMENTIA, DEPRESSION, AND MORTALITY 22
dementia at younger age, dementia may have a greater impact on their survival than for those
who had dementia later in life. Older individuals are more likely to have other major health risks
that directly affect their survival than younger individuals, for example, in this sample, older age
was associated with higher prevalence of heart disease, diabetes, and cancer.
Following the steps detailed in Lichtenstein et al. (2002), these case-control results were
compared to a series of twin analyses. Examining the association within dementia discordant MZ
and DZ pairs allowed for controlling the influences of shared environmental and some genetic
factors. Results indicated that twins with dementia had risk of death about 4.5 times greater than
their healthy co-twin. Comparisons within dementia discordant MZ pairs tested the effect of
dementia on survival when all unmeasured environmental and genetic factors were taken into
account. Findings showed that the risk was not attenuated in MZ pairs in comparison with all
pairs and that risk of death was about 4.8 times greater in MZ twins with dementia, compared
with their healthy co-twins. Finally, non-demented co-twin controls were compared with
unrelated controls, and the findings showed no difference. Thus, while it has previously been
established that having a twin with dementia increases risk of developing dementia in the healthy
twin (Gatz et al., 2006), it does not otherwise affect the healthy twin’s risk of mortality.
As Study 1 offered the opportunity to look at median survival time within twin pairs, it
was of interest to compare the results of the present study to prior estimates of survival once
diagnosed with dementia. In the present study, the median survival time since the age of disease
onset for patients with dementia was 7.9 years. Specifically, median survival for patients with
AD was 8.2 years, whereas patients with VaD had 6.7 years. These estimates are consistent with
some of the findings from the previous studies, suggesting that patients with VaD have shorter
survival time than those with AD (e.g., Fitzpatrick et al., 2005), and within the ranges the
DEMENTIA, DEPRESSION, AND MORTALITY 23
previous studies reported, summarized to be between 3 to 9 years (Helzner et al., 2008). The
present results are perhaps most comparable to the population-based Canadian Study of Health
and Aging (CSHA), which reported an estimate of 6.6 years median survival after the informant-
reported age of onset of dementia before adjusting for length bias in their sample (Wolfson et al.,
2001). If anything, the CSHA’s report may be an underestimate because individuals excluded
from their analyses due to missing age of onset data were less impaired. Following Fitzpatrick et
al. (2005)’s reasoning, there may be a few plausible explanations of methodological differences
for the rather large variability in reported estimates of survival time in dementia patients. In
particular, age of onset may be based on a formal diagnosis documented in medical records, first
signs of dementia observed by clinicians or neurocognitive measures, or a structured interview
with an informant who is asked to describe when they noted definite and enduring symptoms of
dementia (as was done in both HARMONY and CSHA). Although retrospective estimation of
dementia onset may be problematic, informant may be best positioned to characterize reliable
and persistent changes in a patient’s cognition and daily functioning.
In the analysis where twins diagnosed with dementia were compared with their healthy
co-twins in terms of survival time since the index twin’s age of dementia onset, the difference in
median survival between demented twins and cognitively normal co-twins was 4.9 years for all
dementia and 5.3 years within AD discordant twin pairs. This difference was similar to
Fitzpatrick et al. (2005)’s report on incident dementia cases and their age- and gender-matched
controls, where individuals with AD had a median survival of 7.1 years, whereas controls had a
median survival of 11.0 years.
Thus, the results of Study 1 support the effect of dementia on survival after controlling
for genetic influences, sex, early environment, educational attainment, and co-morbid medical
DEMENTIA, DEPRESSION, AND MORTALITY 24
illnesses, and lend further credence to estimates of survival after diagnosis. These observations
now warrant further investigation of the underlying mechanisms by which dementia shortens
lifespan, such as how the disease unfolds (e.g., biological processes and functional impairments)
to increase one’s likelihood to succumb to death.
Study 2 of the current study failed to replicate the previous findings that co-morbid
depressive symptoms in patients with dementia increase the risk of death beyond the risk posed
by having dementia. It is possible that the effect may be too small in the current study population
and may require a larger sample to detect. Or, it may be that the NPI in this sample fell short on
specificity and was overly inclusive. For example, observers may confuse manifestations of
apathy with symptoms of depression or assume that the patients feel sad because they have
dementia. In the current study sample, 40% of individuals with dementia were reported to have
depressive symptoms. Although there have been reports of such high prevalence of depression in
dementia patients in smaller samples (See Enache, Winblad, and Aarsland, 2011, for review), it
is substantially greater than that of the ADAMS study (Okura et al., 2011), where depression was
found in about 25% in their sample. Finally, a large number of individuals with dementia did not
have NPI data, and thus could not be included in the current study, which may have created bias
in the sample, as well as reducing the sample size.
Strengths of the current study are noteworthy. Study 1 used a large population-based
sample of individuals with dementia as well as individuals with normal cognition to look at the
association between dementia and risk of death. In addition, as highlighted earlier, this is the first
study to use the co-twin design to investigate this association within dementia discordant twin
pairs while holding constant a myriad of measured and unmeasured variables, including genetic
and early environmental influences. However, Study 1 also has some limitations. Firstly, a
DEMENTIA, DEPRESSION, AND MORTALITY 25
relatively large number of individuals with dementia were excluded from the analyses due to
missing covariate data. In order to address this problem, a secondary analysis was performed to
run the same model using the entire sample and subsequently, using the sample with complete
covariate data. Considering that excluded cases tended to be older, it is possible that the risk of
death for cases in the sample with complete covariate data might be an underestimate.
Nonetheless, similarly elevated risk of death for cases in this sample suggested that the effect of
exclusion due to missing data did not substantially affect the results. Secondly, researchers argue
that evaluating the duration of survival using prevalent cases may lead to length bias, a likely
overestimation of survival because individuals with rapidly progressive dementia who died
before the study could not be included (Wolfson et al., 2001). Nevertheless, the current study
provides valuable information for the patient, caregivers, and families with respect to planning
and making decisions for the patient’s remaining life, for “how much time” is one of the first and
most frequently asked questions at the time of diagnosis. Further, data from the current study can
aid future research interested in estimating projected prevalence or examining variables
modifying the prognosis and ultimately, survival in dementia.

DEMENTIA, DEPRESSION, AND MORTALITY 26
References
Alzheimer’s Association (2012). Alzheimer’s Association Report 2012 Alzheimer’s disease facts
and figures. Alzheimer’s & Dementia, 8, 131-168.
Apostolova, L. G. & Cummings, J. L. (2008). Neuropsychiatric manifestations in mild cognitive
impairment: a systematic review of the literature. Dementia Geriatric Cognitive
Disorders, 25, 115-126.
Arve, S., Lehtonen, A., & Tilvis, R. S. (1998). Prognosis of depression with or without dementia.
Archives of Gerontology and Geriatrics, 27(2), 141-146.
Blessed, G., Tomlinson, B. E., & Roth, M. (1968). The association between quantitative
measures of dementia and senile changes in the cerebral gray matter of elderly subjects.
The British Journal of Psychiatry, 114, 797-811.
Cummings, J. (1997). The neuropsychiatric inventory: Assessing psychopathology in dementia
patients. Neurology, 48, 10-16.
Dewey, M. E. & Saz, P. (2001). Dementia, cognitive impairment and mortality in persons aged
65 and over living in the community: a systematic review of the literature. International
Journal of Geriatric Psychiatry, 16, 751-761.
Enache, D., Winblad, B., & Aarsland, D. (2011). Depression in dementia: Epidemiology,
mechanisms, and treatment. Current Opinion in Psychiatry, 24(6), 461-472.
Ferri, C. P., Prince, M., Brayne, C., Brodaty, H., Fratiglioni, L., Ganguli, M., . . . Alzheimer’s
Disease International. (2005). Global prevalence of dementia: a Delphi consensus study.
Lancet, 366, 2112-2117.
Fitzpatrick, A. L., Kuller, L. H., Lopez, O. L., Kawas, C. H., & Jagust, W. (2005). Survival
following dementia onset: Alzheimer’s disease and vascular dementia. Journal of
DEMENTIA, DEPRESSION, AND MORTALITY 27
Neurological Science, 229, 43-49.
Gatz, M., Fratiglioni, L., Johansson, B., Berg, S., Mortimer, J. A., Reynolds, C. A., . . . Pedersen,
N. L. (2005). Complete ascertainment in the Swedish Twin Registry: The HARMONY
study. Neurobiology of Aging, 26, 439-447.
Gatz, M., Pedersen, N. L., Berg, S., Johansson, B., Johansson, K., Mortimer, J. A., . . . Ahlbom,
A. (1997). Heritability for Alzheimer’s disease: The study of dementia in Swedish twins.
Journals of Gerontology Series A – Biological Sciences and Medical Sciences, 52, 117-
125.
Gatz, M., Reynolds, C. A., Fratiglioni, L., Johansson, B., Mortimer, J. A., Berg, S., . . . Pedersen,
N. L. (2006). The role of genes and environments for explaining Alzheimer’s disease.
Archives of General Psychiatry, 63, 1-8.
Gatz, M., Reynolds, C., Nikolic, J., Lowe, B., & Karel, M. (1995). An empirical test of
telephone screening to identify potential dementia cases. International Psychogeriatrics,
9, 429-437.
Gilley, D. W., Bienias, J. L., Wilson, R. S., Bennett, D. A., Beck, T. L., & Evans, D. A. (2004).
Influence of behavioral symptoms on rates of institutionalization for persons with
Alzheimer’s disease. Psychological Medicine, 34, 1129-1135.
Helzner, E. P., Scarmeas, N., Cosentino, S., Tang, M. X., Schupf, N., & Stern, Y. (2008).
Survival in Alzheimer’s disease: A multiethnic, population-based study of incident cases.
Neurology, 71, 1489-1495.
Hoch, C. C., Reynolds III, C. F., Buysse, D. J., Fasiczka, A. L., Houck, P. R., Mazumdar, S., &
Kupfer, D. J. (1993). Two-year survival in patients with mixed symptoms of depression
and cognitive impairment: Comparison with major depression and primary degenerative
DEMENTIA, DEPRESSION, AND MORTALITY 28
dementia. American Journal of Geriatric Psychiatry, 1, 59-66.
Kales, H. C., Chen, P., Blow, F., Welsh, D. E., & Mellow, A. M (2005). Rates of clinical
depression diagnosis, functional impairment, and nursing home placement in coexisting
dementia and depression. American Journal of Geriatric Psychiatry, 13, 441-449.
Knopman, D. S., Rocca, W. A., Cha, R. H., Edland, S. D., & Kokmen, E. (2003). Survival study
of vascular dementia in Rochester, Minnesota. Archive of Neurology, 60, 85-90.
Korczyn, A. D. & Halperin, I. (2009). Depression and dementia. Journal of Neurological
Science, 283, 139-142.
Larson, E. B., Shadlen, M-F, Wang, L., McCormack, W. C., Bowen, J. D., Teri, L., & Kukull,
W. A. (2004). Survival after initial diagnosis of Alzheimer’s disease. Annals of Internal
Medicine, 140, 501-509.
Lavery, L. L., Dodge, H. H., Snitz, B., & Ganguli, M. (2009). Cognitive decline and mortality in
a community-based cohort: The movies project. Journal of American Geriatric Society,
57, 94-100.
Lee, H. B. & Lyketsos, C. G. (2003). Depression in Alzheimer’s disease: heterogeneity and
related issues, Biological Psychiatry, 54, 353-362.
Lichtenstein, P., DeFaire, U., Floderus, B., Svartengren, M., Svedberg, P., & Pedersen,
N. L. (2002). The Swedish Twin Registry: A unique resource for clinical,
epidemiological and genetic studies. Journal of Internal Medicine, 252, 184-205.
Linn, B. S., Linn, M. W., & Gurel, L. (1968). Cumulative illness rating scale. Journal of
American Geriatrics Society, 16, 622-626.
Lund and Manchester Groups. (1994). Clinical and neuropathological criteria for
frontotemporal dementia. Journal of Neurology and Neurosurgical Psychiatry, 57, 416-
DEMENTIA, DEPRESSION, AND MORTALITY 29
418.
Lyketsos C. G., Lopez, O., Jones, B., Fitzpatrick, A. L., Breitner, J., & DeKosky, S. (2002).
Prevalence of neuropsychiatric symptoms in dementia and mild cognitive impairment.
Journal of American Medical Association, 288, 1475-1483.
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 under the auspices of Department of Health and Human Services task force on
Alzheimer’s disease. Neurology, 34, 939-944.
McKeith, I. G., Galasko, D., Kosaka, K., Perry, E. K., Dickson, D. W., Hansen, L. A., . . . Perry,
R. H. (1997). Consensus guidelines for the clinical and pathological diagnosis of
dementia with Lewy bodies (DLB): Report of the consortium on DLB international
workshop. Neurology, 47, 1113-1124.
Mehta, K. M., Yaffe, K., Langa, K. M., Sands, L., Whooley, M. A., & Covinsky, K. E. (2003).
Additive effects of cognitive function and depressive symptoms on mortality in elderly
community-living adults. Journals of Gerontology Series A – Biological Sciences and
Medical Sciences, 58, 461-467.
Mölsä, P. K., Marttila, R. J., & Rinne, U. K. (1986). Survival and cause of death in Alzheimer’s
disease and multi-infarct dementia. Acta Neurologica Scandinavica, 74, 103-107.
Morris, J. C., Heyman, A., Mohs, R. C., Hughes, J. P., van Belle, G., Fillenbaum, G., . . . Clark,
C. (1989). The Consortium to Establish a Registry for Alzheimer’s Disease (CERAD):
Part I. Clinical and neuropsychological assessment of Alzheimer’s disease. Neurology,
39, 1159-1165.
Morris, J. C. (1993). The clinical dementia rating (CDR): Current version and scoring rules.
DEMENTIA, DEPRESSION, AND MORTALITY 30
Neurology, 43, 2412-2414.
Neale, R., Brayne, C., & Johnson, A. L. (2001). Cognition and survival: an exploration in a large
multicentre study of the population aged 65 years and older. International Journal of
Epidemiology, 30, 1383-1388.
Neary D., Snowden, J. S., Gustafson, L., Passant, U., Stuss, D., Black, S., . . . Benson, D. F.
(1998). Frontotemporal lobar degeneration: A consensus on clinical diagnostic criteria.
Neurology, 51, 1546-1554.
Okura, T., Plassman, B. L., Steffens, D. C., Llewellyn, D. J., Potter, G. G., & Langa, K. M.
(2011). Neuropsychiatric symptoms and the risk of institutionalization and death: The
aging, demographics, and memory study. Journal of American Geriatrics Society, 59,
478-481.
Rapp, M. A., Schnaider-Beeri, M., Wysocki, M., Guerrero-Berroa, E., Grossman, H. T., Heinz,
A., & Haroutunian, V. (2011). Cognitive decline in patients with dementia as a function
of depression. American Journal of Geriatric Psychiatry, 19, 357-363.
Román, G. C., Tatemichi, T. K., Erkinjuntti, T., Cummings, J. L., Masdeu, J. C., Garcia,
J. H., . . . Scheinberg, P. (1993). Vascular dementia: Diagnostic criteria for research
studies. Report of the NINDS-AIREN international workshop. Neurology, 43, 256-260.
Saz, P. & Dewey, M. E. (2001). Depression, depressive symptoms, and mortality in persons aged
65 and over living in the community: a systematic review of the literature. International
Journal of Geriatric Psychiatry, 16, 622-630.
Schneider, L. S., Tariot, P. N., Lyketsos, C. G., Dagerman, K. S., Davis, K. L., Hsiao, J. K. . . .
Lieberman, J. A. (2001). NIMH clinical antipsychotic trials of intervention effectiveness
(CATIE): Alzheimer disease trial methodology. American Journal of Geriatric
DEMENTIA, DEPRESSION, AND MORTALITY 31
Psychiatry, 9(4), 346-360.
Schulz, R., Drayer, R. A., & Rollman, B. L. (2002). Depression as a risk factor for non-suicide
mortality in the elderly. Society of Biological Psychology, 52, 205-225.
Waring, S. C., Doody, R. S., Pavlik, V. N., Massman, P. J., & Chan, W. (2005). Survival among
patients with dementia from a large multi-ethnic population. Alzheimer Disease and
Associated Disorders, 19, 178-183.
Wolfson, C., Wolfson, D. B., Asgharian, M., M’Lan, C. E., Østbye, T., Rockwood, K., & Hogan,
D. B. (2001). A reevaluation of the duration of survival after the onset of dementia. The
New England Journal of Medicine, 344(15), 1111-1116.

DEMENTIA, DEPRESSION, AND MORTALITY 32

Table 1

Study 1 participant characteristics of the case-control analyses (Step 1) by cognitive status and diagnosis

Cases Dementia by Diagnosis

Controls Cases Questionable
Dementia
Dementia AD VaD Other

Model 1
Total n 12,304 781 181 600 373 122 105
Died n 4,837 739 167 572 356 114 102
(%) (39.3) (94.6) (92.3) (95.3) (95.4) (93.4) (97.1)
Age at screening* M 73.01 81.77 79.98 82.31 82.91 81.90 80.64
(SD) (6.27) (6.72) (6.90) (6.58) (6.45) (6.40) (6.96)
Sex
Female n 6,840 491 89 402 268 70 64
(%) (55.6) (62.9) (49.2) (67.0) (71.9) (57.4) (61.0)
Male n 5,464 290 92 198 105 52 41
(%) (44.4) (37.1) (50.8) (33.0) (28.1) (42.6) (39.0)
Zygosity
MZ n 2,952 242 52 190 127 33 30
(%) (24.0) (31.0) (28.7) (31.7) (34.1) (27.0) (28.6)
Same-sex DZ n 5,145 366 77 289 182 62 45
(%) (41.8) (46.9) (42.5) (48.2) (48.8) (50.8) (42.9)
Opposite-sex DZ n 4,207 173 52 410 64 27 30
(%) (34.2) (22.1) (28.7) (20.2) (17.2) (22.1) (28.5)

Model 3
Total n 11,942 592 164 428 267 83 78
Died n 4,619 554 150 404 253 76 75
(%) (38.7) (93.6) (91.5) (94.4) (94.8) (91.6) (96.2)
Age at screening* M 72.93 81.01 79.78 81.49 82.21 80.46 80.10
(SD) (6.21) (6.67) (6.86) (6.54) (6.36) (6.14) (7.23)
Years of education* M 8.62 7.29 7.60 7.17 7.30 6.67 7.26
(SD) (3.00) (2.35) (2.49) (2.29) (2.43) (1.63) (2.31)
Sex
Female n 6,607 357 81 276 187 43 46
(%) (55.3) (60.3) (49.4) (64.5) (70.0) (51.8) (59.0)
Male n 5,335 235 83 152 80 40 32
(%) (44.7) (39.7) (50.6) (35.5) (30.0) (48.2) (41.0)
Zygosity
MZ n 2,863 175 46 129 87 22 20
(%) (24.0) (29.5) (28.1) (30.1) (32.6) (26.5) (25.6)
Same-sex DZ n 4,984 265 69 196 127 37 32
(%) (41.7) (44.8) (42.1) (45.8) (47.6) (44.6) (41.0)
Opposite-sex DZ n 4,095 152 49 103 53 24 26
(%) (34.3) (25.7) (29.9) (24.1) (19.8) (28.9) (33.3)
Heart Disease
Yes n 7,030 496 115 381 235 78 68
(%) (58.9) (83.7) (70.1) (89.0) (88.0) (94.0) (87.2)
No n 4,912 96 49 47 32 5 10
(%) (41.1) (16.3) (29.9) (11.0) (12.0) (6.0) (12.8)
DEMENTIA, DEPRESSION, AND MORTALITY 33

Lung Disease

Yes n 859 29 12 17 11 2 4
(%) (7.2) (4.9) (7.3) (4.0) (4.1) (2.4) (5.1)
No n 11,083 563 152 411 256 81 74
(%) (92.8) (95.1) (92.7) (96.0) (95.9) (95.6) (94.9)
Diabetes
Yes n 1,155 121 29 92 47 23 22
(%) (9.7) (20.4) (17.7) (21.5) (17.6) (27.7) (28.2)
No n 10,787 471 135 336 220 60 56
(%) (90.3) (79.6) (82.3) (78.5) (82.4) (72.3) (71.8)
Cancer
Yes n 1,251 53 8 45 23 10 12
(%) (10.5) (9.0) (4.9) (10.5) (8.6) (12.1) (15.4)
No n 10,691 539 156 383 244 73 66
(%) (89.5) (91.0) (95.1) (89.5) (91.4) (87.9) (84.6)

Note. * Case versus control significant at p<.0001. Only 16 individuals from Model 1 were excluded from Model 2
analyses due to missing education data. Therefore, Model 2 sample information was not provided here.

DEMENTIA, DEPRESSION, AND MORTALITY 34
Table 2

Correlation matrix between variables for Study 1 case-control analyses (Step 1)

TTD
a
case
b
birthyr
c
sex
d
zygos
e
educ
f
heart
g
lung
h
diab
i
cancer
j
TTD 1
case -.3479 1
birthyr .4225 -.3105 1
sex .0887 .0884 -.0721 1
zygos .0338 -.1002 .0563 -.0461 1
educ .1163 -.1057 .2080 -.0797 -.0171 1
heart -.1634 .3282 -.1583 .0137 -.0006 -.0700 1
lung -.0517 -.0814 -.0052 -.0026 -.0119 .0004 .1446 1
diabetes -.1284 .2188 -.0352 -.0526 .0011 -.0551 .3423 .0152 1
cancer -.0965 -.0325 -.0360 .1134 -.0293 .0274 .0532 .0517 .0460 1

Note.
a
Time to death;
b
0=control, 1=case;
c
birth year;
d
1=male, 2=female;
e
zygosity: 1=MZ, 2=DZ;
f
years of
education;
g
history of heart disease: 0=no, 1=yes;
h
history of lung disease: 0=no, 1=yes;
i
history of diabetes: 0=no,
1=yes;
j
history of cancer: 0=no, 1=yes.

DEMENTIA, DEPRESSION, AND MORTALITY 35
Table 3.1
Comparisons of Cox proportional hazards regression models for Study 1 case-control analyses (Step 1)

Model 1 (n = 13,085) Model 2 (n = 13,069) Model 3 (n = 12,534)
Parameter Estimates and Standard Errors
Case 0.8858 (0.0611)* 0.8704 (0.0615)* 0.8649 (0.0704)*
Birth year -0.1068 (0.0023)* -0.1050 (0.0023)* -0.1063 (0.0025)*
Sex -0.4836 (0.0290)* -0.4926 (0.0292)* -0.5397 (0.0297)*
Zygosity -0.0074 (0.0340) -0.0164 (0.0340) 0.0010 (0.0357)
Years of education -0.0243 (0.0050)* -0.0225 (0.0054)*
Heart disease 0.3387 (0.0311)*
Lung disease 0.3331 (0.0550)*
Diabetes 0.4275 (0.0498)*
Cancer 0.4369 (0.0462)*

Hazard Ratios and 95% Confidence Intervals
Case 2.43 [2.15, 2.73] 2.39 [2.12, 2.69] 2.38 [2.07, 2.73]
Birth year 0.90 [0.90, 0.90] 0.90 [0.90, 0.90] 0.90 [0.90, 0.90]
Sex 0.62 [0.58, 0.65] 0.61 [0.57, 0.65] 0.58 [0.55, 0.62]
Zygosity 0.99 [0.93, 1.06] 0.98 [0.92, 1.05] 1.00 [0.93, 1.07]
Years of education 0.98 [0.97, 0.98] 0.98 [0.97, 0.99]
Heart disease 1.40 [1.32, 1.49]
Lung disease 1.40 [1.25, 1.55]
Diabetes 1.53 [1.39, 1.69]
Cancer 1.55 [1.41, 1.70]

Goodness-of-fit Statistics
n parameters 4 5 9
-2LL 96762.09 96514.18 89168.68
AIC 96770.09 96524.18 89186.68
BIC 96796.59 96557.30 89245.64s

Note. * p<.0001.

DEMENTIA, DEPRESSION, AND MORTALITY 36
Table 3.2

Comparisons of Cox proportional hazards regression models for Study 1 case-control analyses (Step 1), excluding
questionable dementia from cases.

Model 1 (n = 12,904) Model 2 (n = 12,889) Model 3 (n = 12,370)
Parameter Estimates and Standard Errors
Case 0.8525 (0.0745)* 0.8106 (0.0760)* 0.8410 (0.0913)*
Birth year -0.1084 (0.0023)* -0.1074 (0.0024)* -0.1079 (0.0026)*
Sex -0.4823 (0.0295)* -0.4993 (0.0299)* -0.5389 (0.0301)*
Zygosity -0.0063 (0.0345) -0.0182 (0.0345) 0.0015 (0.0363)
Years of education -0.0283 (0.0050)* -0.0227 (0.0056)*
Heart disease 0.3378 (0.0316)*
Lung disease 0.3371 (0.0563)*
Diabetes 0.4386 (0.0507)*
Cancer 0.4329 (0.0467)*

Hazard Ratios and 95% Confidence Intervals
Case 2.35 [2.03, 2.71] 2.31 [1.99, 2.68] 2.32 [1.94, 2.77]
Birth year 0.90 [0.89, 0.90] 0.90 [0.90, 0.90] 0.90 [0.89, 0.90]
Sex 0.62 [0.58, 0.65] 0.61 [0.58, 0.65] 0.58 [0.55, 0.62]
Zygosity 0.99 [0.93, 1.06] 0.98 [0.92, 1.05] 1.00 [0.93, 1.08]
Years of education 0.98 [0.97, 0.99] 0.98 [0.97, 0.99]
Heart disease 1.40 [1.32, 1.49]
Lung disease 1.40 [1.26, 1.57]
Diabetes 1.55 [1.40, 1.71]
Cancer 1.54 [1.41, 1.69]

Goodness-of-fit Statistics
n parameters 4 5 9
-2LL 93902.63 93670.86 86618.61
AIC 93910.63 93680.86 86636.61
BIC 93937.01 93713.83 86695.31

Note. * p<.0001.

DEMENTIA, DEPRESSION, AND MORTALITY 37
Table 3.3

Comparing results of Model 1 using the original sample (n = 13,085) and Model 3 sample (n = 12,534)

Model 1

Model 1 using Model 3 sample
Parameters Estimate (SE) HR [95% CI]

Estimate (SE) HR [95% CI]
Case 0.8525 (0.0745)* 2.35 [2.03, 2.71]

0.9699 (0.0667)* 2.64 [2.31, 3.01]
Birth year -0.1084 (0.0023)* 0.90 [0.89, 0.90]

-0.1100 (0.0024)* 0.90 [0.89, 0.90]
Sex -0.4823 (0.0295)* 0.62 [0.58, 0.65]

-0.5051 (0.0292)* 0.60 [0.57, 0.64]
Zygosity -0.0063 (0.0345) 0.99 [0.93, 1.06]

0.0126 (0.0352) 1.01 [0.95, 1.09]

DEMENTIA, DEPRESSION, AND MORTALITY 38
Table 4

Cox proportional hazards regression analyses predicting mortality by sex and age

By sex* Males (N=5,570) Females (N=6,964)

HR 95% CI HR 95% CI

Control

1.00 -- -- 1.00 -- --
Case

2.54 2.08 3.10 2.29 1.90 2.76

By age§ 65 ≤ Age < 70 (N=4,336) 70 ≤ Age < 75 (N=3,424) 75 ≤ Age < 80 (N=2,535)

HR 95% CI HR 95% CI HR 95% CI

Control

1.00 -- -- 1.00 -- -- 1.00 -- --
Case

4.49 2.72 7.44 4.05 2.97 5.54 3.40 2.68 4.30

80 ≤ Age < 85 (N=1,418) 85 ≤ Age < 90 (N=639) Age ≥ 90 (N=182)

HR 95% CI HR 95% CI HR 95% CI

Control

1.00 -- -- 1.00 -- -- 1.00 -- --
Case

2.33 1.84 2.93 2.06 1.55 2.73 1.67 1.14 2.43

Note. * Covariates: birth year, years of education, zygosity, chronic medical illnesses; § Covariates: sex, years of
education, zygosity, chronic medical illnesses

DEMENTIA, DEPRESSION, AND MORTALITY 39
Table 5

Study 2 participant characteristics of the case-control analyses

No Depressive Symptoms

Depressive Symptoms

Total n

344

230
Died

n (%)

320 (93.0)

219 (95.2)
Age at screening

M (SD)

80.94 (6.77)

81.08 (6.39)
Sex

n (%)

Female

207 (60.2)

148 (64.3)
Male

137 (39.8)

82 (35.7)
Zygosity

n (%)

Monozygotic

101 (29.4)

76 (33.0)
Dizygotic

243 (69.7)

154 (67.0)
Years of education

M (SD)

7.41 (2.53)

7.30 (2.36)
CIRS Severity Index

n

337

224

M (SD)

1.54 (0.32)

1.58 (0.32)
CDR

n

343

230
Mild (CDR total ≤ 1)

n (%)

166 (48.4)

85 (37.0)
Moderate (CDR total = 2)

78 (22.7)

91 (39.6)
Severe (CDR total = 3)

99 (28.9)

54 (23.5)

Note. CIRS = Cumulative Illness Rating Scale. CDR = Clinical Dementia Rating Scale

DEMENTIA, DEPRESSION, AND MORTALITY 40

Figure 1. Participants for
Study 1 case-control analyses

20,269 aged 65+

5,733 dropouts
• 253 died before screening
• 1,182 untraceable
• 4,298 refusal

30 missing cognitive data

594 not interviewable

6,261 without screening outcome info

687 screened positive but no clinical diagnosis

488 refused work-up
z
186 not referred by design

13 untraceable

13,321 cognitive status determined

1,557 by clinical work-up

620 Dementia

185 Questionable
Dementia

752 Not
demented

11,764 screened negative, not referred

242 MZ
24
Indeterminate

539 DZ

2,952 MZ
212
Indeterminate

9,352 DZ

781 Cases 12,304 Controls
DEMENTIA, DEPRESSION, AND MORTALITY 41
Figure 2. Study 1 hazard ratios comparing cases and controls from the case-control (Step 1), MZ and DZ co-twin
control (Step 2), and MZ only co-twin control analyses (Step 3).

14
13
12
11
10
9
8
7
6
5
4
3
2
1
0
Case-control
(N=12,534
individuals)
MZ and DZ
Co-twin control
(N=189 pairs)
MZ
Co-twin Control
(N=41 pairs)
Controls

Cases

DEMENTIA, DEPRESSION, AND MORTALITY 42
Appendix

Results from stratified Cox hazards regression analyses using three models in Step 2 co-twin control analyses.
Model 1 included sex as a covariate (N pairs = 237). Model 2 included sex and education as covariates (N pairs =
237). Model 3 included sex, education, and chronic illnesses as covariates (N pairs = 189).

Model 1 Model 2 Model 3

HR 95% CI HR 95% CI HR 95% CI

Control 1.00 -- -- 1.00 -- -- 1.00 -- --
Case 4.45 3.11 6.36 4.44 3.10 6.36 4.46 2.88 6.91

Abstract (if available)

Abstract Dementia and cognitive impairment are known to increase the risk of mortality in older adults. It has been suggested as well that depressive symptoms in persons with dementia further elevate their risk of mortality. However, these associations have not been investigated in a genetically informative sample. Using a population-based sample of twins in the Study of Dementia in Swedish Twins (HARMONY), Study 1 of the present study examined the association between dementia and mortality in the entire sample as well as within pairs of twins discordant for dementia status. Study 2 tested the association between co-morbid depressive symptoms and mortality in twins with dementia. In Study 1, Cox proportional hazards regression analysis revealed that individuals with dementia had more than doubled risk of death, compared with individuals with normal cognition (HR=2.38, 95% CI [2.07, 2.73]). As hypothesized, comparisons within dementia discordant pairs also showed significantly greater risk of death for those with dementia, indicating that genetic and early environmental influences did not account for the increased risk of death in individuals with dementia. Indeed, median survival time estimated since the demented twin’s age of onset was 9.8 years for twins with dementia and 14.7 years for their cognitively normal co-twins. Study 2 failed to replicate previous studies and found no effect of co-morbid depressive symptoms on mortality in twins with dementia. The present study is the first to look at survival of dementia patients using the co-twin design. These findings may be informative to patients, families, clinicians, and policy makers in making decisions for dementia care.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

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

PDF

Are life events differentially associated with dementia risk by gender? A twin study

PDF

Sources of stability and change in the trajectory of openness to experience across the lifespan

PDF

Underlying neural mechanisms of depression and dementia

PDF

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

PDF

Neuropsychological functioning in older adult long-term cancer survivors

PDF

Personality and cognitive aging: are positive traits protective?

PDF

Perceptions of dementia among Latinos

PDF

Cognitive functioning following ovarian removal before or after natural menopause

PDF

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

PDF

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

PDF

Untreated hyperglycemia associated with tau pathology and worse cognitive performance in older adults

PDF

Psychosocial adjustment among patients with metastatic colorectal cancer

PDF

Communication and home-based interventions in Alzheimer’s disease -- a review

PDF

Life transitions, leisure activity engagement, and cognition among older adults

PDF

Using observed peer discussions to understand adolescent depressive symptoms and interpersonal interactions

PDF

A study of depression disclosure among Latino older adults in Los Angeles

PDF

Examining the effect of fertility on lifespan using data from the Swedish Twin Registry

PDF

Dedifferentiation of emotion regulation strategies in the aging brain: an MVPA investigation

PDF

Brainstem structural integrity in the progression of Alzheimer's disease

Asset Metadata

Creator Jang, Jung Yun (author)

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

School College of Letters, Arts and Sciences

Degree Master of Arts

Degree Program Psychology

Publication Date 04/18/2013

Defense Date 11/09/2012

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

Tag dementia,Depression,mortality,OAI-PMH Harvest,older adults,Survival,twin

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Gatz, Margaret (committee chair), Knight, Bob G. (committee member), Pedersen, Nancy L. (committee member)

Creator Email jungyjan@usc.edu,jungyun.jang@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c3-239177

Unique identifier UC11294673

Identifier etd-JangJungYu-1561.pdf (filename),usctheses-c3-239177 (legacy record id)

Legacy Identifier etd-JangJungYu-1561.pdf

Dmrecord 239177

Document Type Thesis

Format application/pdf (imt)

Rights Jang, Jung Yun

Type texts

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

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA