University of Southern California Dissertations and Theses

Subclinical carotid atherosclerosis, psychosocial measures, and cognitive function in middle- to older-aged adults

(USC Thesis Other)

Subclinical carotid atherosclerosis, psychosocial measures, and cognitive function in middle- to older-aged adults

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content SUBCLINICAL CAROTID ATHEROSCLEROSIS, PSYCHOSOCIAL MEASURES, AND COGNITIVE FUNCTION IN
MIDDLE- TO OLDER-AGED ADULTS

by

Felice Lin, MPH

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

December 2022

Copyright 2022 Felice Lin
ii

ACKNOWLEDGEMENTS
I would like to thank my committee chair, Dr. Mack, for her invaluable advice, support, and patience
during the course of my doctoral studies. My appreciation extends to my committee members - Drs.
Han, Hodis, Karim, and Pa - for their expertise and mentorship. I would also like to thank my family - my
son Alex, our dog Chloe, and especially my husband Kenny - for his understanding and patience, and
without whom this would not have been possible. My gratitude also goes out to my parents and friends
for their encouragement and support throughout my studies.

iii

TABLE OF CONTENTS

ACKNOWLEDGEMENTS ................................................................................................................................. ii
LIST OF TABLES ............................................................................................................................................. vi
ABBREVIATIONS .......................................................................................................................................... viii
ABSTRACT ...................................................................................................................................................... x

CHAPTER 1: INTRODUCTION AND BACKGROUND ON ATHEROSCLEROSIS AND COGNITIVE FUNCTION ..... 1
Cognitive impairment in elderly persons .................................................................................................. 1
Dementia as a growing public health concern ..................................................................................... 1
Prevalence and incidence of mild cognitive impairment ..................................................................... 2
Risk factors for mild cognitive impairment ........................................................................................... 4
Subclinical atherosclerosis and mild cognitive impairment ................................................................. 5
Subclinical atherosclerosis and cognitive function ............................................................................... 6
CIMT and executive function ........................................................................................................... 7
CIMT and verbal memory ................................................................................................................ 8
CIMT and visual memory ................................................................................................................. 9
CIMT and global cognition ............................................................................................................... 9

CHAPTER 2: SUBCLINCIAL CAROTID ATHEROSCLEROSIS AND COGNITIVE FUNCTION IN MIDDLE- TO
OLDER-AGED ADULTS ................................................................................................................................. 12
INTRODUCTION ....................................................................................................................................... 12
MATERIALS AND METHODS .................................................................................................................... 14
Study design ........................................................................................................................................ 14
BVAIT design .................................................................................................................................. 14
WISH design ................................................................................................................................... 15
ELITE design ................................................................................................................................... 16
Carotid artery intima-media thickness assessment ............................................................................ 17
Cognitive function assessment ........................................................................................................... 17
Statistical methods ............................................................................................................................. 18
Cross-sectional associations between CIMT and cognition at baseline and 2.5 years .................. 19
Associations of CIMT with cognitive change over 2.5 years .......................................................... 21
RESULTS ................................................................................................................................................... 21
Study sample ....................................................................................................................................... 21
Associations between CIMT and cognitive function at baseline and 2.5 years .................................. 24
Associations between baseline CIMT and change in cognitive function assessed over 2.5 years ..... 27
DISCUSSION ............................................................................................................................................. 31
iv

CONCLUSION ........................................................................................................................................... 34

CHAPTER 3: INTRODUCTION AND BACKGROUND ON SOCIAL SUPPORT AND COGNITIVE FUNCITON ...... 36
Social support and cognitive function..................................................................................................... 36
Measures of social support ..................................................................................................................... 37
Validated instruments ........................................................................................................................ 37
Indirect measures of social support.................................................................................................... 38
Previous findings ..................................................................................................................................... 38
Associations between social support and overall cognitive function ................................................. 38
Associations between social support and cognitive domains ............................................................ 39
Associations between dimensions of social support and cognitive domains..................................... 40
Association between social support and cognitive function may differ depending on the source of
support ................................................................................................................................................ 41
Social isolation, loneliness, and cognitive function ............................................................................ 42
Study variations that may contribute to different findings .................................................................... 42
Potential issues with reverse causation .................................................................................................. 43

CHAPTER 4: SOCIAL SUPPORT AND COGNITIVE FUNCTION IN POST-MENOPAUSAL WOMEN .................. 44
INTRODUCTION ....................................................................................................................................... 44
MATERIALS AND METHODS .................................................................................................................... 45
Study design ........................................................................................................................................ 45
ELITE design ................................................................................................................................... 45
ELITE psychosocial substudy .......................................................................................................... 46
Social support assessment .................................................................................................................. 47
Cognitive function assessment ........................................................................................................... 48
Statistical methods ............................................................................................................................. 49
RESULTS ................................................................................................................................................... 53
Study sample ....................................................................................................................................... 53
Associations between perceived social support and cognitive function ............................................ 56
DISCUSSION ............................................................................................................................................. 62
CONCLUSION ........................................................................................................................................... 65

CHAPTER 5: INTRODUCTION AND BACKGROUND ON STRESS AND COGNITIVE FUNCITON ...................... 66
Stress and cognitive function .................................................................................................................. 66
Previous findings ..................................................................................................................................... 66
Stressful life events and cognitive function ........................................................................................ 66
Global levels of stress and cognitive function .................................................................................... 68
v

Potential mechanisms ............................................................................................................................. 69
Limitations of previous studies ............................................................................................................... 71

CHAPTER 6: PERCEIVED STRESS AND COGNITIVE FUNCTION IN POST-MENOPAUSAL WOMEN ............... 72
INTRODUCTION ....................................................................................................................................... 72
MATERIALS AND METHODS .................................................................................................................... 74
Study design ........................................................................................................................................ 74
Perceived stress assessment............................................................................................................... 74
Assessment of stressful life events ..................................................................................................... 75
Cognitive function assessment ........................................................................................................... 75
Statistical methods ............................................................................................................................. 77
Associations between perceived stress and cognition .................................................................. 79
Associations between stressful life events and cognition ............................................................. 80
RESULTS ................................................................................................................................................... 82
Study sample ....................................................................................................................................... 82
Associations between PSS total and cognitive function ..................................................................... 86
Associations between PERI life events and cognitive function .......................................................... 87
DISCUSSION ............................................................................................................................................. 90
CONCLUSION ........................................................................................................................................... 95
Limitations of previous studies ............................................................................................................... 71

REFERENCES ................................................................................................................................................ 96

APPENDICES .............................................................................................................................................. 111
APPENDIX A: Medical Outcomes Study-Social Support Survey (MOS-SSS) ...................................... 112
APPENDIX B: Perceived Stress Scale (PSS) ........................................................................................ 113
APPENDIX C: PERI Events Scale ......................................................................................................... 114

vi

LIST OF TABLES
Table 2.1 Visits by study .................................................................................................................... 19
Table 2.2 Baseline characteristics for study participants by clinical trial (n = 1,495) .............................. 23
Table 2.3 Cross-sectional associations between CIMT and cognitive function at baseline
and 2.5 years from multivariable linear regression models .................................................................. 25
Table 2.4 Cross-sectional associations between CIMT and cognitive function at baseline
and 2.5 years from multivariable linear regression models excluding contributors to CIMT
(BMI, SBP, HDL, LDL) ......................................................................................................................... 27
Table 2.5 Associations between baseline CIMT and change in cognitive function assessed
over 2.5 years from multivariable linear regression models ................................................................. 29
Table 2.6 Longitudinal associations between CIMT and cognitive function over 2.5 years
from multivariable linear regression models excluding contributors to CIMT
(BMI, SBP, HDL, LDL) ......................................................................................................................... 30

Table 4.1 Cognitive and psychosocial visits completed within 6 months of each other .......................... 51
Table 4.2a Summary of the number of completed psychosocial visits in the ELITE cohort ..................... 53
Table 4.2b Summary of the number of completed psychosocial visits in the ELITE cohort
when restricted to cognitive-psychosocial visits completed within 6 months of each other ................... 53
Table 4.3 Baseline characteristics for ELITE women included in the cognitive-psychosocial
analysis (n = 335) .............................................................................................................................. 55
Table 4.4 Average transformed social support scores (possible range 0-100) for
cognitive-psychosocial paired visits completed within 6 months of each other (n = 392) ....................... 56
Table 4.5a Associations from linear mixed effects models between transformed social
support scores and cognitive outcomes, limited to cognitive-psychosocial visits completed
within 6 months of each other (n = 322) ............................................................................................ 57
Table 4.5b Associations from linear mixed effects models between transformed social
support scores and cognitive outcomes without marital status, limited to
cognitive-psychosocial visits completed within 6 months of each other (n = 322) ................................. 58
Table 4.6a Summary of interaction by age groups from linear mixed effects models
between transformed positive social interaction score and cognitive outcomes, limited
to cognitive-psychosocial visits completed within 6 months of each other (n = 322) ............................. 59
Table 4.6b Associations from linear mixed effects models between transformed positive
social interaction score and cognitive outcomes by age group ............................................................. 60
Table 4.7a Correlations (r) between number of close friends and relatives and transformed
social support scores (n = 382) .......................................................................................................... 62
Table 4.7b Correlations (r) between number of close friends and relatives and cognitive
outcomes (n = 382) ........................................................................................................................... 62

Table 6.1 Cognitive and psychosocial visits completed within 6 months of each other .......................... 79
Table 6.2a Summary of the number of completed psychosocial visits in the ELITE cohort ..................... 82
Table 6.2b Summary of the number of completed psychosocial visits in the ELITE cohort
vii

when restricted to cognitive-psychosocial visits completed within 6 months of each other ................... 82
Table 6.3 Baseline characteristics for ELITE women included in the cognitive-psychosocial
analysis (n = 335) .............................................................................................................................. 84
Table 6.4 Summary of the average number of PERI events overall and by category that
were rated as not undesirable or undesirable among women who reported an event .......................... 85
Table 6.5 Associations from univariate mixed effects models between PSS total both as a
continuous variable and as a categorical variable based on quintiles and the total number
of PERI life events from cognitive-psychosocial paired visits completed within 6 months
of each other (n = 394) ..................................................................................................................... 86
Table 6.6 Associations from linear mixed effects models between PSS total both as a
continuous variable and as a categorical variable based on quintiles and cognitive outcomes
for cognitive-psychosocial paired visits completed within 6 months of each other (n = 286).................. 87
Table 6.7 Associations from linear mixed effects models between the total number of PERI
events, the different PERI categories, and cognitive outcomes for cognitive-psychosocial
paired visits completed within 6 months of each other (n = 286) ......................................................... 89
Table 6.8 Associations from linear mixed effects models between the total number of
not undesirable events, the total number of undesirable events, and cognitive outcomes
for cognitive-psychosocial paired visits completed within 6 months of each other (n = 286).................. 90

viii

ABBREVIATIONS
AD Alzheimer’s disease
ADRD Alzheimer’s disease and related dementias
aMCI Amnestic mild cognitive impairment
aMCI-MD Amnestic mild cognitive impairment multiple domain
ApoE Apolipoprotein E
ARIC Atherosclerosis Risk in Communities
BLSA Baltimore Longitudinal Study of Aging
BMI Body mass index
BNT Boston Naming Test
BRFSS Behavioral Risk Factor Surveillance System
BVAIT B-Vitamin Atherosclerosis Intervention Trial
CCA Common carotid artery
CCA IMT Common carotid artery intima-media thickness
CERAD-NAB Consortium to Establish a Registry for Alzheimer’s Disease-Neuropsychological
Assessment Battery
CES-D Center for Epidemiologic Studies-Depression scale
CIMT Carotid intima-media thickness
CVLT California Verbal Learning Test
DHEA Dehyroepiandosterone
DR Delayed recall
DSMB Data and Safety Monitoring Board
ELITE Early versus Late Intervention Trial with Estradiol
ELSA-Brasil Estudo Longitudinal de Saude do Adulto-Brasil (Brazilian Longitudinal Study of Adult
Health)
ET2DS Edinburgh Type 2 Diabetes Study
HANDLS Healthy Aging in Neighborhoods of Diversity across the Life Span
HDL High-density lipoprotein
HPA Hypothalamic-pituitary-adrenocortical
HVLT Hopkins Verbal Learning Test
ICA Internal carotid artery
ICA IMT Internal carotid artery intima-media thickness
IGF-1 Insulin like growth factor-1
IMT Intima-media thickness
IR Immediate recall
ISEL Interpersonal Support Evaluation List
ISP Isoflavone-rich soy protein
JLO Judgment of Line Orientation, Form H
LDL Low-density lipoprotein
LNS Letter-Number Sequencing
LBC1936 Lothian Birth Cohort 1936
MCI Mild cognitive impairment
MMSE Mini-Mental State Examination
MoCA Montreal Cognitive Assessment
MOS-SSS Medical Outcomes Study-Social Support Survey
MRI Magnetic resonance imaging
MSPSS Multi-Dimensional Scale of Perceived Social Support
ix

naMCI Non-amnestic mild cognitive impairment
NCODE Neurocognitive Outcomes of Depression in the Elderly
NOMAS Northern Manhattan Study
PERI Psychiatric Epidemiological Research Institute
PSS Perceived Stress Scale
SDMT Symbol Digit Modalities Test
SES Socioeconomic status
SHOP Singapore Heart Failure Outcomes and Phenotypes
SBP Systolic blood pressure
tHcy Total homocysteine
Trails B Trail Making Test Part B
WAIS-III Wechsler Adult Intelligence Scale, 3
rd
Edition
WISH Women’s Isoflavone Soy Health
WMS-III Wechsler Memory Scale, 3
rd
Edition
WMS-R Wechsler Memory Scale-Revised
WRAP Wisconsin Registry for Alzheimer’s Prevention

x

ABSTRACT
The projected number of people aged 65 years and older with Alzheimer’s disease is expected to
increase substantially in the coming decades. This can be attributed to the combined effects of longer
life expectancy and the large number of baby boomers reaching old age. Alzheimer’s and associated
dementias result in an enormous socioeconomic burden that is also overwhelming on a personal level
for caregivers and family members. In this context, the discovery of modifiable risk factors for cognitive
impairment and dementia is important.
Previous studies have shown mixed results for associations between subclinical atherosclerosis,
psychosocial factors such as perceived social support, perceived stress, stressful life events and reduced
cognitive function. The associations between these factors and particular cognitive domains are also not
well known. There are several possible reasons for the lack of consistency in previous findings. One is
that various instruments used to evaluate the same cognitive domain differ in their sensitivity to the
assessment of cognitive decline. Furthermore, some studies used only a single instrument while others
employed multi-test batteries to assess cognitive function. Studies employing batteries may be more
likely to detect impairments in particular cognitive domains. In addition, a variety of tools have been
used to assess stress, most being one-dimensional and only measuring an aspect of stress (e.g.,
psychological or physiological). As a result, there is wide variation in how different measures of stress
are associated with cognitive outcomes.
This dissertation aims to examine the associations between subclinical atherosclerosis as
measured by carotid artery intima-media thickness (CIMT) and four cognitive domains: executive
function, verbal memory, visual memory, and global cognition in a population of participants enrolled in
three clinical trials: the BVAIT, WISH, and ELITE studies (n = 1,495, mean age = 61 years). Cognitive
function was assessed by a battery of 14 cognitive and neuropsychological tests that are validated and
sensitive to age-associated change in middle- to older-aged adults. This dissertation will also examine
xi

the associations between perceived social support, perceived stress, and stressful life events with the
same four cognitive domains in a population of women enrolled in the ELITE psychosocial substudy (n =
448, mean age = 61 years). We hypothesize that higher CIMT, high levels of perceived stress and a
greater number of stressful life events experienced are associated with poorer cognitive function. We
also hypothesize that high levels of perceived social support have a protective effect on cognitive
function.

1

CHAPTER 1
INTRODUCTION AND BACKGROUND ON ATHEROSCLEROSIS AND COGNITIVE IMPAIRMENT IN ELDERLY
PERSONS

Dementia as a growing public health concern
The number of people with Alzheimer’s disease (AD) and related dementias (ADRD) is expected
to dramatically rise with an aging world population. In 2010, an estimated 35.6 million people suffered
from Alzheimer’s disease and other dementias worldwide. This number is expected to jump to 66 million
by the year 2030 and 115 million by the year 2050 (Wortmann, 2012). Indeed, prevalence rates of
dementia in persons aged 65 years or older increased drastically from 1990-2000 in European countries
(Lobo et al., 2000). However, data from some recent studies in the United States such as the Health and
Retirement Study have shown that prevalence of dementia among persons aged 65 years or older has
decreased in recent years (Langa et al., 2017). There is evidence that this decline in the United States
can be in part attributed to improvements in general education levels (lower education is associated
with higher dementia incidence) as well as better awareness about controlling cardiovascular risk factors
that contribute to dementia, especially in middle-aged adults (Larson & Langa, 2017). Despite this, in
light of the ongoing obesity epidemic, a rapidly aging US population, and the lack of effective dementia
treatments or prevention strategies, there should be a plan for addressing the anticipated increasing
incidence and prevalence of late-life dementia in decades to come.
Alzheimer’s disease is the 5
th
leading cause of death in the United States. An estimated 6.5
million Americans aged 65 years and older are currently living with Alzheimer’s disease, with that
number projected to more than double to 13.8 million by the year 2060 (Gaugler et al., 2022). Care of
dementia patients is emotionally, physically, and financially exhausting. The burden of providing care
has negative effects on both the mental and physical health of dementia caregivers (Richardson et al.,
2013). Even when dementia patients transition to a nursing home, dementia caregivers continue to
2

provide more care over time compared to non-dementia caregivers (Nikzad-Terhune et al., 2010).
Health care and long-term care for dementia patients is incredibly costly. The average cost per year for
Medicare beneficiaries aged 65 years and older with dementia is estimated to be around $41,757. This is
almost three times the average cost per year of $14,026 for Medicare beneficiaries aged 65 years and
older without dementia (Gaugler et al., 2022). Even with Medicare and other sources of financial
assistance, patients with dementia can still expect to pay a substantial amount out-of-pocket for their
care. Compared to patients without dementia, those with dementia have been estimated to spend
$46,418 (in 2021 dollars) more out-of-pocket between age 65 and death (Gaugler et al., 2022).
Currently, no effective treatments exist for preventing or recovering from dementia. As such,
there has been much research dedicated to studying modifiable risk factors for cognitive impairment.
Prevalence and incidence of mild cognitive impairment
Compared to dementia, mild cognitive impairment (MCI), the condition that potentially
precedes dementia, may not be as well recognized by the general population. Since it was introduced in
the late 1980s, MCI has been depicted as an intermediate stage between the expected cognitive decline
of normal aging and the more serious decline of dementia (Petersen et al., 2014). MCI is a condition in
which an individual may still go about their daily lives normally, but experience changes in cognitive
function that are noticeable to those closest to them (Gaugler et al., 2022). Globally, the prevalence of
MCI has been estimated to be anywhere from 16%-20% among adults 60 years and older and is
expected to increase dramatically in the future with more than a quarter of the world’s population
becoming over 65 years of age (Petsko, 2006; Roberts & Knopman, 2013; Sachdev et al., 2015).
Only a few studies have estimated the incidence of MCI. Incidence rates have been found to
differ by clinical subtype. Individuals with a subtype of MCI that involves memory impairment called
amnestic MCI (aMCI) are more likely to develop Alzheimer’s disease or other dementias than their
cognitively normal peers (Gaugler et al., 2022). While other subtypes of MCI such as non-amnestic MCI
3

(naMCI) have not been shown to be consistently associated with progression to dementia, evidence that
the subtype of MCI can influence the subsequent type of dementia that develops has been controversial
(Michaud et al., 2018). A 2008 study in a multiethnic community of elderly Caribbean Hispanic, black, or
non-Hispanic white adults aged 65 years and older in Northern Manhattan, New York reported an
incidence rate of 7.4% per 100 person-years for individuals with aMCI, and an incidence rate of 4.1% per
100 person-years for individuals with naMCI (Manly et al., 2008). A 2012 study on adults ≥70 years of
age in Minnesota also found higher, albeit less substantial, incidence rates for aMCI compared to naMCI.
This study reported an incidence rate of 3.8% per 100 person-years for individuals with aMCI, and an
incidence rate of 1.5% per 100 person-years for individuals with naMCI (Roberts et al., 2012). This is of
particular interest since aMCI is associated with an increased risk of progression to dementia.
The prognosis for individuals with MCI varies. For some, the symptoms of MCI remain
unchanged for years or may even improve over time while for others, the symptoms progress to
dementia. The estimated proportion of individuals who progress from MCI to dementia differs and is
influenced by the definition of MCI and the type of setting or population under study. In a longitudinal
follow-up study of 111 older individuals diagnosed as having MCI, the annual conversion rate to
dementia was 13% in clinic-based study settings, and 3% in community-based study settings (Farias et
al., 2009). A meta-analysis of 41 cohort studies found that the annual conversion rate of individuals with
MCI to dementia and Alzheimer’s disease was 9.6% and 8.1%, respectively, in clinic settings and 4.9%
and 6.8%, respectively, in community studies (Mitchell & Shiri-Feshki, 2009). Research has shown that
greater carotid intima-media thickness (CIMT), a marker of subclinical atherosclerosis, is associated with
increased risk of conversion from MCI to dementia (Viticchi et al., 2012). A better understanding of the
preclinical phase, particularly modifiable risk factors, is important in addressing the public health
concern of dementia.
Risk factors for mild cognitive impairment
4

MCI is common in old age. An estimated 15-42% of people aged 65 years or older have MCI
(Petersen et al., 2014). Aside from age, cardiovascular disease risk factors such as hypertension,
dyslipidemia, and diabetes have been associated with reduced cognitive function (Kumari & Marmot,
2005; Solomon et al., 2009; Tervo et al., 2004). Past studies on elderly patients with severe carotid
artery stenosis have shown a clear association between carotid atherosclerosis and cognitive
impairment (Bo et al., 2005; Bo et al., 2006). Greater amounts of carotid plaque have been linked to
elevated risk of developing dementia (Wendell et al., 2012). These findings have been corroborated by
pathological studies of carotid arteries (Hulette 1997, Suemoto 2011). Even mild carotid atherosclerosis
considered clinically asymptomatic, also known as subclinical atherosclerosis, has been shown in cross-
sectional studies to be associated with poorer cognitive performance in middle-aged adults after
adjusting for other vascular risk factors (i.e., diabetes, hypertension, and smoking) (Romero et al., 2009).
Lower education level (≤12 years) has also been found to be associated with an increased risk of incident
MCI (Roberts et al., 2012). Other pathological studies of the brain have found a strong association
between intracranial atherosclerosis, or atherosclerosis in the brain, and Alzheimer’s pathology (Honig
et al., 2005; Roher et al., 2003).
Since MCI has been considered a precursor of dementia, research has focused on known genetic
markers for AD among patients with MCI. One of the greatest risk factors for AD in the general
population is inheriting copies of the apolipoprotein E (ApoE) e4 allele (Gaugler et al., 2022). In general,
the risk of AD increases with the number of copies inherited. Individuals who inherit one copy of the e4
allele have three times the risk of developing AD compared to those with the more common e3 allele,
while those who inherit two copies have eight to twelve times the risk (Holtzman et al., 2012; Loy et al.,
2014; Michaelson, 2014). Having the ApoE e4 allele can also increase the risk of dementia with Lewy
bodies (MedlinePlus, 2021). The ApoE gene codes for the apolipoprotein E protein which is responsible
for transporting cholesterol in the bloodstream (Gaugler et al., 2022). The ApoE e4 allele has been found
5

to be associated with an increased number of amyloid plaques between neurons which are a hallmark of
AD brain pathology (MedlinePlus, 2021). It is unsurprising then that the presence of at least one ApoE e4
allele also increases risk for MCI (DeCarli et al., 2001; Tervo et al., 2004).
Subclinical atherosclerosis and mild cognitive impairment
CIMT, a measurement of the thickness of the two inner layers of the carotid artery easily
assessed by ultrasound, is often used to identify and quantify atherosclerosis in the subclinical stages
(Stein et al., 2008). While this study will not be looking at transition between cognitive states as an
outcome, it is nonetheless worthwhile to examine literature where subclinical atherosclerosis has
contributed to clinically-defined cognitive states such as MCI. In two Korean studies of older adults,
greater CIMT was found to be associated with both incident and prevalent MCI (Moon et al., 2015; Park
et al., 2019). More specifically, in the 2015 study, participants with cognitive impairment who
progressed to MCI or dementia had greater CIMT compared to participants who were cognitively normal
or whose cognitive status was unchanged during the 5-year study period. In the 2019 cross-sectional
study, greater CIMT was found to be positively associated with prevalent MCI. Although both of these
studies considered performance on batteries of neuropsychological tests as part of the definition of
MCI, the criteria used to diagnose MCI differed. The 2015 study used diagnostic criteria proposed by the
International Working Group on MCI to define the outcome whereas the 2019 study referred to a
dementia specialist to make the diagnosis of MCI.
The association between CIMT and MCI has also been reported to differ by MCI subtype. A study
of Italian adults aged 45 years and older reported a positive association between CIMT and incident
amnestic MCI multiple domain (aMCI-MD), a subtype of MCI in which memory and one or more
additional non-memory domains are impaired (Camarda et al., 2018). This study did not find an
association between CIMT and aMCI, the subtype of MCI most likely to progress to dementia. This
contrasts with the results from an earlier study of Spanish adults aged 65 years and older without
6

history of clinical vascular disease in which a strong association between greater CIMT and prevalent
aMCI was found (Casado Naranjo et al., 2015). It should be noted that the definition of aMCI in the 2015
study differed slightly from the 2018 study. The 2015 study did not find a statistically significant
difference with regard to CIMT measures between participants with aMCI or aMCI-MD. As such, the two
groups were combined into a single group denoted as aMCI during analyses.
Subclinical atherosclerosis and cognitive function
The majority of studies examining the association between subclinical atherosclerosis and
cognitive function have not used diagnostic criteria for MCI as the outcome, but rather performance on
specific neuropsychological tests among apparently cognitively unimpaired persons. As such, there is
comparatively more information on the association between subclinical atherosclerosis and cognitive
function. Subclinical atherosclerosis as measured by CIMT has been found to have a significant positive
association with brain magnetic resonance imaging (MRI) markers for ischemia in middle-aged adults
(Romero et al., 2009). It is not surprising then that greater CIMT has also been shown to be associated
with reduced cognitive functions and increased risk of cognitive impairment, particularly in older adults
(Johnston et al., 2004; Moon et al., 2015; Wendell et al., 2009).
While many studies have shown an association between greater CIMT and poorer cognitive
performance, the findings vary on associations between greater CIMT and particular cognitive domains.
Cognitive domains studied have included executive function, verbal memory, visual memory, and global
cognition. Executive function refers to an individual’s ability to plan and coordinate multiple tasks
(Diamond, 2013; Sachdev et al., 2014). Verbal memory encompasses immediate and delayed recall and
recognition of verbal material while visual memory refers to the same of visual material (Sachdev et al.,
2014).

7

CIMT and executive function
A few studies have reported an inverse association between greater CIMT and executive
function. A 2009 study on 804 adults aged 20-93 who were participants in the Baltimore Longitudinal
Study of Aging (BLSA) found that greater CIMT was associated with poorer performance on the Category
Fluency test and Trail Making Test Parts A and B, which are measures of executive function, after
adjusting for age (Wendell et al., 2009). A 2013 study on 831 diabetic adults aged 60-75 years who were
participants in the Edinburgh Type 2 Diabetes Study (ET2DS) also found an inverse association between
greater CIMT and executive function (Feinkohl et al., 2013).
There are also a few studies that have found no association between CIMT and executive
function. A 2009 study on 504 healthy hyperhomocysteinemic adults aged 40 years and older who were
participants in the B-Vitamin Atherosclerosis Intervention Trial (BVAIT) found that CIMT had no
association with executive function (Gatto et al., 2009). A second 2009 study on 1,975 Framingham
Offspring Study participants with a mean age of 58 years also did not find an association between
common carotid artery intima-media thickness (IMT) and executive function (Romero et al., 2009).
These findings were echoed in later studies. A 2015 study on 8,208 adults with a mean age of 49.6 years
in the Brazilian Longitudinal Study of Adult Health (ELSA-Brasil) did not find any significant association
with executive function. However, this same study found that there was a stronger inverse association
between CIMT and executive function among participants who reported current alcohol use (Suemoto
et al., 2015). Two 2016 studies, one on 308 adults with a mean age of 63 years without a history of heart
attack, heart failure, or stroke who served as controls in the Singapore Heart Failure Outcomes and
Phenotypes (SHOP) study, and another on 1,712 adults aged 30-64 years in the Healthy Aging in
Neighborhoods of Diversity across the Life Span (HANDLS) study, also found that CIMT had no
association with executive function (Lim et al., 2016; Wendell et al., 2016). However, an inverse
association between CIMT and executive function was found in the HANDLS study once a three-way
8

interaction between IMT, race, and poverty was accounted for in the final mixed-effects model (Wendell
et al., 2016). More specifically, this inverse association was most evident among white participants and
those with higher socioeconomic status (SES). A 2017 study on 1,166 adults with a mean age of 70 years
in the Northern Manhattan Study (NOMAS) also did not find any association between CIMT and
executive function, even after stratifying analysis by ApoE4+/ApoE4- status (Gardener et al., 2017).
CIMT and verbal memory
In the 2009 study on BVAIT participants, greater CIMT was found to have a weak inverse
association with performance on the California Verbal Learning Test (CVLT), a measure of verbal
memory (Gatto et al., 2009). However, this same study did not find an association between greater
CIMT and other components of verbal memory such as logical memory or semantic memory (Gatto et
al., 2009). This may be attributed to variability in tests used to measure verbal memory. The tests used
to measure logical memory and semantic memory are less sensitive to episodic memory decline
compared to the CVLT (Rabin et al., 2009). Inverse associations between greater CIMT and verbal
memory were also found in the 2009 BLSA study, the 2013 ET2DS study, and the 2015 ELSA-Brasil study
(Feinkohl et al., 2013; Suemoto et al., 2015; Wendell et al., 2009). The 2016 HANDLS study found an
association between greater CIMT and poorer performance on the delayed recall portion of the CLVT
once a three-way interaction between IMT, race, and poverty was accounted for in the final mixed-
effects model. The association was most apparent in white participants with high SES (Wendell et al.,
2016). The 2017 NOMAS study found an inverse association between greater CIMT and two components
of verbal memory (Gardener et al., 2017). More specifically, an inverse association between greater
CIMT and episodic memory was found prior to and after stratifying analysis by ApoE4+/ApoE4- status.
The inverse association between greater CIMT and semantic memory was found only for participants
who were ApoE4+. A 2018 study on 206 patients with a mean age of 65.1 years who visited the Tokyo
University Hospital for health screenings from 2009 to 2013 found that greater maximum IMT was
9

inversely associated with the immediate recall component but not the delayed recall component of
verbal memory (Matsumoto et al., 2018).
The 2009 Framingham Offspring Study did not find that common carotid artery IMT was
associated with poorer verbal memory performance (Romero et al., 2009). A 2012 study on 4,371 adults
aged 25 years and older in Norway also did not find any association between mean common carotid
artery far-wall IMT and verbal memory (Arntzen et al., 2012). Similarly, neither the 2016 SHOP study nor
the 2016 HANDLS study found any association between CIMT and verbal memory (Lim et al., 2016;
Wendell et al., 2016). As previously mentioned, while the 2017 NOMAS study did find inverse
associations between CIMT and components of verbal memory, no association was found between CIMT
and semantic memory prior to stratifying analysis by ApoE4+/ApoE4- status (Gardener et al., 2017).
After stratifying analysis by ApoE4+/ApoE4- status, an inverse association was found between greater
CIMT and semantic memory among ApoE4+ participants.
CIMT and visual memory
Studies reporting on the association between CIMT and visual memory are almost evenly split.
The 2009 BLSA study and the 2013 ET2DS study found inverse associations between CIMT and visual
memory (Feinkohl et al., 2013; Wendell et al., 2009) whereas the 2009 BVAIT study, the 2009
Framingham Offspring study, and the 2016 SHOP study found no associations between CIMT and visual
memory (Gatto et al., 2009; Lim et al., 2016; Romero et al., 2009).
CIMT and global cognition
There has not been as much focus on global cognition as an outcome in past studies. Global
cognition has previously been assessed in a variety of ways. The 2013 ET2DS study found an inverse
association between CIMT and global cognition as measured by the Mini-Mental State Examination
(MMSE) (Feinkohl et al., 2013) whereas the 2009 BVAIT study and the 2018 study on Tokyo University
Hospital patients found no associations between CIMT and global cognition as measured by a weighted
10

composite of scores in the neuropsychological battery in the BVAIT study and the MMSE and the
Hasegawa Dementia Scale-revised in the Tokyo University Hospital study, respectively (Gatto et al.,
2009; Matsumoto et al., 2018).
There are several potential reasons for the lack of consistency in the findings. One reason is that
it is not known how well the commonly used tests for cognition agree with each other. For example,
there are several different tests that can be used to assess verbal learning such as the CVLT, the Hopkins
Verbal Learning Test (HVLT), and the Word List task from the Consortium to Establish a Registry for
Alzheimer’s Disease-Neuropsychological Assessment Battery (CERAD-NAB). Not only do these tests
differ in difficulty, but the CVLT and the CERAD-NAB have also been found to differ in sensitivity in
detecting episodic memory impairments (Beck et al., 2012). Some studies only use a single tool to assess
cognitive impairment while others use an entire battery. It is likely that studies using extensive batteries
may be more likely to detect impairments in particular cognitive domains.
The association between subclinical atherosclerosis and cognitive decline may also differ
depending on the operational definition of subclinical atherosclerosis. While some studies use average
CIMT to define subclinical atherosclerosis (which is averaged over multiple measures of CIMT over a
length or specific sites of the carotid artery), others use maximum CIMT (which is a measure of CIMT at a
single site). The association also differs depending on the site of IMT measurement. For example, at the
left versus right common carotid artery (CCA), or at the common carotid artery versus the internal
carotid artery (ICA) (Johnston et al., 2004; Romero et al., 2009). Previous studies have shown that
internal carotid artery IMT (ICA IMT) is a better marker for cognitive impairment than common carotid
artery IMT (CCA IMT) (Romero et al., 2009). However, ICA IMT is difficult to reliably measure. While the
ICA is more prone to atherosclerotic changes, the branching angle of the ICA from the CCA makes it
much more challenging obtain accurate measurements (Sillesen, 2014). The number and size of plaques
defined as protrusion into the vessel lumen with thickening of the vessel wall in the CCA and ICA may
11

also be better markers for cognitive impairment. A 2012 study found that subclinical atherosclerosis
measured in this way, but not CCA far-wall IMT, were predictors for cognitive decline (Arntzen et al.,
2012). In comparison to early intima-media thickening, plaques are a sign of more advanced disease and
are not always present in the CCA of individuals with subclinical atherosclerosis
To further elucidate the association between subclinical atherosclerosis and cognitive decline,
this current study combines data from three clinical trials: the B-Vitamin Atherosclerosis Intervention
Trial (BVAIT), the Women’s Isoflavone Soy Health study (WISH), and the Early versus Late Intervention
Trial with Estradiol (ELITE) for a larger sample size. Participants in each of these clinical trials were given
a thorough battery of standardized neuropsychological tests at baseline and at follow-up visits, which
allows for examinations of longitudinal associations. The scores from the various neuropsychological
tests are combined into composite scores for executive function, verbal memory, visual memory, and
global (overall) cognition. These clinical trials also have a well-characterized measure of subclinical
atherosclerosis, measured as average CIMT, through high resolution B-mode ultrasound of the right
CCA. Additionally, we controlled for potential confounding effects of cardiovascular disease risk factors
and examined effect modification by ApoE4 genotype. Thus, this study analyzed cognitive decline
outcomes that were assessed through a battery of neuropsychological tests sensitive to age-related
changes in middle- to older-aged adults and correlated these cognitive outcomes with a well-validated
measure of subclinical atherosclerosis with long-term follow-up, information on various biochemical and
behavioral factors, and a larger sample size than previous studies.

12

CHAPTER 2
SUBCLINICAL CAROTID ATHEROSCLEROSIS AND COGNITIVE FUNCTION IN MIDDLE- TO OLDER-AGED
ADULTS

INTRODUCTION
Alzheimer’s disease, one of the leading causes of death in the United States, affects an
estimated 6.5 million Americans (Gaugler et al., 2022). This figure is projected to increase to 13.8 million
by the year 2060 (Gaugler et al., 2022). Because there are no current effective treatments for preventing
or treating dementia, there has been significant interest in identifying modifiable risk factors for
cognitive impairment. Past studies have shown a clear association between carotid atherosclerosis and
cognitive impairment (Bo et al., 2005; Bo et al., 2006). Indeed, greater amounts of carotid arterial plaque
have been linked to an increased risk of developing dementia (Wendell et al., 2012). Even mild carotid
atherosclerosis considered clinically asymptomatic, also known as subclinical atherosclerosis, has been
shown in cross-sectional studies to be associated with poorer cognitive function in middle-aged adults
after adjusting for other vascular risk factors (i.e., diabetes, hypertension, and smoking) (Romero et al.,
2009).
Carotid artery intima-media thickness (CIMT), a measurement of the thickness of the two inner
layers of the carotid artery easily assessed by ultrasound, is often used to identify atherosclerosis in the
subclinical stages (Stein et al., 2008). Subclinical atherosclerosis as measured by CIMT has been shown
to be associated with increased risk of cognitive impairment, particularly in older adults (Johnston et al.,
2004; Moon et al., 2015; Wendell et al., 2009).
While many studies have shown an association between greater CIMT and poorer cognitive
performance, the findings vary on associations between greater CIMT and particular cognitive domains.
For example, while some studies did not find any association between greater CIMT and executive
function (Gardener et al., 2017; Gatto et al., 2009; Lim et al., 2016), others found an inverse association
13

between greater CIMT and executive function (Feinkohl et al., 2013; Suemoto et al., 2015; Wendell et
al., 2009). Disparate results have also been found for associations between CIMT and other cognitive
domains.
There are several potential reasons for the lack of consistency in the findings. One is that it is
not known how well the commonly used tests for cognition are in agreement with each other. Another
potential reason is that some studies only use a single tool to assess cognitive impairment while others
use a multi-test battery. Differences in study populations may also contribute to the lack of consistency
in the findings. Finally, the association between subclinical atherosclerosis and mild cognitive
impairment may differ depending on the operational definition of subclinical atherosclerosis.
To evaluate the association between subclinical atherosclerosis and cognitive function in
middle- to older-aged adults, this current study had a larger study population using combined data from
three clinical trials: the B-Vitamin Atherosclerosis Intervention Trial (BVAIT), the Women’s Isoflavone
Soy Health study (WISH) trial, and the Early versus Late Intervention Trial with Estradiol (ELITE).
Participants in each of these clinical trials completed a thorough battery of standardized
neuropsychological tests at baseline and at follow-up visits, allowing for examination of longitudinal
associations. The scores from the various neuropsychological tests were combined into composite
scores for executive function, verbal memory, visual memory, and global (overall) cognition. These
clinical trials also have a well-characterized measure of atherosclerosis, measured as average CIMT,
through high resolution B-mode ultrasound of the right common carotid artery (CCA); ultrasound
imaging and CIMT measurement were completed using the same standardized protocol in all three
trials. Additionally, we controlled for potential confounding effects of cardiovascular disease and
demographic risk factors.

14

MATERIALS AND METHODS
Study design
This study combined data from three clinical trials: the B-Vitamin Atherosclerosis Intervention
Trial (BVAIT) (NCT00114400), the Women’s Isoflavone Soy Health (WISH) trial (NCT00118846), and the
Early versus Late Intervention Trial with Estradiol (ELITE) (NCT00114517).
BVAIT design
BVAIT was a double-blind, placebo-controlled, randomized clinical trial conducted from
November 2000 to June 2006. The trial was designed to examine whether B-vitamin supplementation
would reduce the progression of early atherosclerosis in individuals over 40 years of age with higher
levels of total homocysteine (tHcy). Eligible men and women had fasting tHcy≥8.5 μmol/L and did not
have diabetes or clinical signs/symptoms of cardiovascular disease. A total of 506 participants were
randomized to daily high-dose vitamin B supplementation (folic acid 5 mg + vitamin B 12 0.4 mg + vitamin
B 6 50 mg) or matching placebo in a 1:1 ratio within two strata of baseline carotid artery intima-media
thickness (CIMT) (<0.75 mm, ≥0.75 mm). Cognitive function was measured using a battery of 14
neuropsychological tests at baseline and again at 2.5 years. As the trial was in progress, the initial
treatment period of 2.5 years was extended 1-2 years as recommended by the study’s Data and Safety
Monitoring Board (DSMB).
In the primary outcome analysis, high-dose B-vitamin supplementation was found to
significantly reduce progression of subclinical atherosclerosis in a subgroup of healthy participants with
fasting tHcy≥9.1 μmol/L without clinical evidence of cardiovascular disease (Hodis et al., 2009). A post-
trial cross-sectional analysis of baseline measures showed that higher CIMT was weakly associated with
lower verbal learning abilities (Gatto et al., 2009). CIMT was not found to have any association with
global cognition in the same analysis.

15

WISH design
WISH was a double-blind, placebo-controlled, randomized trial conducted from April 2004 to
March 2009. WISH was designed to determine whether isoflavone-rich soy protein (ISP) supplements
had an effect on the progression of subclinical atherosclerosis in healthy post-menopausal women.
Similar to BVAIT, eligible participants did not have diabetes or clinical signs/symptoms of cardiovascular
disease. Menopause in WISH was defined as absence of menses >1 year and serum estradiol level lower
than 20 pg/ml. A total of 350 women were randomized to treatment (25 g soy protein containing 91 mg
aglycone equivalents of naturally occurring isoflavones and respective glycosides [154 mg total
isoflavone conjugates plus aglycone equivalents] of genistein, daidzein, and glycitein) or matching
placebo in a 1:1 ratio within two strata of baseline CIMT (<0.75 mm, ≥0.75 mm). Cognitive function was
measured using a battery of 14 neuropsychological tests at baseline and again at 2.5 years. As with
BVAIT, the initial treatment period of 2.5 years was increased by the study’s DSMB, in this case, to 3
years with an optional additional 6 month study visit.
In the primary outcome analysis, ISP supplementation was not associated with less progression
of subclinical atherosclerosis compared to placebo among healthy post-menopausal women (Hodis et
al., 2011). Examination of global cognitive composite score between the two treatment groups after 2.5
years showed that long-term dietary soy isoflavone supplementation had no effect on global cognition.
However, improvements were observed in the visual memory composite score after 2.5 years among
participants randomized to treatment but not among participants randomized to placebo. There were
no significant differences over time in composite scores between the two groups for executive function
or verbal memory (Henderson et al., 2012). Follow-up analysis utilizing data on change in urine excretion
of isoflavonoids in relation to change in cognitive measures supported the previous finding that long-
term dietary soy isoflavone supplementation had no effect on global cognition. Additionally, changes in
16

urine excretion of isoflavonoids was found to be negatively associated with changes in general
intelligence but not changes in memory (St John et al., 2014).
ELITE design
ELITE was a double-blind, placebo-controlled, randomized clinical trial conducted from July 2005
to February 2013. It was designed to examine the effects of oral 17β-estradiol (estrogen) on the
progression of subclinical atherosclerosis and cognitive decline in healthy post-menopausal women.
Similar to BVAIT and WISH, eligible participants did not have diabetes or clinical signs/symptoms of
cardiovascular disease. Menopause was defined as absence of menses for at least 6 months or bilateral
oophorectomy, as well as a serum estradiol level lower than 25 pg/ml (92 pmol per liter). ELITE was
specifically designed to test the hypothesis that estrogen supplementation would reduce atherosclerosis
progression when given in early menopause (<6 years after menopause), but not in late menopause
(≥10 years after menopause). Secondary cognitive outcomes evaluated a similar hypothesis. A total of
643 women were randomized to treatment (oral 17B-estradiol 1 mg/day plus micronized progesterone
[45 mg] as a 4% vaginal gel for women with a uterus) or placebo in a 1:1 allocation ratio, within strata
defined by length of time since menopause (early menopause being <6 years past menopause and late
menopause being ≥ 10 years past menopause). Other stratification factors included presence or absence
of a uterus at randomization. Women with a uterus also received progesterone as a 4% vaginal gel or
matched placebo. Cognitive function was measured using a battery of 14 neuropsychological tests
administered at baseline, 2.5 years, and 5 years after randomization.
In the primary outcome analysis, oral estradiol therapy was associated with less progression of
subclinical atherosclerosis (as measured by CIMT) compared to placebo in the early menopause group;
estrogen and placebo groups did not differ on atherosclerosis progression in the late menopause group
(Hodis et al., 2016). Oral estradiol therapy was also not found to affect verbal memory, executive
functions, or global cognition in either the early or late menopause groups (Henderson et al., 2016).
17

Carotid artery intima-media thickness assessment
In each of the three clinical trials, high-resolution B-mode ultrasound images of the right distal
common carotid artery were obtained as previously described at baseline and every 6 months during
follow-up (Hodis et al., 2001). From the images, CIMT was measured using software that was developed
at USC for longitudinal measurements of atherosclerotic changes (Selzer et al., 1994; Selzer et al., 2001).
More specifically, CIMT was the average of multiple individual measurements between the intima-
lumen and media-adventitia interfaces along a 1-cm length of the far wall of the right common carotid
artery (Gatto et al., 2009). This method ensured that the same measurements were being taken in all
study participants.
Cognitive function assessment
The same battery of cognitive and neuropsychological assessments were administered by a
single trained psychometrist in a standardized order to participants in each of the three clinical trials.
The battery included the following 14 assessments:
 Symbol Digit Modalities Test (SDMT)
 Trail Making Test Part B (Trails B)
 Judgment of Line Orientation, Form H (JLO)
 Block Design [Wechsler Adult Intelligence Scale, 3
rd
Edition (WAIS-III)]
 Letter-Number Sequencing [Wechsler Memory Scale, 3
rd
Edition (WMS-III)] (LNS)
 Category fluency (animal naming, 60 seconds) (Animals)
 Boston Naming Test, 30-item version (BNT)
 Shipley Institute of Living Scale (Shipley), Abstraction Scale
 California Verbal Learning Test, 2
nd
edition (CVLT-II), immediate recall (IR) and delayed recall
(DR)
 Logical Memory, immediate recall and delayed recall
18

 Faces I (IR) and II (DR) (WMS-III)
The assessments in this battery were selected to be sensitive to age-related changes in cognition
and were chosen to evaluate different cognitive functions and abilities with a focus on executive
function, verbal memory, and visual memory. Participants in all three clinical trials were given the
battery at baseline and at 2.5 years. Participants in ELITE were also given the battery at 5 years.
Results from the 14 assessments were used to generate composite scores for executive function,
verbal memory, visual memory, and global cognition. The executive function composite score was
calculated as a weighted average of the following cognitive tests: Symbol Digit Modalities Test, Trail
Making Test Part B, Shipley Institute of Living Abstraction Scale, Letter-Number Sequencing, and
Category fluency. The verbal memory composite score was calculated as a weighted average of the
CVLT-II (immediate and delayed recall) and Logical Memory (immediate and delayed recall) test. The
visual memory composite score was calculated as a weighted average of the Faces I (immediate recall)
and Faces II (delayed recall) tests. The global cognition composite score was calculated as a weighted
average of all 14 assessments. For each participant, all test scores were converted to a standardized Z
score at baseline, 2.5 years, and 5 years. The Z scores were calculated using the baseline mean and
standard deviation for each test, and were specific to each trial. To calculate the composite scores, the Z
scores for the appropriate cognitive tests were summed and then weighted by the inverse inter-test
correlation matrix (Gatto et al., 2009; Henderson et al., 2016; Henderson et al., 2012; Henderson et al.,
2013).
Statistical methods
Because some cognitive or CIMT data were not captured exactly at the 2.5- and 5-year visits
(due primarily to missed study visits), follow-up CIMT and cognitive measures were defined over a range
of visits for the two time points; the 2.5-year variable encompassed visit months 24-36, and the 5-year
19

variable encompassed visit months 52-66. Table 2.1 summarizes the number of visits completed at each
month by study.
Table 2.1 Visits by study
Visit (month)
0 24 28 30 32 33 34 36 52 54 56 58 60 62 64 66
BVAIT 504 1 0 353 0 52 0 34 N/A N/A N/A N/A N/A N/A N/A N/A
WISH 349 1 1 201 42 0 16 12 N/A N/A N/A N/A N/A N/A N/A N/A
ELITE 642 0 0 345 105 0 38 17 40 26 171 78 32 2 4 6
Cognitive data were scheduled to be completed at visit months 0 and 30.
Cross-sectional associations between CIMT and cognition at baseline and 2.5 years. Pairwise associations
between measures of cognitive function and participant variables were first examined by fitting simple
linear regression models with executive function, verbal memory, visual memory, and global cognition
composite scores as the dependent variables. Participant variables included the primary dependent
variable of interest, CIMT, as well as demographics (age, sex, race, education level, income level, and
marital status), laboratory values (low-density lipoprotein (LDL)/high-density lipoprotein (HDL)
cholesterol, glucose), lifestyle factors (smoking status, alcohol consumption, and physical activity
summary measures), use of anti-hypertensive and cholesterol lowering medications at baseline, and
vitals (body mass index (BMI) and blood pressure). Multivariable linear regression models were then
used to examine the cross-sectional association at baseline and at 2.5 years between CIMT and cognitive
function domains. Covariates that displayed a significant association with the cognitive variables were
evaluated as potential confounders of the association between cognitive measures and CIMT. These
covariates were entered individually into linear regression models of the association between CIMT
analyzed as a continuous variable and each of the dependent cognitive variables. Addition of covariates
that resulted in a >10% change in the CIMT regression coefficient were considered confounders and
included for adjustment in the final model. In models where height or weight were found to be a
20

confounder, they were replaced with BMI as it incorporates both height and weight and is a better
indicator of body composition (Gallagher et al., 1996). Due to associations of vascular factors with
atherosclerosis and cognition (Anstey et al., 2008; Baumgart et al., 2015; Toth, 2008), vascular risk
factors (BMI, systolic blood pressure (SBP), HDL, LDL and current smoking status) were included as
relevant confounders in all models, regardless of whether or not they met the change-in-estimate
criterion. Indicator variables for study were also included in all models to account for differences
between clinical trials. To assess potential trial treatment effects, a variable for trial treatment
assignment (placebo, active B-vitamin [BVAIT], active soy [WISH], or active hormone treatment [ELITE])
was used. For the multiple linear regression model with verbal memory weighted composite score as
the dependent variable, the final model only included covariates that were historically known to be
important confounders and those with p-values <0.20.
In addition to assessment of CIMT as a continuous independent variable, CIMT was categorized
into quartiles to allow for examination of trends and possible non-linearities (e.g., thresholds) in
associations. A few covariates were modified to improve interpretability and simplify the analysis. Age
was centered on the mean, while categories for race (Native American and Other, respectively) were
combined due to small sample size in these categories. Both education and income levels, that were
captured as categories with given ranges, were simplified into one-degree of freedom variables.
Education categories were replaced with a standard number of years of education (e.g., high school
graduate, 12 years). The median level of a reported income category was assigned as a given
participant’s reported annual income. A dichotomous alcohol variable (yes/no) was created to reflect
overall consumption of alcohol by combining the separate responses to consumption of different types
of alcohol (beer, red wine, white wine, hard liquor). Reflecting the elevated risk for Alzheimer’s disease
associated with the ApoE4 allele, ApoE genotype was modeled as a dichotomous variable; ApoE4+ if a
participant had at least one copy of the ApoE4 allele, and ApoE4- if not. To test if CIMT associations with
21

cognitive function differed by the presence of an ApoE4+ genotype, interactions between ApoE4
genotype and CIMT as well as categorical CIMT were tested for each of the four fully adjusted models.
Product terms were also used to test for interactions between CIMT and sex and between CIMT and age
group (participants <65 years old and participants ≥65 years old). Indicator variables for study were also
included in all models to account for difference between clinical trials. To assess for potential trial
treatment effects, a variable for trial treatment assignment (placebo, active B-vitamin [BVAIT], active
isoflavone-rich soy protein [WISH], or active hormone treatment [ELITE]) was used.
Because BMI, SBP, and HDL are potential determinants of CIMT in the cognitive function
pathway, the cross-sectional association models were also examined without adjustment for these
variables.
Associations of CIMT with cognitive change over 2.5 years. A change variable was created for each
dependent cognitive variable as the difference between the 2.5-year and baseline scores. Associations
between baseline CIMT and the cognitive change variables were tested in models adjusting only for
baseline cognitive scores and in multivariable linear regression models. Potential confounders were re-
evaluated in the final models. Covariates that resulted in a >10% change in the CIMT regression
coefficient were considered confounders and included for adjustment in the final models. Modeling
procedures were completed as described above with the change in cognitive measure as the dependent
variable. To examine the CIMT-cognition association by age, analysis of associations between baseline
CIMT and the cognitive change variables were also stratified by age group (<65 and ≥65 years old). The
associations between change in CIMT and the cognitive change variables were also assessed.
RESULTS
Study sample
Baseline characteristics for study participants are summarized in Table 2.2 by study. Since ELITE
and WISH enrolled only women, the majority of study participants in the total analysis sample were
22

women (79.4%). The majority of study participants were also non-Hispanic White (66.0%) and married
(59.4%). Study participants had an average (SD) age of 61.1 (8.1) years, were highly educated (mean =
15.9 (2.0) years of education correlating with a Bachelor’s degree), had an average annual income of
$66,700 ($30,700), and were overweight (mean BMI = 27.4 (5.2) kg/m
2
). Study participants in BVAIT (the
only trial that included men) were slightly older (61.5 (9.9) years) and slightly more overweight (mean
BMI = 28.1 (4.9) kg/m
2
) while those in ELITE were slightly more educated (mean = 16.0 (1.9) years of
education) and had the highest average annual income ($69,000 ($31,000)). Very few participants
reported as current smokers (3.2%), and about half reported consuming alcohol (45.9%). Over one third
of the sample (527, 35.3%) reported current use of anti-hypertensive medication and 279 (18.7%)
participants reported current use of cholesterol lowering medication. Across all three clinical trials, the
average systolic blood pressure was 120.4 (15.2) mmHg, and the average diastolic blood pressure was
76.0 (8.8) mmHg. A little over a quarter of participants (26.9%) were positive for the ApoE4 genotype.
The average Center for Epidemiologic Studies-Depression (CES-D) scale score at baseline was low (7.4
out of a possible 60) and only a small proportion (13.2%) scored 16 or above (the cut off score indicating
concern for clinical depression) (American Psychological Association, 2011). The average CIMT did not
substantially differ across all three clinical trials (0.77 (0.12) mm).

23

Table 2.2 Baseline characteristics for study participants by clinical trial (n = 1,495)
Variable BVAIT
(n = 504)
a

WISH
(n = 349)
a

ELITE
(n = 642)
a

Combined
Mean ± SD or Number (%)
Age (years) 61.5 ± 9.9 61.4 ± 7.1 60.6 ± 6.9 61.1 ± 8.1
Sex
Male 308 (61.1%) 0 (0%) 0 (0%) 308 (20.6%)
Female 196 (38.9%) 349 (100%) 642 (100%) 1,187 (79.4%)
Race
Non-Hispanic White 326 (64.7%) 222 (63.6%) 439 (68.4%) 987 (66.0%)
Non-Hispanic Black 75 (14.9%) 21 (6.0%) 60 (9.3%) 156 (10.4%)
Hispanic 55 (10.9%) 55 (15.8%) 90 (14.0%) 200 (13.4%)
Asian or Pacific
Islander
45 (8.9%) 38 (10.9%) 53 (8.3%) 136 (9.1%)
Other 3 (0.6%) 13 (3.7%) 0 (0%) 16 (1.1%)
Education (# of
years)
15.7 ± 2.0
b
15.8 ± 2.0 16.0 ± 1.9 15.9 ± 2.0
Annual income $64.6k ± $29.7k
b
$65.3k ± $31.3k
b
$69.0k ± $31.0k
b
$66.7k ± $30.7k
Marital status
Single, never
married
45 (8.9%) 33 (9.5%) 50 (7.8%) 128 (8.6%)
Married 323 (64.1%) 197 (56.4%) 368 (57.3%) 888 (59.4%)
Separated 5 (1.0%) 7 (2.0%) 12 (1.9%) 24 (1.6%)
Divorced 89 (17.7%) 80 (22.9%) 171 (26.6%) 340 (22.8%)
Widowed 41 (8.1%)
2
32 (9.2%) 41 (6.4%) 114 (7.6%)
Current smoker 17 (3.4%)
2
8 (2.3%) 22 (3.4%) 47 (3.2%)
Alcohol use 221 (43.8%) 152 (43.6%) 313 (48.8%) 686 (45.9%)
Use of anti-
hypertensives at
baseline
190 (37.7%) 117 (33.5%) 220 (34.3%) 527 (35.3%)
Use of cholesterol
lowering medication
at baseline
81 (16.1%) 72 (20.6%) 126 (19.6%) 279 (18.7%)
BMI (kg/m
2
) 28.1 ± 4.9 26.6 ± 5.2 27.2 ± 5.4
b
27.4 ± 5.2
Blood pressure
(mmHg)

Systolic 127.0 ±15.6 117.9 ± 13.9 116.5 ± 13.7
b
120.4 ± 15.2
Diastolic 78.7 ± 8.9 75.0 ± 8.6 74.5 ± 8.4
b
76.0 ± 8.8
ApoE4+ 103 (22.2%)
b
86 (25.7%)
b
197 (30.9%)
b
386 (26.9%)
CES-D score
c
6.2 ± 6.6 7.3 ± 6.8
b
8.3 ± 8.6 7.4 ± 7.6
CES-D score ≥ 16
d
44 (8.7%) 40 (11.5%) 113 (17.6%) 197 (13.2%)
Average CIMT (mm,
over 1 cm segment)
0.75 ± 0.15 0.81 ± 0.10 0.77± 0.11 0.77± 0.12
Executive function
composite score
-0.012 ± 1.369 -0.005 ± 1.323 -0.002 ± 1.360 -0.006 ± 1.353
Verbal memory
composite score
-0.005 ± 1.317 -0.006 ± 1.357 0.001 ± 1.348 -0.003 ± 1.339
24

Visual memory
composite score
0.001 ± 1.108 0.001 ±1.107 0 ± 1.103 0.001 ± 1.105
Global cognition
composite score
0.014 ± 1.745 0.002 ± 1.715 -0.001 ± 1.827 0.005 ± 1.773
a. 2 participants in BVAIT and 1 participant in both WISH and ELITE did not complete the baseline cognitive exam
b. Differing sample sizes: Education, BVAIT: n = 503. Income, BVAIT: n = 472, WISH: n = 319, ELITE: n = 597. Marital
status, BVAIT: n = 503. Current smoker, BVAIT: n = 503. BMI, ELITE: n = 641. Blood pressure, ELITE: n = 640. ApoE4
genotype, BVAIT: n = 463, WISH: n = 334, ELITE: n = 637. CES-D score, WISH: n = 348
c. CES-D scores range from 0-60, with higher scores indicating more depressive symptomatology
d. A score of 16 or higher on the CES-D indicates concern for clinical depression

Associations between CIMT and cognitive function at baseline and 2.5 years
At baseline in unadjusted models, higher average CIMT was associated with significantly lower
mean changes in cognitive scores per mm of CIMT. More specifically, higher average CIMT was
associated with significantly lower executive function composite score (β = -0.1292 units per 0.1 mm
CIMT, 95% CI: -0.1849, -0.0736, p<0.0001), verbal memory composite score (β = -0.0845 units per 0.1
mm CIMT, 95% CI: -0.1398, -0.0292, p = 0.0028), visual memory composite score (β = -0.0683 units per
0.1 mm CIMT, 95% CI: -0.1140, -0.0226, p = 0.0034), and global cognition composite score (β = -0.1032
units per 0.1 mm CIMT, 95% CI: -0.1764, -0.0300, p = 0.0058). However, these associations became non-
significant after adjusting for various confounders (Table 2.3).
In adjusted models, CIMT was not significantly associated with executive function composite
score at baseline (β = 0.0221 units per 0.1 mm CIMT, 95% CI: -0.0331, 0.0773, p = 0.43) or at 2.5 years (β
= -0.0340 units per 0.1 mm CIMT, 95% CI: -0.1062, 0.0382, p = 0.36), after adjusting for age, sex, race,
education, income, marital status, ApoE4 genotype, BMI, SBP, HDL, LDL, current smoking status and
study. CIMT was also not significantly associated with verbal memory composite score at baseline (β =
0.0015 units per 0.1 mm CIMT, 95% CI: -0.0457, 0.0749, p = 0.64) or at 2.5 years (β = -0.0488 units per
0.1 mm CIMT, 95% CI: -0.1259, 0.0292, p = 0.22) after adjusting for age, sex, race, education, income,
ApoE4 genotype, CES-D score, BMI, SBP, HDL, LDL, current smoking status and study. CIMT was not
significantly associated with visual memory composite score at baseline (β = 0.0462 units per 0.1 mm
CIMT, 95% CI: -0.0059, 0.0983, p = 0.08) or at 2.5 years (β = 0.0158 units per 0.1 mm CIMT, 95% CI: -
25

0.0479, 0.0796, p = 0.63), adjusting for age, sex, race, education, income, marital status, BMI, SBP, HDL,
LDL, current smoking status and study. CIMT was not significantly associated with global cognition
composite score at baseline (β = 0.0440 units per 0.1 mm CIMT, 95% CI: -0.0303, 0.1183, p = 0.25),
adjusting for age, sex, race, education, CES-D score, BMI, SBP, HDL, LDL, smoking status and study. Trial
treatment assignment did not affect the estimate of the cross-sectional association between CIMT and
cognitive function at 2.5 years.
Table 2.3 Cross-sectional associations between CIMT and cognitive function at baseline and 2.5 years
from multivariable linear regression models
Baseline 2.5 years
Cognitive
Domain
β (95% CI)
a
p-value β (95% CI)
a
p-value Model covariates
Executive
function
0.0221
(-0.0331, 0.0773)
0.43 -0.0340
(-0.1062, 0.0382)
0.36 Age, sex, race,
education, income,
marital status, ApoE
genotype, BMI, SBP,
HDL, LDL, current
smoking status, study
Verbal
memory
0.0146
(-0.0457, 0.0749)
0.64 -0.0488
(-0.1259, 0.0292)
0.22 Age, sex, race,
education, income,
ApoE genotype, CES-
D score, BMI, SBP,
HDL, LDL, current
smoking status, study
Visual
memory
0.0462
(-0.0059, 0.0983)
0.08 0.0158
(-0.0479, 0.0796)
0.63 Age, sex, race,
education, income,
marital status, BMI,
SBP, HDL, LDL,
current smoking
status, study
Global
cognition
0.0440
(-0.0303, -0.1183)
0.25 -0.0338
(-0.1292, 0.0616)
0.49 Age, sex, race,
education, CES-D
score, BMI, SBP, HDL,
LDL, current smoking
status, study
a. Beta estimates are reported in units per 0.1 mm of CIMT
For baseline executive function and verbal memory, a similar trend was found in adjusted
analyses when comparing cognitive measures in participants with CIMT in the lowest quartile (≤0.6885
mm) to participants with CIMT in the highest quartile (0.8393 – 2.1360 mm). That is, there was no
26

significant difference in executive function composite scores (β = 0.0081 units per 0.1 mm CIMT, 95% CI:
-0.0112, 0.0273, p = 0.43), or verbal memory composite scores (β = 0.0082 units per 0.1 mm CIMT, 95%
CI: -0.0127, 0.0291, p = 0.19) between participants with CIMT in the lowest quartile and participants
with CIMT in the highest quartile. There was a weak difference in visual memory composite scores with
participants with CIMT in the highest quartile scoring slightly higher per mm of CIMT than participants
with CIMT in the lowest quartile (β = 0.0136 units per 0.1 mm CIMT, 95% CI: -0.0044, 0.0316, p = 0.047).
There was also a significant difference in global cognition composite scores at baseline with participants
with CIMT in the highest quartile scoring slightly higher per mm of CIMT than participants with CIMT in
the lowest quartile (β = 0.0110 units per 0.1 mm CIMT, 95% CI: -0.0148, 0.0368, p = 0.03).
Examining the association between categorical CIMT and cognitive function by sex did little to
explain this unexpected trend. Among men, there was no significant difference in visual memory
composite scores between men in the highest compared to lowest quartile of CIMT (β = 0.0156 units per
0.1 mm CIMT, 95% CI: -0.0185, 0.0496, p = 0.47). However, among women, the weak difference in visual
memory composite scores between women with CIMT in the highest compared to lowest quartile
remained (β = 0.0149 units per 0.1 mm CIMT, 95% CI: -0.0063, 0.0362, p = 0.046). There was also no
significant difference in global cognition composite scores between men in the highest compared to
lowest quartile of CIMT (β = -0.0078 units per 0.1 mm CIMT, 95% CI: -0.0550, 0.0395, p = 0.29). Among
women, the significant difference in global cognition composite scores between women with CIMT in
the highest compared to lowest quartile remained (β = 0.0153 units per 0.1 mm CIMT, 95% CI: -0.0153,
0.0458, p = 0.02). Given the results from the models with categorical CIMT, no further modeling
including vascular risk factors or treatment assignment was done.
As elevated BMI, SBP, HDL, and LDL contribute to greater CIMT in the cognitive function
pathway and possibly attenuate the association between CIMT and cognitive function, the association at
baseline and 2.5 years was also re-examined without these variables. There were no significant
27

associations between higher average CIMT and any of the cognitive outcomes at baseline or at 2.5 years
in these repeated models (Table 2.4).
Table 2.4. Cross-sectional associations between CIMT and cognitive function at baseline and 2.5 years
from multivariable linear regression models excluding contributors to CIMT (BMI, SBP, HDL, LDL)
Baseline 2.5 years
Cognitive
Domain
β (95% CI)
a
p-value β (95% CI)
a
p-value Model covariates
Executive
function
0.0139
(-0.0404, 0.0682)
0.62 -0.0387
(-0.1094, 0.0319)
0.28 Age sex, race,
education, income,
marital status, ApoE4
genotype, current
smoking status, study
Verbal
memory
0.0040
(-0.0553, 0.0633)
0.89 -0.0654
(-0.1415, 0.1071)
0.09 Age, sex, race,
education, income,
ApoE4 genotype,
CES-D score, current
smoking status, study
Visual
memory
0.0374
(-0.0138, 0.0886)
0.15 -0.0131
(-0.0492, 0.0753)
0.68 Age, sex, race,
education, income,
marital status,
current smoking
status, study
Global
cognition
0.0286
(-0.0445, 0.1017)
0.44 -0.0536
(-0.1463, 0.0390)
0.26 Age, sex, race,
education, CES-D
score, current
smoking status, study
a. Beta estimates are reported in units per 0.1 mm of CIMT
No significant interaction by ApoE4 genotype was found between cognitive associations with
CIMT modeled as a continuous or categorical variable (all interaction p-values > 0.19).
Associations between baseline CIMT and change in cognitive function assessed over 2.5 years
Higher CIMT at baseline was associated with significantly decreased cognitive function over 2.5
years after adjusting only for baseline cognitive composite scores. Higher CIMT at baseline was
significantly associated with decreased executive function composite score (β = -0.0488 units per 0.1
mm CIMT, 95% CI: -0.0807, -0.0170, p = 0.003), verbal memory composite score (β = -0.0668 units per
0.1 mm CIMT, 95% CI: -0.1157, -0.0178, p = 0.008), visual memory composite score (β = -0.0700 units
28

per 0.1 mm CIMT, 95% CI: -0.1057, -0.0343, p = 0.0001), and global cognition composite score (β = -
0.1210 units per 0.1 mm CIMT, 95% CI: -0.1713, -0.0707, p<0.0001).
The average CIMT at 2.5 years across all three clinical trials (0.79 ± 0.12 mm) increased by 0.02
mm from the average CIMT at baseline. There were no significant associations between change in CIMT
and change in cognition in models adjusting only for baseline cognitive function scores (all p-values >
0.06).
The associations between higher CIMT at baseline and 2.5-year change in some of the cognitive
composite scores were altered considerably and no longer significant after adjusting for various
confounders (Table 2.5). Higher CIMT at baseline was not significantly associated with a decrease in
executive function composite score assessed over 2.5 years (β = -0.0179 units per 0.1 mm CIMT, 95% CI:
-0.0548, 0.0190, p = 0.34) after adjusting for baseline executive function composite score, age, sex, race,
education, income, marital status, ApoE genotype, baseline BMI, baseline SBP, baseline HDL, baseline
LDL, current smoking status, and study. There was also no significant association between higher CIMT
at baseline and change in verbal memory composite score assessed over 2.5 years (β = -0.0236 units per
0.1 mm CIMT, 95% CI: -0.0793, 0.0321, p = 0.41) after adjusting for baseline verbal memory composite
score, age, sex, race, education, income, ApoE genotype, baseline CES-D, baseline BMI, baseline SBP,
baseline HDL, baseline LDL, current smoking status, and study. Likewise, higher CIMT at baseline was not
significantly associated with a decrease in visual memory composite score assessed over 2.5 years (β = -
0.0231 units per 0.1 mm CIMT, 95% CI: -0.0640, 0.0177, p = 0.27) after adjusting for baseline visual
memory composite score, age, sex, race, education, income, marital status, baseline SBP, baseline HDL,
baseline LDL, current smoking status, and study. The significant association seen in the earlier model
between higher CIMT at baseline and a change in global cognition composite score assessed over 2.5
years weakened but remained (β = -0.0557 units per 0.1 mm CIMT, 95% CI: -0.1104, -0.0010, p = 0.046)
after adjusting for baseline global cognition composite score, age, sex, race, education, baseline CES-D,
29

baseline BMI, baseline SBP, baseline HDL, baseline LDL, current smoking status, and study. As with the
cross-sectional analysis at 2.5 years, trial treatment assignment did not affect the estimate of the
longitudinal association between CIMT and cognitive function over 2.5 years.
Table 2.5 Associations between baseline CIMT and change in cognitive function assessed over 2.5 years
from multivariable linear regression models
Cognitive domain β (95% CI)
a
p-value Model covariates
Executive function -0.0179 (-0.0548, 0.0190) 0.34 Baseline executive function composite
score, age, sex, race, education,
income, marital status, ApoE4
genotype, baseline BMI, baseline SBP,
baseline HDL, baseline LDL, current
smoking status, study
<65 years old -0.0199 (-0.0638, 0.0241) 0.37
≥65 years old -0.0129 (-0.0842, 0.0584) 0.72
Verbal memory -0.0236 (-0.0793, 0.0321) 0.41 Baseline verbal memory composite
score, age, sex, race, education,
income, ApoE4 genotype, baseline CES-
D score, baseline BMI, baseline SBP,
baseline HDL, baseline LDL, current
smoking status, study
<65 years old -0.0144 (-0.0800, 0.0513) 0.67
≥65 years old -0.0491 (-0.1572, 0.0590) 0.37
Visual memory -0.0231 (-0.0640, 0.0177) 0.27 Baseline visual memory composite
score, age, sex, race, education,
income, marital status, baseline BMI,
baseline SBP, baseline HDL, baseline
LDL, current smoking status, study
<65 years old -0.0356 (-0.0831, 0.0119) 0.14
≥65 years old -0.0160 (-0.0983, 0.0664) 0.70
Global cognition -0.0557 (-0.1103, -0.0010) 0.046 Baseline global cognition composite
score, age, sex, race, education,
baseline CES-D score, baseline BMI,
baseline SBP, baseline HDL, baseline
LDL, current smoking status, study
<65 years old -0.0351 (-0.1005, 0.0303) 0.29
≥65 years old -0.1078 (-0.2100, -0.0056) 0.039
a. Beta estimates are reported in units per 0.1 mm of CIMT.
These associations remained unchanged in the longitudinal models that excluded adjustment
for contributors to CIMT with the exception of the model where change in global cognition composite
score over 2.5 years was the outcome (Table 2.6). There was no significant association between higher
CIMT at baseline and change in executive function composite score assessed over 2.5 years (β = -0.0121
units per 0.1 mm CIMT, 95% CI: -0.0484, 0.0242, p = 0.51); or between higher CIMT at baseline and
change in verbal memory composite score assessed over 2.5 years (β = -0.0226 units per 0.1 mm CIMT,
95% CI: -0.0773, 0.0321, p = 0.42); or between higher CIMT at baseline and change in visual memory
composite score assessed over 2.5 years (β = -0.0254 units per 0.1 mm CIMT, 95% CI: -0.0654, 0.0145, p
30

= 0.21). The weak association seen between higher CIMT at baseline and change in global cognition
composite score assessed over 2.5 years was remained of borderline significance (β= -0.0517 units per
0.1 mm CIMT, 95% CI: -0.1054, 0.0020, p = 0.059).
Table 2.6 Longitudinal associations between CIMT and cognitive function over 2.5 years from
multivariable linear regression models excluding contributors to CIMT (BMI, SBP, HDL, LDL)
Cognitive domain β (95% CI)
a
p-value Model covariates
Executive function -0.0121 (-0.0484, 0.0242) 0.51 Baseline executive function composite
score, age sex, race, education, income,
marital status, ApoE4 genotype,
current smoking status, study
Verbal memory -0.0226 (-0.0773, 0.0321)

0.42 Baseline verbal memory composite
score, age, sex, race, education,
income, ApoE4 genotype, current
smoking status, study
Visual memory -0.0254 (-0.0654, 0.0145)

0.21 Baseline visual memory composite
score, age, sex, race, education,
income, marital status, current
smoking status, study
Global cognition -0.0517 (-0.1054, 0.0020) 0.059 Baseline global cognition composite
score, age, sex, race, education, CES-D
score, current smoking status, study
a. Beta estimates are reported in units per 0.1 mm of CIMT
There were no significant associations between higher CIMT at baseline and change in any of
the cognitive outcomes by sex assessed over 2.5 years (all p-values > 0.09).
Past longitudinal studies have shown that greater CIMT is associated with cognitive decline in
adults aged 65 years and older. However, the results have been mixed in middle-aged adults. The
associations between higher CIMT at baseline and 2.5 year change in cognitive composite scores were
re-examined by age group (<65 versus ≥65 years old) and are summarized in Table 2.5. There was a
significant inverse association between higher CIMT at baseline and change in global cognition
composite score assessed over 2.5 years among adults ≥65 years (β = -0.1078 units per 0.1 mm CIMT,
95% CI: -0.2100, -0.0056, p = 0.039) after adjusting for baseline global cognition composite score, age,
sex, race, education, baseline CES-D score, baseline BMI, baseline SBP, baseline HDL, baseline LDL,
31

current smoking status, and study. No significant associations between CIMT and change in other
cognitive domain composite scores were observed for either age group.
DISCUSSION
This post hoc longitudinal analysis of three randomized trials revealed a weak inverse
association between CIMT at baseline and change in global cognition composite score assessed 2.5 years
later. The global cognition composite score includes cognitive tests comprising the executive function,
visual memory, and verbal memory composite scores, which may cumulatively strengthen the overall
association between CIMT at baseline and change in global cognition composite score from baseline to
2.5 years compared to associations with specific domains.
When longitudinal analysis was stratified by age group, the weak inverse association was
statistically significant among participants aged 65 years and older. Age was inversely associated with a
decline in global cognition at 2.5 years in our data (Pearson’s r = -0.19), and older-aged adults typically
exhibit greater levels of cognitive decline compared to their middle-aged counterparts (Arntzen et al.,
2012). Increasing age is also known to be associated with higher CIMT (van den Munckhof et al., 2018)
and is evident in our data (Pearson’s r = 0.37). It is possible that associations between CIMT and
cognition become evident once CIMT reaches a certain thickness, as in the older participants in this
sample.
When looking at the association between categorical CIMT and cognitive function at baseline,
we observed a weak difference in visual memory composite scores between participants with CIMT in
the highest quartile compared to participants with CIMT in the lowest quartile. Unexpectedly,
participants with CIMT in the highest quartile had higher visual memory composite scores compared to
participants with CIMT in the lowest quartile. A slightly stronger difference in global cognition composite
score was observed between participants with CIMT in the highest quartile scoring higher compared to
participants with CIMT in the lowest quartile. Stratifying this analysis by sex did not fully explain this
32

unexpected finding. A significant positive association between higher CIMT and visual memory as well as
higher CIMT and global cognition was seen for women but not for men. Overall, women had higher
visual memory composite scores (mean (SD): 0.0230 (1.1031)) compared to men (-0.0853 (1.1091)) at
baseline. This sex difference at baseline may have accentuated the association between higher CIMT
and visual memory that was observed for women. However, men had higher global cognition scores
(0.1097 (1.6530)) compared to women (-0.0227 (1.8022)) at baseline. While these baseline analyses
adjusted for a number of confounders, other possible confounding factors such as diet, chronic stress,
diagnosis of chronic obstructive pulmonary disease, and family history of cardiovascular disease that can
affect both CIMT and cognitive function, were not considered.
Stronger associations of cognition with CIMT were observed longitudinally over 2.5 years than
cross-sectionally at baseline or at 2.5 years. This is likely due to the fact that we were looking at change
in cognitive test scores over time while adjusting for baseline cognitive test scores. As such, in this
population of healthy middle- to older-aged adults, changes in cognition may be a more sensitive
indicator of vulnerability to vascular health than a measure of cognition at a single time point. Cross-
sectional associations between CIMT and cognitive functions at baseline and at 2.5 years were not
observed. In a previous cross-sectional analysis using data from BVAIT, baseline measures showed that
thicker CIMT was weakly associated with lower verbal learning abilities, but not with global cognition
(Gatto et al., 2009). Because of clinical trial inclusion/exclusion criteria and the demands of
participation, persons with higher levels of cognitive impairment would not have been enrolled despite
possibly having cardiovascular disease risk factors. This may have created a healthy selection bias that
reduced cross-sectional baseline associations.
Atherosclerosis may adversely affect cognition through several mechanisms, including cerebral
hypoperfusion, as blood vessel stenosis may result in reduced blood flow and oxygen supply to the
brain. Reduced intracerebral perfusion can damage brain tissue and lead to declines in cognitive
33

function (Appelman et al., 2010). Indeed, low cerebral blood flow velocity has been found to be
associated with cognitive decline (Ruitenberg et al., 2005). Higher CIMT is a risk factor for subclinical
brain infarcts, or silent strokes, which are in turn associated with declines in cognitive function and an
increased risk for dementia (Reitz et al., 2006; Schneider et al., 2003; Thong et al., 2013). Silent strokes
may not have noticeable symptoms and are thought to be common among older people, with the
estimated prevalence ranging from 5 to 62% in populations with mean ages of 54 to 79 years old,
respectively (Fanning et al., 2014). This mechanisms is entirely plausible in this study population of
otherwise healthy middle- to older-aged adults. Although this study did not directly measure
cerebrovascular stenosis through methods such as brain MIR or transcranial Doppler, carotid artery
atherosclerosis serves as a surrogate marker for intracerebral atherosclerosis (Arntzen et al., 2012).
There are several potential explanations for the lack of agreement in findings from this study
and from previous studies. It is not understood how well the commonly used tests for cognition are in
agreement with each other. For example, there are several different tests that can be used to assess
verbal memory such as the CVLT, the Hopkins Learning Test (HVLT), and the Word List task from the
Consortium to Establish a Registry for Alzheimer’s Disease-Neuropsychological Assessment Battery
(CERAD-NAB). Not only do these tests differ in difficulty, but the CVLT and CERAD-NAB have also been
found to differ in sensitivity in detecting episodic memory impairments (Beck et al., 2012). Another
potential reason is that some studies only use a single tool to assess cognitive impairment while others
use an entire battery. It is possible that studies using extensive batteries may be more likely to detect
impairments in particular cognitive domains. Finally, the association between subclinical atherosclerosis
and cognitive impairment may differ depending on the operational definition of subclinical
atherosclerosis. While some studies may use average CIMT to define subclinical atherosclerosis, others
may use maximum CIMT. This association may also differ depending on the site of IMT measurement.
For example, studies may select either the left or right common carotid artery, or the common versus
34

internal carotid artery on which to base their measurements (Johnston et al., 2004; Romero et al.,
2009). CIMT used in this study was developed and validated with serial quantitative coronary
angiography as a measure of generalized atherosclerosis (Mack et al., 2000) and CIMT is a predictor of
clinical cardiovascular events; however, unilateral measurement of CIMT from plaque-free common
carotid artery may underestimate the atherosclerosis burden of cerebrovascular arteries.
Strengths of this study include the longitudinal design which allows us to address the
directionality of the association between CIMT and cognitive function, a large sample size, and a
thorough battery of neuropsychological tests.
Aspects of this study that would limit the generalizability include the inclusion of more women
than men due to the convenience of available data from three clinical trials, two of which only enrolled
postmenopausal women. Because of the focus on postmenopausal women in the WISH and ELITE
studies, the study sample included few individuals who were middle-aged (e.g., 40-49) or elderly (>80
years of age). In addition, given clinical trial selection criteria, individuals with a history of cardiovascular
disease and diabetes were excluded. As such, the findings of this study may not be generalized to
individuals with cardiovascular disease, premenopausal women, or elderly individuals >80 years. Lastly,
the findings may not be generalizable to populations outside of the USA.
CONCLUSION
In summary, in this longitudinal study of 1,495 healthy adults with a mean age of 61 years,
greater carotid artery intima-media thickness at baseline had a weak, statistically significant inverse
association with change in global cognition assessed over 2.5 years. No cross-sectional associations were
seen between higher CIMT and declines in cognitive function at baseline or at 2.5 years. This study
provides evidence that subclinical atherosclerosis of the carotid artery may be a modifiable correlate of
cognitive decline in middle and older age. Further research is needed to expand and support these
findings in more representative samples. Given that the number of people with AD and related
35

dementias is expected to rise with an aging world population and increasing life expectancy, detection
and management of cardiovascular disease early on could reduce the risk for cognitive impairment in
middle- to older-aged adults.

36

CHAPTER 3
INTRODUCTION AND BACKGROUND ON SOCIAL SUPPORT AND COGNITIVE FUNCTION

Social support and cognitive function
In addition to subclinical atherosclerosis, there has been interest in other modifiable risk factors
for cognitive impairment and dementia in middle- to older-aged adults such as social relationships. If
there is a causal association, improving social relationships offers a potentially simple method of
promoting better cognitive outcomes and decreasing dementia risk. Social relationships can be
characterized by a number of factors such as social activities, networks, and support (Berkman et al.,
2000; Kelly et al., 2017). The structural aspects of social relationships are represented by social activities
and networks while the functional aspects are represented by social support (Kuiper et al., 2015; Kuiper
et al., 2016).
Social support can be further delineated by five dimensions: emotional support, informational
support, instrumental support (synonymous with tangible support), positive social interaction, and
affectionate support (Sherbourne & Stewart, 1991). Emotional support can be defined by expressions of
empathy, love, trust, and caring. Informational support can be defined by offerings of advice,
suggestions, and information. Instrumental support can be defined by the provision of tangible help
(including financial aid) and services. Positive social interaction is defined by the availability of other
persons to engage in enjoyable activities with. Finally, affectionate support is defined by expressions of
affection and love (Priede et al., 2018).
Social support (both perceived and actual) is an aspect of social relationships that may affect
cognitive function in middle- to older-aged adults. Past studies examining the association between
overall amount of perceived social support and cognitive function have generally found a protective
association with some exceptions. Several explanations for the beneficial association of social support
37

on cognitive function have been postulated. As early as the 1980s, it was thought that social support
impacted cognitive function by buffering the negative effects of stress on cognition (Cohen & Wills,
1985). It has also been hypothesized that individuals who have a good degree of social support
experience less physiological arousal (i.e., fewer escalations in blood pressure, respiratory rates). Being
in a state of heightened physiological arousal has been shown to be associated with poor cognitive
function and cognitive decline (Seeman et al., 2001). In addition, it has been postulated that social
activities (including the provision and receipt of social support), may actually provide cognitive
stimulation which can help individuals increase their cognitive reserve (Scarmeas & Stern, 2003).
Research has shown that individuals with more cognitive reserve may be better equipped to cope with
symptoms of degenerative brain changes associated with dementia (Stern, 2002). An alternative
explanation is rooted in the social control theory. That is, retaining close relationships with friends and
family can encourage healthier behaviors which in turn may help to maintain good cognitive function
(Umberson, 1987).
Measures of social support
Validated instruments
Social support instruments typically assess perceived availability of social support rather than
the actual received amount of social support. This is because the actual amount of social support
received is likely to be related to an individual’s needs at a particular point in time and may not reflect
the actual amount of support available (Sherbourne & Stewart, 1991). Depending on the instrument
used to assess perceived social support, the names for support dimensions vary. For example, the
Interpersonal Support Evaluation List (ISEL) measures the following four dimensions: appraisal,
belonging, self-esteem, and tangible support. Appraisal support, the perceived availability of someone
to discuss problems with, is closest to emotional/informational support. Belonging, the perceived
availability of other persons to do things with, is closest to positive social interaction. Self-esteem, the
38

perceived availability of having a positive comparison when comparing one’s self to others, is not one of
the usual domains of social support (Sims et al., 2014). Other social support instruments may only
provide an overall index of social support or assess specific support dimensions. For example, the Two-
Way Social Support Scale only measures perceived emotional and instrumental support. While all five
dimensions are covered in the Medical Outcomes Study-Social Support Survey (MOS-SSS), the emotional
and information support domains are sometimes combined given the overlap between their respective
items (Sherbourne & Stewart, 1991).
Indirect measures of social support
Other studies have examined the association of cognitive function with indirect measures of
social support such as social isolation, loneliness, and marital status. Although the terms social isolation
and loneliness may appear interchangeable, there are important differences. Social isolation is one very
specific aspect of social relationships, while experiences and feelings of loneliness are not entirely
dependent on the lack of social interactions with others (Kelly et al., 2017). However, loneliness is
something that could be taken into account in scenarios where there is little perceived social support.
Previous findings
Associations between social support and overall cognitive function
A 2017 systematic review performed by Kelly et al concluded the literature suggests a positive
association between social support and global cognition (Kelly et al., 2017). Cross-sectional analysis of a
cohort of 4,993 Taiwanese adults aged 65 years and older found that strong social support was
associated with higher scores (indicating better cognitive function) on the Short Portable Mental Status
Questionnaire (Yeh & Liu, 2003). This association was also found in other populations. A study on 1,091
Scottish adults aged 70 in the Lothian Birth Cohort 1936 (LBC1936) found that more social support was
associated with better verbal reasoning but not memory at age 70 (Gow et al., 2013). Another cross-
sectional study on 2,211 Mexican adults aged 50 and older found that medium and high levels of social
39

support were associated with lower prevalence of cognitive impairment for adults aged 71-80 years old
(Zamora-Macorra et al., 2017). Unlike the findings from the LBC1936 study, this study did not find the
same association for adults <71 years old or over 80 years old. A more recent cross-sectional study on
17,206 adults aged 45 years and older from the Behavioral Risk Factor Surveillance System (BRFSS)
cohort found that participants who reported insufficient social support were more likely to report
experiencing confusion or memory loss (Weng et al., 2020).
Some studies have not found an association between perceived social support and cognitive
function. In a cross-sectional study on 273 older Hispanic immigrants aged 70-100 years old living in
Miami, Florida, social support was not related to cognitive function (Brown et al., 2009). Two
longitudinal studies also did not find an association between social support and cognitive function over
time. A study on 13,119 adults from the Atherosclerosis Risk in Communities (ARIC) study with a mean
age of 57.1 years found that at baseline, more social support was associated with better global
cognition. However, this association did not remain at the 20-year follow-up (Kats et al., 2016). Another
longitudinal study on 6,863 adults with a mean age of 55.8 years from the Whitehall II cohort found that
social support had no influence on cognitive changes from baseline over 10-years of follow-up (Liao et
al., 2018).
Associations between social support and cognitive domains
Findings from studies on the associations between social support and specific cognitive domains
have been mixed. The 2017 systematic review performed by Kelly et al summarized the literature
suggested a positive association between social support and episodic memory, but not between social
support and attention or processing speed (Kelly et al., 2017). A cross-sectional study on 623 adults with
a mean age of 56.7 years from the Wisconsin Registry for Alzheimer’s Prevention (WRAP) found that
social support was positively associated with speed and flexibility but not with immediate memory,
verbal memory, or working memory. The relationship between social support and speed and flexibility
40

also diminished when a participant had one or two copies of the ApoE e4 allele (Zuelsdorff et al., 2013).
The previously mentioned study on the LBC1963 cohort found an association between social support
and faster processing speed, contradicting the findings summarized in the systematic review performed
by Kelly et al. However, this association disappeared once depression was taken into account (Gow et
al., 2013). A cross-sectional study on 175 adults with a mean age of 66.32 years found that social
support was actually associated with worse nonverbal memory and response inhibition after adjusting
for age, sex, education, depressive symptomatology (Beck Depression Inventory), BMI, SBP, total
cholesterol and fasting glucose (Sims et al., 2014). This same study also found that tangible support
(synonymous with instrumental support) was associated with poorer visuospatial ability,
visuoconstructional ability, nonverbal memory, perceptuo-motor speed, executive function, attention
and working memory, and verbal memory. Given these unexpected findings, the authors postulated that
for some individuals, the receipt of social support that cannot be reciprocated can be stressful and
upsetting.
Associations between dimensions of social support and cognitive domains
Fewer studies have investigated the associations between the individual dimensions of social
support and cognitive function or between the individual dimensions of social support and specific
cognitive domains. Low perceived instrumental support in a cohort of 223 adults aged 60 years and
older from the Neurocognitive Outcomes of Depression in the Elderly (NCODE) study was associated
with worse verbal memory, processing speed, and executive function at 1-year follow-up (Dickinson et
al., 2011). In another study on 802 adults from the Glostrup 1914 Cohort, instrumental support was not
associated with cognitive function over 30-year follow-up from age 50 to 80 (Gow & Mortensen, 2016).
Emotional support in particular has been linked to better cognitive function in some longitudinal
studies. In a study on 1,189 adults aged 70-79 years from the MacArthur Studies of Successful Aging,
greater emotional support at baseline was associated with better cognitive function at both baseline
41

and at 7.5-year follow-up (Seeman et al., 2001). Emotional support was also associated with higher
MMSE scores at 12-year follow-up in a community-based sample of 354 adults aged 50 years and older
(Holtzman et al., 2004), and with higher Montreal Cognitive Assessment (MoCA) scores at 1-year follow-
up in a cohort of 121 Japanese adults with a mean age of 73.9 years from the Togo Town Study (Noguchi
et al., 2019). Conversely, a study on 175 adults with a mean age of 66.3 years did not find that appraisal
support (similar to emotional/informational support) had any association with response inhibition,
visuospatial ability, visuoconstructional ability, nonverbal memory, perceptuo-motor speed, executive
function, attention, or verbal memory (Sims et al., 2014). The previously mentioned Whitehall II cohort
study also did not find that any of the individual social support domains had any association with
cognitive outcomes. More specifically, confiding support (similar to a combination of emotional and
informational support) and practical support (similar to instrumental support) were not found to be
protective against declines in executive function or memory over 10 years (Liao et al., 2018).
Association between social support and cognitive function may differ depending on the source of support
Past studies have also shown that the benefits for cognitive function may differ depending on
the source of social support. In a study on 120 Chinese adults aged 60 years and older, social support
from family members had a significant positive association on cognitive function (Zhu et al., 2012). In
this same study, neither social support from friends nor significant others had the same association on
cognitive function. Other studies have shown the opposite. The previously mentioned study on adults in
the Togo Town Study found that social support from family (either co-residing or not) was not
associated with MoCA scores at 1-year follow-up (Noguchi et al., 2019). However, social support,
especially emotional support, from friends and neighbors was positively associated with MoCA scores at
1-year follow-up. The authors reasoned that this unexpected finding might be explained by the fact that
friends are typically close in age and share similar life experiences and lifestyles.

42

Social isolation, loneliness, and cognitive function
Much of the previous research has found that social isolation and loneliness are negatively
associated with cognitive function. A longitudinal study on adults aged 75 years and older from the
Helsinki Aging Study found that feelings of loneliness were not associated with a low MMSE score at
baseline, at 1-year, or at 5-year follow-up. However, this study did find that feelings of loneliness
increased the risk for cognitive decline at 10-year follow-up (Tilvis et al., 2004). A study on Chinese
adults aged 65 years and older from the Chinese Longitudinal Healthy Longevity Survey found that social
isolation, indicated by marital status and living arrangement, was associated with cognitive decline over
9 years (Zhong et al., 2017). A more recent study using data from the same cohort found that loneliness
was associated with cognitive impairment at 3-year follow-up, and that this association did not differ by
sex (Zhou et al., 2019).
As for association with specific cognitive domains, a study on 823 adults with a mean age of 82.7
years found that loneliness was associated with worse global cognition, semantic memory, perceptual
speed, and visuospatial ability at both baseline and 4-year follow-up (Wilson et al., 2007). While there
were also associations between loneliness and worse episodic memory and working memory at
baseline, these particular associations did not remain at 4-year follow-up. Another study using data from
the English Longitudinal Study of Ageing found that social isolation and loneliness were significantly
associated with decreases in verbal fluency, immediate recall, and delayed recall at 4-year follow-up
(Shankar et al., 2013).
Study variations that may contribute to different findings
In general, past studies have found that social support is positively associated with cognitive
function in middle- to older-aged adults. However, some past studies have failed to find this association.
One reason for the lack of consistency in the findings may be the different measures used for perceived
social support. While some studies used validated instruments such as the Duke Social Support Index,
43

ISEL, MOS-SSS, or the Multi-Dimensional Scale of Perceived Social Support (MSPSS), other studies
adapted a few select questions from full instruments, or simply asked about marital status/living
arrangements, or frequency of social contact. Questions about marital status/living arrangements and
frequency of social contact may only be indirect measures of perceived social support at best.
How cognitive function is assessed also varies greatly by study. While some studies use entire
neuropsychological batteries, others only used a single instrument such as the MoCA or the Short
Portable Mental Status Questionnaire to assess cognitive outcomes. More research is needed on how
the individual dimensions of social support affect the different domains of cognitive function as only a
handful of studies examined these associations.
Potential issues with reverse causation
Declines in social relationships, and subsequently social support, can be an early sign of
cognitive dysfunction. According to the health selection hypothesis, poor cognitive function can affect
both the quantity and quality of social relationships, and subsequently social support (Liao et al., 2018).
Indeed, for dementia patients, the beginnings of cognitive decline and behavior changes can make
maintaining and reciprocating social relationships rather difficult (Amieva et al., 2010). One way to
address this issue of reverse causation is to examine the association between social support and
cognitive decline longitudinally. The ELITE study allows for this with assessments of cognitive function at
2.5 and 5 years.

44

CHAPTER 4
SOCIAL SUPPORT AND COGNITIVE FUNCTION IN POST-MENOPAUSAL WOMEN

INTRODUCTION
Social support, the perception of the availability of assistance or support from others, has been
shown to positively impact cognitive function and decrease dementia risk in older adults (Holtzman et
al., 2004; Seeman et al., 2001; Yeh & Liu, 2003; Zhu et al., 2012). One explanation for the beneficial
association is that social support helps to buffer the negative effects of stress (Cohen & Wills, 1985). The
transactional nature of social support (both provision and receipt of social support) may also provide
cognitive stimulation which in turn may contribute to an individual’s cognitive reserve (Scarmeas &
Stern, 2003). Cognitive reserve refers to individual differences in cognitive processes or neural networks
affecting the performance of tasks. Cognitive reserve may afford resilience against age-related brain
changes or dementia-related pathology such that individuals with greater cognitive reserve may be
better able to cope with symptoms of degenerative brain changes (Stern, 2002). A third explanation is
that maintaining close relationships with friends and family can foster healthier behaviors such as eating
well and exercising together which in turn may help to maintain good cognitive function (Umberson,
1987).
In addition to overall perceived amount of social support, studies examining the association
between social support and cognitive function in older adults often measure dimensions of social
support: emotional support, informational support, instrumental support, positive social interaction,
and affectionate support.
Past studies of social support and overall cognitive function have generally found a positive
association. Findings from studies of social support and specific cognitive domains have been mixed and
only a few studies have examined the associations between the individual dimensions of social support
45

and cognitive function. Reasons for the lack of consistency in the findings include use of different
measures for perceived social support and for cognitive function.
This study adds to existing knowledge on perceived social support and cognitive function by
utilizing data gathered from the Early versus Late Intervention Trial with Estradiol (ELITE) and its
psychosocial substudy. Participants in ELITE completed a thorough battery of standardized
neuropsychological test at baseline and at follow-up visits. The scores from the various
neuropsychological tests were combined into composite scores for executive function, verbal memory,
visual memory, and global (overall) cognition.
ELITE participants who agreed to participate in the optional psychosocial substudy completed a
series of questionnaires that assessed psychosocial characteristics and life events every six months. The
questionnaires included the Medical Outcomes Study Social Support Survey (MOS-SSS) which gauges an
individual’s perceived availability of social support. Since its development in the 1990s, the MOS-SSS has
been widely used to assess perceived social support (Khuong et al., 2018). An overall social support
score was calculated from the MOS-SSS with higher scores indicating greater amounts and availability of
perceived social support. MOS-SSS subscales were also used to measure dimensions of social support
including emotional, instrumental, and informational support.
MATERIALS AND METHODS
Study design
ELITE design
ELITE was a double-blind, placebo-controlled, randomized clinical trial conducted from July 2005
to February 2013. The trial was designed to examine the effects of oral 17β-estradiol (estrogen) on the
progression of subclinical atherosclerosis and cognitive decline in healthy post-menopausal women.
Similar to BVAIT and WISH, eligible participants did not have diabetes or clinical signs/symptoms of
cardiovascular disease. Menopause was defined as absence of menses for at least 6 months or bilateral
46

oophorectomy, as well as a serum estradiol level lower than 25 pg/ml (92 pmol per liter). ELITE was
specifically designed to test the hypothesis that estrogen supplementation would reduce atherosclerosis
progression when given in early menopause (<6 years after menopause), but not in late menopause
(≥10 years after menopause). Secondary cognitive outcomes evaluated a similar hypothesis. A total of
643 women were randomized to treatment (oral 17B-estradiol 1 mg/day plus micronized progesterone
[45 mg] as a 4% vaginal gel for women with a uterus) or placebo in a 1:1 allocation ratio, within strata
defined by length of time since menopause (early menopause being <6 years past menopause and late
menopause being ≥ 10 years past menopause). Other stratification factors included presence or absence
of a uterus at randomization. Women with a uterus also received progesterone as a 4% vaginal gel or
matched placebo. Cognitive function was measured using a battery of 14 neuropsychological tests
administered at baseline, 2.5 years, and 5 years after randomization.
In the primary outcome analysis, oral estradiol therapy was associated with less progression of
subclinical atherosclerosis (CIMT) compared to placebo in the early menopause group; estrogen and
placebo groups did not differ on atherosclerosis progression in the late menopause group (Hodis et al.,
2016). Oral estradiol therapy was also not found to affect verbal memory, executive functions, or global
cognition in either the early or late menopause groups (Henderson et al., 2016).
ELITE psychosocial substudy
Women from the ELITE clinical trial were invited to participate in the psychosocial substudy,
which was initiated while the ELITE trial follow-up was in progress and had been completely randomized
(n = 643). At the time the psychosocial substudy began in January 2009, there were 512 active
participants in ELITE. Of these, 448 women (87.5%) agreed to participate in the psychosocial substudy.
Consenting women were given a packet of questionnaires to complete at home, that included
assessment of social support. The questionnaires assessed psychosocial characteristics and life events
and were given every six months until the substudy ended in December 2012.
47

Social support assessment
The MOS-SSS is a 20-item self-administered questionnaire that measures the degree to which an
individual perceives the availability of functional social support in their lives (Sherbourne & Stewart,
1991). The first item is an open-ended question which asks an individual about the size of their social
network by indicating the number of close friends and relatives they have. Responses to the remaining
19 items are captured on a 5-point Likert scale ranging from (1) “None of the time” to (5) “All of the
time”. An overall social support score ranging from 19-95 is calculated either by averaging scores across
all items or by transforming the total score. Transformed scores are calculated by taking the difference
between the observed score and minimum possible score, dividing by the difference between the
maximum possible score and minimum possible score, and multiplying by 100:

Transformed scores have a possible range of 0-100. Higher average or transformed scores indicate a
greater amount and availability of perceived social support (Sherbourne & Stewart, 1991).
The MOS-SSS can also be divided into subscales that represent five dimensions of social support:
emotional support, informational support, tangible support (synonymous with instrumental support),
positive social interaction, and affection support (Priede et al., 2018). Emotional support can be defined
by expressions of empathy, love, trust, and caring. Informational support can be defined by offerings of
advice, suggestions, and information. Instrumental support can be defined by the provision of tangible
help (including financial aid) and services. Positive social interaction is defined by the availability of other
persons to engage in enjoyable activities with. Finally, affectionate support is defined by expressions of
affection and love (Priede et al., 2018).
(observed score – minimum possible score)
(maximum possible score – minimum possible score)
x 100
48

Previous psychometric evaluation of the MOS-SSS, including that of a modified, eight-item
version, has shown good internal reliability in different populations (Khazaee-Pool et al., 2018; Khuong
et al., 2018; Moser et al., 2012; Zucoloto et al., 2019) (all Cronbach’s alpha ≥0.91). The scores from the
instrument were also found to be reliable with stable reassessment over a 1-year period (Sherbourne &
Stewart, 1991). The different dimensions of social support measured by the MOS-SSS have been shown
to be distinct from structural measures of social support such as social network size and from related
health measures.
Cognitive function assessment
A battery of cognitive and neuropsychological assessments were administered by a single
trained psychometrist in a standardized order to all ELITE participants. The battery included the
following 14 assessments:
 Symbol Digit Modalities Test (SDMT)
 Trail Making Test Part B (Trails B)
 Judgment of Line Orientation, Form H (JLO)
 Block Design [Wechsler Adult Intelligence Scale, 3
rd
Edition (WAIS-III)]
 Letter-Number Sequencing [Wechsler Memory Scale, 3
rd
Edition (WMS-III)] (LNS)
 Category fluency (animal naming, 60 seconds) (Animals)
 Boston Naming Test, 30-item version (BNT)
 Shipley Institute of Living Scale (Shipley), Abstraction Scale
 California Verbal Learning Test, 2
nd
edition (CVLT-II), immediate recall (IR) and delayed recall
(DR)
 Logical Memory, immediate recall and delayed recall
 Faces I (IR) and II (DR) (WMS-III)
49

The assessments in this battery were selected to be sensitive to age-related changes in cognition
and were chosen to evaluate different cognitive functions and abilities with a focus on executive
function, verbal memory, and visual memory. Participants in ELITE completed the battery at baseline,
2.5 years, and 5 years.
Results from the 14 tests were used to generate composite scores for executive function, verbal
memory, visual memory, and global cognition. The executive function composite score was calculated
as a weighted average of the following cognitive tests: Symbol Digit Modalities Test, Trail Making Test
Part B, Shipley Abstraction Scale, Letter-Number Sequencing, and Category fluency. The verbal memory
composite score was calculated as a weighted average of the CVLT-II (immediate and delayed recall) and
Logical Memory (immediate and delayed recall) tests. The visual memory composite score was
calculated as a weighted average of the Faces I (immediate recall) and Faces II (delayed recall) tests. The
global cognition composite score was calculated as a weighted average of all 14 assessments. For each
participant, all test scores were converted to a standardized Z score at baseline, 2.5 years, and 5 years.
The Z scores were calculated using the baseline mean and standard deviation for each test. To calculate
the composite scores, the Z scores for the appropriate cognitive tests were summed and then weighted
by the inverse inter-test correlation matrix (Gatto et al., 2009; Henderson et al., 2016; Henderson et al.,
2012; Henderson et al., 2013).
Statistical methods
Before proceeding with analyses, the psychosocial dataset was modified. If a participant skipped
a question on the MOS, an overall social support score could not be calculated. As such, within-
participant average scores across non-missing MOS items for each participant were imputed for missing
items (0.4% of all MOS items). Variables for the raw overall social support score as well as the five social
support dimensions were calculated. Total scores that contained decimals following imputation of
average scores were rounded to the closest integers. From these total scores, average and transformed
50

scores were calculated for overall social support and the five social support dimensions. Transformed
scores were used in statistical analyses.
Since the psychosocial substudy was initiated sometime after the main ELITE study had begun, a
three-level variable was created to categorize the timing of visits in both the psychosocial and ELITE
datasets for merging purposes. In the psychosocial dataset, the three levels defined psychosocial visits
that were completed at 6-18, 24-46, and 48-72 months of trial follow-up. In the modified ELITE dataset,
the three levels comprised cognitive assessment visits that were completed at baseline, 30-46
(corresponding to approximate 3-year assessment), and 52-64 months (corresponding to approximate 5-
year assessment). The psychosocial dataset was then merged with the modified ELITE dataset on
participant ID and this three-level variable. After merging, a window variable was created by taking the
difference between the timing of the psychosocial and ELITE visits. Because participants completed the
psychosocial questionnaire packets at home, psychosocial completion dates often did not coincide
exactly with cognitive dates. This window variable was used to indicate any psychosocial and ELITE visits
that were completed within 6 months of each other and would be included in the analysis. The numbers
of cognitive-psychosocial paired visits that were completed within 6 months of each other are
summarized in Table 4.1. Any duplicate cognitive visits in the merged datasets were dropped. A total of
392 6-month psychosocial-cognitive assessments were identified.

51

Table 4.1 Cognitive and psychosocial visits completed within 6 months of each other
Psychosocial visit (month)
6 24 30 36 48 54 60 66 Total
Cognitive visit (month)
0 6 6
30 102 56 94 252
32 13 18 31
34 4 9 13
36 2 1 3
52 27 4 31
54 14 3 17
56 1 1
58 1 1
60 21 10 4 35
64 2 2
Total 6 102 75 122 41 30 12 4 392

Univariate linear regression was used to examine the unadjusted associations between overall
social support as well as the different social support dimensions and variables associated with cognitive
function. Covariates that were significantly associated with social support (at p<0.05) included marital
status, BP medication usage on trial, BMI, CES-D, education, HDL cholesterol, and income. These
covariates were included in linear mixed effects models examining the associations between cognitive
function and transformed social support scores. Because BMI and HDL are correlated (r = -0.35), only
BMI was included in the mixed effects models. CES-D was eventually excluded as it was a potential
mediator between social support and cognition. Although age and race were not significantly associated
with social support, they were included in the linear mixed effects models based on prior knowledge.
Univariate linear regression was also used to examine the unadjusted associations between various
domains of social support (overall, emotional, informational, instrumental, positive social interaction,
affectionate) and cognitive function (executive function, verbal memory, visual memory, global
cognition).
Analyses were performed with linear mixed effects modeling since some participants had
multiple cognitive-psychosocial paired visits. The mixed effects models included a random intercept at
the participant level, which modeled individual participant deviations from the sample mean scores for
52

the cognitive outcome variables. Fixed effects included age, marital status, use of blood pressure
medications on trial, BMI, education, income, race, treatment assignment (placebo or active treatment)
and social support scores. As marital status is a potential indicator of social support, the associations
between social support and cognitive outcomes were also examined without marital status in the mixed
effects models. Because positive social interaction, or the availability of other persons to engage in
enjoyable activities with, is known to decrease as people age, the interaction by age group (<65, ≥65
years old) was also examined for the association between positive social interaction and cognitive
outcomes. Since the majority of participants were <65 years old, the interaction by age group using a
different cutoff (<60, ≥60 years old) was examined as well. Lastly, one past study has shown that greater
social support was associated with better speed and flexibility as measured by performance on the Trails
A, Trails B, and Stroop Color-Word tests, and that this relationship was diminished by the presence of
the ApoE4 allele (Zuelsdorff et al., 2013). The interaction by ApoE4 status was also examined among the
ELITE psychosocial substudy participants for the association between and social support and
performance on the Trails B test as well as with all cognitive outcomes. Adjustments for multiple testing
was not used because there were some hypotheses about associations overall with cognition but not
with specific cognitive domains. Table 4.2a summarizes the number of psychosocial visits completed by
all 448 psychosocial substudy participants. Restricting analyses to cognitive and psychosocial visits that
were completed within 6 months of each other reduced the number of observations in the analysis to
392 paired cognitive-psychosocial visits completed by 335 psychosocial substudy participants. Table 4.2b
summarizes the number of psychosocial visits completed by the 335 psychosocial substudy participants.
Previous studies have shown that the number of contacts in an individual’s social network likely
decreases with increasing age. As such, correlations between the number of friends and MOS-SSS scores
were also examined (Bhattacharya et al., 2016).

53

Table 4.2a Summary of the number of completed psychosocial visits in the ELITE cohort
Number of
psychosocial visits
# of participants Cumulative # of
participants
1 46 46
2 47 93
3 49 142
4 75 217
5 53 270
6 75 345
7 67 412
8 36 448

Table 4.2b Summary of the number of completed psychosocial visits in the ELITE cohort when restricted
to cognitive-psychosocial visits completed within 6 months of each other
Number of
psychosocial visits
# of participants Cumulative # of
observations
1 278 278
2 57 392

RESULTS
Study sample
Baseline characteristics for the 335 ELITE women providing cognitive and psychosocial data for
the analyses are summarized in Table 4.3. As with the main ELITE cohort, the majority of participants
were non-Hispanic White (69.3%) and married (57.9%). Study participants included in the analyses had
an average (SD) age of 60.9 (7.1) years, and were highly educated (16.0 (1.9) years of education
correlating with a Bachelor’s degree), had an average annual income of $69,200 ($31,700), and were
overweight (mean BMI = 27.0 (5.5) kg/m
2
). Very few women reported being current smokers (3.0%), and
about half reported consuming alcohol regularly (yes/no) (50.2%). Approximately one third (33.1%)
reported current use of anti-hypertensive medication and a smaller proportion (18.5%) reported current
use of cholesterol lowering medication. The average systolic blood pressure was 115.4 (12.9) mmHg,
and the average diastolic blood pressure was 74.4 (8.4) mmHg. Almost a third of participants (31.4%)
were positive for the ApoE4 genotype. The average CES-D score at baseline was low (7.8 out of a
possible 60) and 17.6% scored 16 or above (the cut off score indicating concern for clinical depression)
54

(American Psychological Association, 2011). One participant included in the analysis did not complete a
cognitive assessment at baseline but did complete cognitive and psychosocial assessments at
subsequent visits.
55

Table 4.3 Baseline characteristics for ELITE women included in the cognitive-psychosocial analysis (n =
335)
Variable Mean ± SD or Number (%)
Age (years) 60.9 ± 7.1
Age ≥65 years at baseline 86 (25.7%)
Age ≥60 years at baseline 168 (50.2%)
Race
Non-Hispanic White 232 (69.3%)
Non-Hispanic Black 25 (7.5%)
Hispanic 43 (12.8%)
Asian or Pacific Islander 35 (10.5%)
Other 0 (0%)
Education (# of years) 16.0 ± 1.9
Annual income $69.2 ± $31.7
1

Marital status
Single, never married 26 (7.8%)
Married 194 (57.9%)
Separated 7 (2.1%)
Divorced 87 (26.0%)
Widowed 21 (6.3%)
Current smoker 10 (3.0%)
Alcohol use 168 (50.2%)
Use of anti-hypertensives at baseline 111 (33.1%)
Use of cholesterol lowering medication at
baseline
62 (18.5%)
BMI (kg/m
2
) 27.0 ± 5.5
1

Blood pressure (mmHg)

Systolic 115.4 ± 12.9
1

Diastolic 74.4 ± 8.4
1

ApoE4+ 104 (31.4%)
1

CES-D score 7.8 ± 7.8
CES-D score ≥ 16
2
59 (17.6%)
Executive function composite score 0.066 ± 1.361
1

Verbal memory composite score 0.085 ± 1.324
1

Visual memory composite score -0.001 ± 1.087
1

Global cognition composite score 0.078 ± 1.892
1

1. Differing sample sizes: income: n = 315, BMI: n = 334, SBP/DBP: n = 333, ApoE4+ status: n = 331, cognitive composite
scores: n = 334.
2. A score of 16 or higher on the CES-D indicates concern for clinical depression.

Average transformed social support scores for the 392 cognitive-psychosocial paired visits
completed within 6 months of each other are summarized in Table 4.4.

56

Table 4.4 Average transformed social support scores (possible range 0-100) for cognitive-psychosocial
paired visits completed within 6 months of each other (n = 392)
Mean ± SD
Overall social support 76.5 ± 20.9
Emotional social support 77.2 ± 22.2
Instrumental social support 70.7 ± 28.2
Informational social support 77.5 ± 21.1
Positive social interaction 78.1 ± 21.5
Affectionate support 79.5 ± 25.0

Associations between perceived social support and cognitive function
Results from the mixed effects models are summarized in Table 4.5a. Instrumental social
support had a borderline significant positive association with visual memory (β = 0.0044 units per point
increase in transformed instrumental social support score, 95% CI: -0.00002, 0.0088, p = 0.051) after
adjusting for age, marital status, use of blood pressure medications on trial, BMI, education, income,
race, and treatment assignment. In addition, positive social interaction had a significant positive
association with global cognition (β = 0.0092 units per point increase in transformed positive social
interaction score, 95% CI: 0.0001, 0.0182, p = 0.047). There were no other significant associations
between perceived overall, emotional, instrumental, informational, positive social interaction, or
affectionate support and any of the cognitive outcomes. When marital status, a potential indicator of
support, was excluded from the models, only the association between positive social interaction and
global cognition remained significant (β= 0.0090 units per point increase in transformed positive social
interaction score, 95% CI: 0.00004, 0.0180, p = 0.049). Other associations between perceived social
support and cognition without marital status are summarized in Table 4.5b.

57

Table 4.5a Associations from linear mixed effects models between transformed social support scores
and cognitive outcomes, limited to cognitive-psychosocial visits completed within 6 months of each
other (n = 322)
Cognitive domain
Executive
function
Verbal memory Visual memory Global cognition
Social support domain
β (95% CI), p-value
Overall social
support
0.0008
(-0.0057, 0.0074)
p = 0.80
0.0025
(-0.0045, 0.0095)
p = 0.48
0.0037
(-0.0024, 0.0099)
p = 0.23
0.0067
(-0.0028, 0.0163)
p = 0.16

Emotional
social
support
-0.0004
(-0.0063, 0.0055)
p = 0.89
0.0005
(-0.0058, 0.0068)
p = 0.87
0.0013
(-0.0042, 0.0068)
p = 0.65
0.0034
(-0.0052, 0.0121)
p = 0.43

Instrumental
social
support
0.0014
(-0.0033, 0.0061)
p = 0.55
0.0022
(-0.0028, 0.0073)
p = 0.39
0.0044
(-0.00002,
0.0088)
p = 0.051

0.0045
(-0.0024, 0.0114)
p = 0.20
Informational
social
support
0.0000
(-0.0061, 0.0061)
p = 0.99
0.0017
(-0.0049, 0.0083)
p = 0.61
0.0027
(-0.0031, 0.0085)
p = 0.36

0.0060
(-0.0029, 0.0150)
p = 0.19
Positive
social
interaction
0.0024
(-0.0038, 0.0086)
p = 0.45
0.0034
(-0.0033, 0.0101)
p = 0.32
0.0025
(-0.0034, 0.0083)
p = 0.40

0.0092
(0.0001, 0.0182)
p = 0.047
Affectionate
support
-0.0009
(-0.0062, 0.0045)
p = 0.76

0.0012
(-0.0046, 0.0071)
p = 0.67
0.0017
(-0.0034, 0.0069)
p = 0.50
0.0020
(-0.0059, 0.0099)
p = 0.62
*Model covariates included age, marital status, use of BP medications on trial, BMI, education, income, race and treatment
assignment.

58

Table 4.5b Associations from linear mixed effects models between transformed social support scores
and cognitive outcomes without marital status, limited to cognitive-psychosocial visits completed within
6 months of each other (n = 322)
Cognitive domain
Executive
function
Verbal memory Visual memory Global cognition
Social support domain
β (95% CI), p-value
Overall social
support
0.0006
(-0.0058, 0.0070)
p = 0.85

0.0019
(-0.0049, 0.0088)
p = 0.58

0.0033
(-0.0027, 0.0094)
p = 0.28
0.0066
(-0.0028, 0.0159)
p = 0.17
Emotional
social
support
-0.0006
(-0.0064, 0.0053)
p = 0.85

0.00004
(-0.0062, 0.0063)
p = 0.99
0.0009
(-0.0046, 0.0064)
p = 0.75
0.0033
(-0.0052, 0.0118)
p = 0.44
Instrumental
social
support
0.0012
(-0.0034, 0.0057)
p = 0.61
0.0018
(-0.0031, 0.0067)
p = 0.47

0.0039
(-0.0004, 0.0083)
p = 0.07
0.0044
(-0.0023, 0.0110)
p = 0.20
Informational
social
support
-0.0002
(-0.0062, 0.0058)
p = 0.95

0.0012
(-0.0035, 0.0078)
p = 0.71
0.0024
(-0.0034, 0.0082)
p = 0.42
0.0058
(-0.0031, 0.0147)
p = 0.20
Positive
social
interaction
0.0022
(-0.0040, 0.0084)
p = 0.48

0.0031
(-0.0035, 0.0097)
p = 0.36
0.0024
(-0.0035, 0.0083)
p = 0.42
0.0090
(0.00004, 0.0180)
p = 0.0491
Affectionate
support
-0.0009
(-0.0061, 0.0043)
p = 0.73

0.0009
(-0.0048, 0.0065)
p = 0.77
0.0017
(-0.0033, 0.0067)
p = 0.51
0.0023
(-0.0054, 0.0099)
p = 0.56
*Model covariates included age, use of BP medications on trial, BMI, education, income, race and treatment assignment.
Table 4.6a summarizes results evaluating interactions between both age groups (<60 and ≥60,
<65 and ≥65) and positive social interaction for all cognitive outcomes. Borderline significant
interactions by both age groups were found between the association with positive social interaction and
visual memory (interaction p-values = 0.06 for both <60 and ≥60, <65 and ≥65). A borderline significant
interaction by age group <65 and ≥65 was also found for the association between positive social
interaction and global memory (interaction p-value = 0.08). There were no other significant interactions
by age groups.
59

Table 4.6a Summary of interaction by age groups from linear mixed effects models between
transformed positive social interaction score and cognitive outcomes, limited to cognitive-psychosocial
visits completed within 6 months of each other (n = 322)
Age group Executive
function
Verbal memory Visual memory Global cognition
Interaction β
(95% CI), p-value
<60, ≥60
years
0.0095
(-0.0026, 0.0215)
p = 0.12

0.0014
(-0.0111, 0.0140)
p = 0.82
-0.0104
(-0.0213, 0.0005)
p = 0.06
0.0033
(-0.0139, 0.0205)
p = 0.71
<65, ≥65
years
-0.0074
(-0.0218, 0.0071)
p = 0.32

-0.0078
(-0.0232, 0.0075)
p = 0.32
-0.0130
(-0.0265, 0.0004)
p = 0.06
-0.0184
(-0.0391, 0.0023)
p = 0.08
*Model covariates included age group, marital status, BP medications on trial, BMI, education, income, race, treatment
assignment, and positive social interaction score*age group. Reference groups for interaction terms: <60 and <65.

When the associations between positive social interaction and cognitive function were
examined separately by the different age groups, a significant direct association was found between
positive social interaction and global cognition among participants who were <65 years old (β = 0.0108
units per point increase in transformed positive social interaction score, 95% CI: 0.0004, 0.0213, p =
0.043). This significant association seen among participants who were <65 years old did not carry over
for the <60 year old age group (Figure 4.1). There were no other significant associations between
positive social interaction and cognitive function by age groups (Table 4.6b).

60

Table 4.6b Associations from linear mixed effects models between transformed positive social
interaction score and cognitive outcomes by age group
Executive
function
Verbal memory Visual memory Global cognition
Positive social interaction β,
95% CI, p-value
<60 years
(n = 177)
-0.0023
(-0.0108, 0.0061)
p = 0.59

0.0027
(-0.0068, 0.0121)
p = 0.58
0.0069
(-0.0012, 0.0150)
p = 0.09
0.0080
(-0.0048, 0.0209)
p = 0.22
≥60 years
(n = 125)
0.0059
(-0.0037, 0.0155)
p = 0.23

0.0026
(-0.0077, 0.0129)
p = 0.62
-0.0041
(-0.0130, 0.0048)
p = 0.36
0.0062
(-0.0073, 0.0197)
p = 0.37
<65 years
(n = 245)
0.0029
(-0.0042, 0.0100)
p = 0.43

0.0033
(-0.0040, 0.0107)
p = 0.37
0.0037
(-0.0030, 0.0103)
p = 0.28
0.0108
(0.0004, 0.0213)
p = 0.043
≥65 years
(n = 77)
-0.0023
(-0.0183, 0.0138)
p = 0.78

-0.0027
(-0.0218, 0.0164)
p = 0.78
-0.0067
(-0.0200, 0.0066)
p = 0.32
-0.0091
(-0.0281, 0.0099)
p = 0.33
*Model covariates included age, marital status, BP medications on trial, BMI, education, income, race, and treatment
assignment.

61

Figure 4.1 Associations from linear mixed effects models between transformed positive social
interaction score and global cognition by age groups

No significant interactions by ApoE4 genotype was found for any of the associations between
social support and performance on the Trails B test (all interaction p-values ≥0.17). A significant
interaction by ApoE4 genotype was found for the association between informational social support and
executive function (interaction p-value = 0.04). When this association was examined separately by
ApoE4 genotype, there was a significant inverse association for women who carried an ApoE4 allele (β =
-0.0129 units per point increase in transformed informational social support score, 95% CI: -0.0238, -
0.0021, p = 0.020) but not for women who did not (β = 0.0051 units per point increase in transformed
informational social support score, 95% CI: -0.0026, 0.0128, p = 0.19) after adjusting for age, marital
status, use of BP medications on trial, BMI, education, income, race, and treatment assignment. No
-0.04
-0.03
-0.02
-0.01
0
0.01
0.02
0.03
β (Positive Social Interaction), 95% CI
Age Group
<60 ≥60 ≥65 <65
62

significant interactions by ApoE4 genotype were found for any of the other associations between social
support and cognitive outcomes (all other interaction p-values ≥0.14).
Correlations between the number of close friends and relatives in a participant’s social network
and overall social support and the different social support dimensions were all of positive but of weak
magnitude (all correlations approximately 0.20) and strongly significant (summarized in Table 4.7a). On
the other hand, none of the correlations between the number of close friends and relatives in a
participant’s social network and cognitive function were significant (Table 4.7b).
Table 4.7a Correlations (r) between number of close friends and relatives and transformed social
support scores (n = 382)
Overall
social
support
Emotional
social
support
Instrumental
social
support
Informational
social support
Positive
social
interaction
Affectionate
support
Number of
friends
0.22
p<0.0001
0.21
p<0.0001
0.19
p<0.0001
0.21
p<0.0001
0.20
p<0.0001
0.19
p<0.0001

Table 4.7b Correlations (r) between number of close friends and relatives and cognitive outcomes (n =
382)
Verbal memory Executive
function
Visual memory Global cognition
Number of
friends
-0.09
p = 0.08
-0.08
p = 0.14
-0.01
p = 0.89
-0.08
p = 0.12

DISCUSSION
The aim of this study was to examine the association between perceived overall social support
as well as different dimensions of social support and cognitive function in a relatively healthy group of
middle-to older-aged women. Only a limited number of studies have investigated associations between
different dimensions of social support and specific cognitive domains. In this study, significant positive
associations were found between instrumental social support and visual memory as well as between
positive social interaction and global cognition. Previous research has generally found a protective
association between social support and cognitive function. Several explanations for this relationship
have been postulated, including one hypothesis that social activities may provide cognitive stimulation
63

which can contribute to individuals’ cognitive reserve (Scarmeas & Stern, 2003). Individuals with greater
cognitive reserve may be better equipped to cope with symptoms of degenerative brain changes
associated with dementia (Stern, 2002).
In this sample of healthy postmenopausal women, the average transformed social support
scores summarized in Table 4.4 are comparable to those of other cohorts. In a 2019 study of perceived
social support in acute coronary syndrome on 3,036 patients in the Alberta Provincial Project for
Outcome Assessment in Coronary Heart Disease Registry, the average transformed overall social
support score was 76.5 (22.8), which is quite similar to the average transformed overall social support
score of 76.5 (20.9) in our study (Wang et al., 2019). The 2019 study also categorized three levels of
social support: low, intermediate, and high. Participants in the intermediate and high social support
groups had average transformed overall social support scores of 63.1 (18.3) and 89.1 (11.0), respectively
(Wang et al., 2019). Based on this categorization, the ELITE women included in these analyses can be
considered to have slightly higher than intermediate levels of perceived social support.
The positive association between instrumental support and visual memory was unexpected.
Previous studies have not found instrumental support to be associated with general cognitive ability
(Gow & Mortensen, 2016; La Fleur & Salthouse, 2017). It has been postulated that this may be because
instrumental support can be viewed in a more controlling, and therefore negative, light by the recipient
(Uchino, 2009). In an attempt to clarify the association between instrumental support and visual
memory found in the ELITE sample, relationships between instrumental support and other tests with a
visual component, Symbol Digit Modalities Test, Judgement of Line Orientation, and Block Design tests
were also examined. None of these relationships were found to be significant (all p-values ≥0.14). In
contrast to our ELITE finding of a positive association, one past study found an inverse association
between total social support, as measured by the Interpersonal Support Evaluation List (ISEL), and
nonverbal memory, as measured by recall of line drawings from the Visual Reproductions I and II of
64

Wechsler Memory Scale-Revised (WMS-R) (Sims et al., 2014). This same study also reported finding
inverse associations between instrumental support and nonverbal memory, visuospatial and
visuoconstructional abilities as measured by the Judgement of Line Orientation and Block Design tests,
respectively.
On the other hand, the association between positive social interaction and global cognition was
consistent and as hypothesized. A greater availability of other persons to engage in enjoyable activities
with could help to maintain good cognitive function by buffering or reducing the negative effects of
stress, increasing cognitive reserve, or fostering healthier behaviors. Even in models without marital
status, the significant association between positive social interaction and global cognition remained. This
may mean that positive social interaction beyond those that arise from marriage are beneficial to
cognitive function.
The association between positive social interaction and global cognition differed by age group
(<65 and ≥65 years), with a stronger association confined to the younger age group. The average
transformed positive social interaction scores were comparable between women <65 years (78.7 (22.0))
and ≥65 years (76.4 (20.0)), so the more positive association seen in the younger age group is not due to
differences in social support by age group. This may mean that social support is important at a slightly
earlier point in life.
The association between informational social support and executive function differed greatly by
ApoE4 genotype. The inverse, significant association for women who carried an ApoE4 allele is double in
magnitude compared to the direct, non-significant association for women who did not. These
associations are not in the hypothesized directions.
The absence of significant associations between the number of close friends and relatives and
cognitive outcomes mean that it is unlikely that the size of the social network confounded the
association we found between positive social interaction and global cognition.
65

Strengths of this study include use of a validated social support instrument capturing different
dimensions of social support and a thorough battery of neuropsychological tests. Limitations of this
study include a lack of men in the sample due to ELITE study design, which only enrolled
postmenopausal women. In addition, because of the focus on postmenopausal women in the ELITE trial,
our study sample did not have many individuals aged 40-49 years or >80 years. As such, we were unable
to evaluate for differential associations over a large age range of young adulthood through older age.
We were also unable to evaluate the long-standing consensus that sex has differential impacts on social
support in our study. The association between social support and cognitive function may differ
depending the frequency and the sources of social support (Holtzman et al., 2004; Noguchi et al., 2019;
Zhu et al., 2012). The MOS does not capture these aspects of perceived social support, which may in
part explain why we did not find other significant associations between social support and cognitive
function. We were unable to perform longitudinal analysis to definitively rule out issues with reverse
causation. That is, that cognitive dysfunction can affect both the quantity and quality of social
relationships, and subsequently, social support (Amieva et al., 2010; Liao et al., 2018). While the ELITE
study assessed both social support and cognitive function at multiple time points, oftentimes, they were
not assessed at the same time. Participants would complete the psychosocial battery at home after
completing a scheduled ELITE visit. Only 57 participants had two time points of paired cognitive-
psychosocial data. This restricted our ability to perform longitudinal analysis.
CONCLUSION
Results from this study show associations between instrumental social support and visual
memory as well as between positive social interaction and global cognition in a population of healthy
middle- to older-aged women. Further research is needed to corroborate and extend these findings.
Providing better or more social support may potentially be an easy way to improve cognitive outcomes
and decrease dementia risk in middle- to older-aged adults.
66

CHAPTER 5
INTRODUCTION AND BACKGROUND ON STRESS AND COGNITIVE FUNCTION

Stress and cognitive function
Aside from cardiovascular risk factors and social support, stress from both chronic and life event
sources is another factor that has been associated with declines in cognitive function among older
adults. Stress is defined as the process of internal or external demands exceeding a person’s adaptive
capacity that may increase the risk of disease (Korten et al., 2017). It has long been known that stress
affects cognitive function and exacerbates age-related cognitive decline (McEwen & Sapolsky, 1995;
Sapolsky, 1996). In addition, both high levels of physiological and psychological stress have been shown
to increase the risk for developing Alzheimer’s disease (Moceri et al., 2001; Wilson et al., 2003).
Previous findings
Stressful life events and cognitive function
Much of the previous research on stress and cognitive function in older adults has focused on
stress from life events, both positive and negative, as the exposure of interest. Life events are
experiences (both positive and negative) that can affect an individual’s daily life and require
readjustment. Although older adults have been shown to have less stress compared to younger adults
(Cohen & Janicki-Deverts, 2012) possibly in part due to retirement or no longer having dependents to
care for, approximately a quarter of healthy older adults aged 57 years or more reported experiencing at
least one stressful life event within a 3-month period (Ormel et al., 2001).
A few previous studies have found an inverse association between the number of stressful life
events and cognitive function (Dickinson et al., 2011; Peters et al., 2010; VonDras et al., 2005). One
study on 428 older adults with a mean age of 73 years did not find any association between the total
number of life events and cognitive function. This may be because using estimates of stress from specific
67

life events versus from an aggregate of life events has been shown to increase the predictive validity of
stress assessment (Sands, 1981). As such, this study of older adults did find associations between
specific life events and cognitive function. For example, a direct association was found between having a
friend who was injured or ill (considered a moderate stress event by the study investigators) and better
performance on attention, psychomotor speed, and episodic memory tasks whereas an inverse
association was found between having less money to live on or being the victim of a crime (considered
high stress events by the study investigators) and performance on psychomotor speed tasks (Rosnick et
al., 2007). A more recent study on 1,356 older adults with a mean age of 73 years also found differential
associations between specific life events and cognitive function. Similar to the earlier study of older
adults, the illness of a partner or relative was associated with better cognitive function and higher stress
events such as the death of a child or grandchild was inversely associated with cognitive function
(Comijs et al., 2011). Higher stress events such as the loss of a spouse have previously been shown to be
associated with greater declines in memory among older adults (Aartsen et al., 2005).
One potential explanation for these associations is the Yerkes-Dodson law that hypothesizes an
inverted-U model of stress and arousal on cognitive performance (Yerkes & Dodson, 1908). In other
words, low or high levels of stress and arousal are associated with poorer cognitive performance
whereas moderate levels of stress are associated with better cognitive performance (Yerkes & Dodson,
1908). However, more recent research has shown that while high levels of stress do impair hippocampus
and prefrontal cortex-related functions such as memory and reasoning, mild stress tends to stimulate
cognitive function on simple tasks that do not require an abundance of cognitive resources (Sandi,
2013). It is also possible that older adults who experience the injury or illness of a friend may be more
motivated to make healthy changes to their lifestyles, such as engaging in increased physical activity,
which in turn may help to prevent cognitive decline (Anstey & Christensen, 2000).

68

Global levels of stress and cognitive function
Studies of stressful life events may not capture everyday levels of stress or sources of stress
outside of a catalogue of events. Questionnaires that evaluate the degree to which individuals perceive
recent situations in their lives as stressful, such as the Perceived Stress Scale (PSS), assess a global level
of stress (Cohen & Janicki-Deverts, 2012). Previous studies examining global levels of stress in older
adults have found inverse associations with cognitive function. One study on 98 adults in Wisconsin
found an inverse association between higher levels of global stress and episodic memory. The declines
were even more pronounced in those aged 60-80 years (VonDras et al., 2005). Another larger study on
6,207 adults aged 65 years and older in Illinois also found that higher levels of perceived stress was
associated with worse cognitive function as well as accelerated cognitive decline over an average of 6.8
years of follow-up (Aggarwal et al., 2014). A second study in Illinois on 3,159 Chinese-American adults
aged 60 years and older found that while higher levels of perceived stress was associated with poorer
performance on cognitive tests, it was not associated with faster cognitive decline over the 2 year study
period (Chen et al., 2019). The authors postulated that the follow-up time was not long enough to
capture any association with cognitive decline.
Results from other recent studies in populations outside of the United States are similar. A study
on 1,099 older adults aged 64 years and older in the Netherlands similarly found that higher levels of
perceived stress was associated with worse processing speed, direct and delayed recall, semantic
fluency, and digit span backwards (Korten et al., 2017). Another study of 1,766 German adults with a
mean age of 54.5 years reported that participants with higher levels of perceived stress performed
significantly worse on executive functioning, processing speed, verbal episodic memory, and working
memory tasks (Oumohand et al., 2020).
It is also important to note that there is variability in how individuals respond to the effects of
stress. Individual differences such as age, gender, personality traits, and previous life experiences can
69

affect how stress impacts cognitive function (Sandi, 2013). Some individuals appear to be more resilient
to the negative effects of stress on cognition. Identifying factors that play a role in resilience may be
important in future research.
Potential mechanisms
There are several biologically plausible mechanisms that may explain the link between stress
and changes in cognition in older age. The first and probably most commonly stated mechanism is that
repeated exposure to stress activates the hypothalamic-pituitary-adrenocortical (HPA) axis which in turn
triggers the release of glucocorticoids (Lupien et al., 2009). Glucocorticoids easily cross the blood-brain-
barrier and elevated levels can contribute to functional and structural changes in the brain (McEwen,
2009; Sandi, 2013). Previous research has shown that the amygdala, prefrontal cortex, and the
hippocampus undergo stress-induced structural remodeling which may be detrimental to mental
flexibility, memory and other cognitive functions, and may possibly be more pronounced in older adults
(Aggarwal et al., 2014; McEwen, 2009). In a recent MRI study of 28 adults aged 65 years or older, higher
levels of perceived stress were shown to be associated with decreased overall prefrontal cortex volume
(Moreno et al., 2017). Older adults with chronic exposure to high levels of glucocorticoids had smaller
hippocampal volumes and worse performance on hippocampus-dependent memory tasks compared to
their counterparts with normal levels of glucocorticoids (Lupien et al., 1998). Decreases in hippocampal
volume occur concurrently with the onset of memory decline (Reiman et al., 1998). Findings from animal
studies have also supported this mechanism. In a study of older macaques, treatment with a
glucocorticoid over a year was shown to increase levels of a certain type of amyloid beta peptide, the
main component of amyloid plaques found in the brains of Alzheimer’s disease patients (Kulstad et al.,
2005). On the other hand, when glucocorticoid levels are artificially kept low as they were in a study of
middle-aged rats, hippocampal changes and memory deficits correlated with brain aging were reduced
(Landfield et al., 1981).
70

Chronic stress can also cause decreased anabolic hormones such as the androgen
dehydroepiandrosterone (DHEA), growth hormone, insulin like growth factor-1 (IGF-1), and
testosterone. DHEA is important in mediating the effects of stress since it serves as an anti-
glucocorticoid (Epel, 2009). Indeed, androgens like DHEA have been shown to affect cerebral
vasculature by increasing vascular tone and reducing vascular damage from oxidative stress (Cai & Li,
2020). High levels of oxidative stress molecules, such as reactive oxygen species, in the hippocampus can
lead to apoptosis of glial cells, hippocampus dysplasia, and subsequently, cognitive decline (Kamat et al.,
2016). This is reflected in animal models where exposure to stress in the form of chronic intermittent
hypoxia in rats resulted in increased levels of oxidative stress in the hippocampus and consequent
memory impairments (Snyder et al., 2018). Like DHEA, testosterone has also been shown to reduce
damage from oxidative stress in neurons (Cai & Li, 2020). A meta-analysis of 14 randomized clinical trials
on testosterone supplementation and the prevention of cognitive decline in older men found that there
were significant improvements in executive function (Tan et al., 2019).
In addition, stress been shown to influence the rate of telomere shortening. In a previous study
of healthy women 20-50 years old, higher perceived stress as captured by the PSS was associated with
shorter telomere length (Epel et al., 2004). Previous research in animal models has shown that
psychological stress can induce oxidative stress (Wang et al., 2007), and that oxidative stress increases
the rate of telomere shortening and subsequently raises the risk for vascular dementia (von Zglinicki et
al., 2000). Indeed, patients with probable or possible vascular dementia have been found to have
significantly shorter leukocyte telomere lengths, an indicator of biological aging, compared to age-
matched patients who had normal cognitive function but had cerebrovascular or cardiovascular disease,
and to healthy patients (von Zglinicki et al., 2000).
Lastly, it is also thought that stress may affect cognitive function through psychological
processes such as anxiety, worry, rumination and related concepts (Brosschot et al., 2006). Past research
71

has shown that anxiety in older adults increases the risk for both age-related cognitive decline and for
dementia (Gulpers et al., 2016; Lenze & Butters, 2016; Sinoff & Werner, 2003). After exposure to a
stressor, it is common for people to experience worrisome thinking. Worry can represent an attempt to
address the source of a stressor and as such, may even help to mediate the negative effects of a
stressor. However, it is also possible that worry may prolong a person’s physiological response to a
stressor, which may ultimately lead to disease (Brosschot et al., 2006). Anxiety and worry have been
thought to hinder the allocation of processing resources in working memory which in turn affects
performance on a variety of cognitive tasks (VonDras et al., 2005).
Limitations of previous studies
It has been noted in population-based studies that there is a lack of consistency and care in how
stress is measured. Previous studies have used a variety of one-dimensional tools to assess stress, such
as those that measure only the psychological or physiological effects of stress (Epel et al., 2018).
However, stress is a multi-dimensional concept and it is not known how different stress measures are
associated with each other. For example, previous research has also shown that different tools such as
the Perceived Stress Scale (subjective measure) and hair cortisol concertation (physiological measure)
are not necessarily associated with each other (Oumohand et al., 2020). Another reason is that there is
wide variation in how the different stress measures are associated with cognitive outcomes (Aggarwal et
al., 2014; Korten et al., 2017; McLennan et al., 2016). In addition, some studies only use a single
measure of cognitive function such as the MMSE while others use multiple tests or an entire battery.
The use of a single test is not ideal and may limit a study’s ability to detect impairments in particular
cognitive domains.

72

CHAPTER 6
PERCEIVED STRESS AND COGNITIVE FUNCTION IN POST-MENOPAUSAL WOMEN

INTRODUCTION
Stress, defined as the process of internal or external demands exceeding a person’s adaptive
capacity (Korten et al., 2017), has been associated with declines in cognitive function among older
adults. Stress not only affects cognitive performance, it also appears to exacerbate age-related cognitive
decline (McEwen & Sapolsky, 1995; Sapolsky, 1996). In addition, high levels of stress have been shown
to increase the risk for developing Alzheimer’s disease (Moceri et al., 2001; Wilson et al., 2003).
Much of the previous research on stress and cognitive function in older adults has focused on
stress from life events, both positive and negative, as the exposure of interest. Approximately a quarter
of healthy older adults aged 57 years or more reported experiencing at least one stressful life event
within a 3-month period (Ormel et al., 2001). A few previous studies have found an inverse association
between the number of stressful life events and cognitive function (Dickinson et al., 2011; VonDras et
al., 2005). Others did not find any association when examining stress as the total number of life events
and cognitive function but did find associations between specific life events and cognitive function
(Comijs et al., 2011; Rosnick et al., 2007). In these studies, higher stress events such as being the victim
of a crime or the death of a child or grandchild were inversely associated with cognitive function
whereas more moderate stress events such as the injury or illness of a friend, partner, or relative was
found to be associated with better cognitive function.
Other previous studies have examined the association between a global level of stress and
cognitive function. Past studies have found inverse associations between higher levels of perceived
stress and cognitive function, with results being more pronounced in older adults aged 60-80 years
compared to adults <60 years (Aggarwal et al., 2014; Korten et al., 2017; VonDras et al., 2005).
73

Historically, much of the evidence for the negative association between stress and cognitive
function is taken from clinical observations, basic neuroscience studies, or randomized controlled trials
with humans rather than from population-based studies (Aggarwal et al., 2014). Findings from the few
population-based studies can be conflicting due to several reasons. One is that there are a variety of
tools used to assess stress, most being one-dimensional and only measuring an aspect of stress (i.e.,
psychological, or physiological effects) (Epel et al., 2018). It is also not known how different stress
measures (e.g., subjective and physiological measures) are associated with each other (Oumohand et al.,
2020). A third reason is that there is wide variation in how the different stress measures are associated
with cognitive outcomes (Aggarwal et al., 2014; Korten et al., 2017; McLennan et al., 2016). There is
even variation in associations between the same stress measure and different assessments of the same
cognitive domain. For example, one study reported an inverse association between the PSS and the
Logical Memory immediate and delayed recall tests of the Wechsler Memory Scale, but no association
between the PSS and the Verbal Paired Associates of the Wechsler Memory Scale (VonDras et al., 2005).
It should be noted that the Verbal Paired Associates test is unstructured and therefore a more difficult
test of verbal memory than the Logical Memory immediate and delayed recall tests. Altogether, these
issues result in a lack of consistency and thoroughness in how stress is assessed. In addition, some
studies only use a single measure of cognitive function such as the MMSE while others use multiple tests
or an entire battery.
Last but not least, most of the previous studies of the effects of stress on cognition have been
conducted in only male animals or humans. This is an issue since past research has also indicated that
males and females do indeed respond differently to stress (Lupien et al., 2009).
We examine the hypothesis that global stress and stressful life events are associated with
poorer cognitive performance in middle- to older-aged women. This study utilizes data gathered from
the Early versus Late Intervention Trial with Estradiol (ELITE) and its psychosocial substudy. Participants
74

in ELITE completed a thorough battery of standardized neuropsychological tests at baseline and at
follow-up visits. The scores from the various neuropsychological tests were combined into composite
scores for executive function, verbal memory, visual memory, and global (overall) cognition. ELITE
participants who agreed to participate in the optional psychosocial sub-study completed a series of
questionnaires that assessed psychosocial characteristics and life events every six months. The
questionnaires included the Perceived Stress Scale (PSS), a measure of global stress, and the Psychiatric
Epidemiological Research Institute (PERI) Life Events Scale, a catalogue of 110 major life events.
MATERIALS AND METHODS
Study design
Women from the ELITE clinical trial were invited to participate in the psychosocial substudy,
which was initiated while the ELITE trial follow-up was in progress. At the time the psychosocial
substudy began in January 2009, there were 512 active participants in ELITE. Of these, 448 women
(87.5%) agreed to participate in the psychosocial substudy. Women were given a packet of
questionnaires to complete at home that included assessments of perceived stress and the occurrence
and impact of stressful life events. Once completed, the participants mailed the packet of questionnaires
back to the clinic. The questionnaires assessed psychosocial characteristics and life events and were
given every six months until the substudy ended in December 2012.
Perceived stress assessment
Perceived stress was assessed using the Perceived Stress Scale (PSS), a widely used psychological
instrument for measuring the perception of stress. While the PSS was originally developed as a 14-item
scale, the shorter 10-item version was used in the ELITE psychosocial substudy. The shorter 10-item
version has been shown to provide an adequate measure of perceived stress compared to the 14-item
scale (Cohen, 1988). The PSS is a self-administered questionnaire that measures the degree to which
situations in a person’s life (in the past 2 weeks) are perceived as stressful. Responses for questions
75

about how one’s life might be perceived as uncontrollable, unpredictable, and overloading are captured
on a 5-point Likert scale that ranges from (1) “None of the time” to (5) “Very often”. An overall PSS
score is calculated by summing up the scores across all items and can range from 10-50 with higher
scores indicating higher amounts of perceived stress (Cohen, 1988). As the PSS is not a diagnostic
instrument, there are no cutoffs for levels of perceived stress. Previous studies have used quartiles or
quintiles to compare individuals in the highest category of scores (high perceived stress) to those in the
lowest category of scores (low perceived stress).
Assessment of stressful life events
The Psychiatric Epidemiological Research Institute (PERI) Life Events Scale was used to capture
the occurrence and impact of stressful life events. The PERI Life Events Scale consists of 110 major life
events (both positive and negative) in various categories: school, work, relationships/marriage, children,
family, housing, crimes/legal matters, finances, social activities, miscellaneous, and health/safety
(Dohrenwend et al., 1978). Participants are asked to indicate which of the 110 major life events they
experienced in the past 12 months. For those events that were experienced, participants were also
asked to rate the perceived undesirability of the event on a 7-point Likert scale that ranges from (1) “Not
undesirable at all” to (7) “Extremely undesirable”. A total number of stressful life events can be
calculated by summing the number of events that participants specified having experienced overall and
across the various categories.
Cognitive function assessment
A battery of cognitive and neuropsychological assessments were administered by a single trained
psychometrist in a standardized order to all ELITE participants. The battery included the following 14
assessments:
 Symbol Digit Modalities Test (SDMT)
 Trail Making Test Part B (Trails B)
76

 Judgment of Line Orientation, Form H (JLO)
 Block Design [Wechsler Adult Intelligence Scale, 3
rd
Edition (WAIS-III)]
 Letter-Number Sequencing [Wechsler Memory Scale, 3
rd
Edition (WMS-III)] (LNS)
 Category fluency (animal naming, 60 seconds) (Animals)
 Boston Naming Test, 30-item version (BNT)
 Shipley Institute of Living Scale (Shipley), Abstraction Scale
 California Verbal Learning Test, 2
nd
edition (CVLT-II), immediate recall (IR) and delayed recall
(DR)
 Logical Memory, immediate recall and delayed recall
 Faces I (IR) and II (DR) (WMS-III)
The assessments in this battery were selected to be sensitive to age-related changes in cognition
and were chosen to evaluate different cognitive functions and abilities with a focus on executive
function, verbal memory, and visual memory. Participants in ELITE were given the battery at baseline,
2.5 years, and 5 years.
Results from the 14 tests were used to generate composite scores for executive function, verbal
memory, visual memory, and global cognition. The executive function composite score was calculated as
a weighted average of the following cognitive tests: Symbol Digit Modalities Test, Trail Making Test Part
B, Shipley Abstraction Scale, Letter-Number Sequencing, and Category fluency. The verbal memory
composite score was calculated as a weighted average of the CVLT-II (immediate and delayed recall) and
Logical Memory (immediate and delayed recall) tests. The visual memory composite score was
calculated as a weighted average of the Faces I (immediate recall) and Faces II (delayed recall) tests. The
global cognition composite score was calculated as a weighted average of all 14 assessments. For each
participant, all test scores were converted to a standardized Z score at baseline, 2.5 years, and 5 years.
The Z scores were calculated using the baseline mean and standard deviation for each test. To calculate
77

the composite scores, the Z scores for the appropriate cognitive tests were summed and then weighted
by the inverse inter-test correlation matrix (Gatto et al., 2009; Henderson et al., 2016; Henderson et al.,
2012; Henderson et al., 2013).
Statistical methods
Before conducting any analyses, the psychosocial dataset was modified. For any missing items
(0.3% of all PSS items), a median score based on all available items for each participant at each visit was
imputed. Because items 7-10 were positively worded, they were reverse coded before an overall index
was calculated by summing scores from items 1-6 and the reverse coded items 7-10. Scores for each
item were also recoded from a scale of 1-5 to a scale of 0-4 to allow for direct comparison to scores in
previous studies. One participant who was included in the analyses did not complete the PSS at her
psychosocial visit 42 but did appear to complete all other assessments in the psychosocial battery at this
visit. PSS total was assessed as both a continuous variable and as a categorical variable based on
quintiles. Based on quintiles, stress level was classified as: “Low” (PSS total score: 0-5) and “High” (PSS
total score: 17-40).
For the PERI Life Events Scale, responses to events that were originally coded as 1 if participants
indicated they occurred or as missing if they did not were recoded as 1 and 0, respectively. The following
indexes were then created by summing the total number of events experienced overall and in each of
the survey categories: PERI total events, PERI total school, PERI total work, PERI total relationships, PERI
total children, PERI total family, PERI total housing, PERI total legal, PERI total finances, PERI total
miscellaneous (entering/leaving the armed services, taking a trip other than a vacation), and PERI total
health. For those events that were experienced, a two-level variable was created to indicate if events
were considered to be either not undesirable at all or any level of undesirability (responses that ranged
from (2) “A little undesirable” to (7) “Extremely undesirable”). The total number of events that were
78

rated as not undesirable and undesirable were also separately summed. PERI items 111-113 were not
included in the current analyses as they were free-response questions and could not be easily coded.
Since the psychosocial substudy was initiated sometime after the main ELITE study had begun, a
three-level variable was created to categorize the timing of visits in both the psychosocial and ELITE
datasets. In the psychosocial dataset, the three levels defined psychosocial visits that were completed at
6-18, 24-46, or 48-72 months of trial follow-up. In the modified ELITE dataset, the three levels comprised
cognitive assessment visits that were completed at baseline, 30-46, and 52-64 months. The modified
ELITE dataset contained cognitive data and other variables from the full ELITE dataset that were either
statistically significantly associated with cognitive function, or were of interest based on prior
knowledge. The psychosocial dataset was then merged with the modified ELITE dataset on participant ID
and this three-level variable. After merging, a window variable was created by taking the difference
between the psychosocial and ELITE visit months. Because participants completed the psychosocial
questionnaire packets at home, psychosocial visit dates often did not coincide with ELITE visit dates. This
window variable was used to indicate any psychosocial and ELITE visits that were completed within 6
months of each other and would be included in the analysis. The number of cognitive-psychosocial
paired visits that were completed within 6 months of each other is summarized in Table 6.1. Any
duplicate cognitive visits in the merged datasets were dropped. A total of 394 paired psychosocial-
cognitive assessments were identified.

79

Table 6.1 Cognitive and psychosocial visits completed within 6 months of each other
Psychosocial visit (month)
6 24 30 36 48 54 60 66 Total
Cognitive visit (month)
0 6 6
30 102 56 94 252
32 13 18 31
34 4 9 13
36 2 1 3
52 27 4 31
54 16 3 19
56 1 1
58 1 1
60 21 10 4 35
64 2 2
Total 6 102 75 122 43 30 12 4 394

Associations between perceived stress and cognition
Univariate linear regression was used to assess the association between PSS total and cognitive
outcomes (executive function, verbal memory, visual memory, global cognition) as well as between PSS
total and variables associated with cognitive function. Covariates that had a significant inverse
association with PSS total included age (β = -0.1713 units per point increase in total PSS score, 95% CI: -
0.2587, -0.0839), use of lipid lowering medications at baseline (β = -0.1713 units relative to participants
who did not use lipid lowering medications at baseline, 95% CI: -4.2583, -1.0598), use of lipid lowering
medications on trial (β = -1.7615 units relative to participants who did not use lipid lowering medications
on trial, 95% CI: -3.0903, -0.4327), use of blood pressure medications at baseline (β = -2.2384 units
relative to participants who did not use blood pressure medications at baseline, 95% CI: -3.5659, -
0.9110), use of blood pressure medications on trial (β = -1.3348 units relative to participants who did
not use blood pressure medications on trial, 95% CI: -2.5859, -0.0838), and income (β = -0.2391 units per
point increase in total PSS score, 95% CI: -0.4431, -0.0351). Covariates that had a significant direct
association with PSS total included race (β = 2.8081 units for Asian or Pacific Islanders relative to Non-
Hispanic Whites, 95% CI: 0.7209, 4.8954), CES-D score (β = 0.3611 units per point increase in total PSS
80

score, 95% CI: 0.2903, 0.4319), and current smoking status (β = 4.1498 units relative to never smokers,
95% CI: 0.3698, 7.9298).
Analyses were performed with linear mixed effects modeling since a number of participants had
multiple cognitive-psychosocial paired visits. The mixed effects models for PSS total included a random
intercept at the participant level, which accounted for participant level variation and the mean scores
for the cognitive outcome variables. Fixed effects included age, use of lipid lowering medications at
baseline, use of blood pressure medications at baseline, CES-D score, race, income, and current smoking
status. Because the use of these blood pressure or lipid lowering medications at baseline and on trial are
highly correlated, only the use of these medications at baseline were included in the models. Although
education level and marital status were not significantly associated with PSS total, their inclusion in the
mixed effects models was also examined based on prior information regarding associations with stress
and cognition. Product terms were used to test for interactions by age group (<65 and ≥65 years old)
and by ApoE4 genotype for the associations between PSS total and cognitive outcomes.
Associations between stressful life events and cognition
Univariate linear regression was also used to assess the associations between total number of
PERI life events and cognitive outcomes and between total number of PERI life events and variables of
interest from Project 1. Covariates that had a significant inverse association with total number of PERI
life events included age (β = -0.0803 units per PERI life event, 95% CI: -0.1242, -0.0364), use of lipid
lowering medications at baseline (β = -0.9435 units relative to participants not using lipid lowering
medications at baseline, 95% CI: -1.7500, -0.1370), use of lipid lowering medications on trial (β = -0.9973
units relative to participants not using lipid lowering medications on trial, 95% CI: -1.6613, -0.3333), use
of blood pressure medications at baseline (β = -1.1012 units relative to participants not using blood
pressure medications at baseline, 95% CI: -1.7665, -0.4360), and use of blood pressure medications on
trial (β = -0.8996 units relative to participants not using blood pressure medications on trial, 95% CI: -
81

1.5234, -0.2758). Covariates that had a significant direct association with total number of PERI life
events included marital status (β= 2.1335 units for separated relative to married women, 95% CI: -
0.2337, 4.5008; β= 0.9765 units for divorced relative to married women, 95% CI: 0.2400, 1.7130), CES-D
score (β = 0.0438 units per PERI life event, 95% CI: 0.0005, 0.0871).
As with PSS total, analyses for total number of PERI life events and the different PERI categories
were performed with mixed effects modeling since a number of participants had multiple cognitive-
psychosocial paired visits. The mixed effects models for total number of PERI life events and the
different PERI categories included a random intercept at the participant level, which accounted for
participant level variation and the mean scores for the cognitive outcome variables. Fixed effects
included age, CES-D score, use of blood pressure medications at baseline, use of lipid lowering
medications at baseline, and marital status. The use of blood pressure medications and lipid lowering
medications at both baseline and on trial were significantly associated with the total number of PERI life
events. Because the use of these medications at baseline and on trial are highly correlated, only the use
of these medications at baseline were included in the models. Although income, education, and race
were not significantly associated with the total number of PERI life events, they were included in the
mixed effects models based on prior knowledge. Because past research has found that associations
between life events and cognitive function were stronger in ApoE4 carriers compared with noncarriers
(Comijs et al., 2011), product terms were used to test for interactions between the total number of PERI
life events, the different categories of PERI life events, and ApoE4 genotype. Product terms were also
used to test for interactions between the total number of PERI life events, the different categories of
PERI life events, and age group (<65 and ≥65 years old). Univariate linear regression was also used to
assess associations between the total number of not undesirable events, the total number of
undesirable events and cognitive outcomes. Analysis for total number of not undesirable events and
82

total number of undesirable events was also performed with mixed effects modeling using the same
covariates from the total number of PERI life events models.
The correlation between the PSS and PERI responses were also examined, as well as univariate
mixed effects models with PSS as a continuous variable and as a categorical variable based on quintiles.
Table 6.2a summarizes the number of psychosocial visits completed by all 448 psychosocial
substudy participants. Restricting analyses to cognitive and psychosocial visits that were completed
within 6 months of each other reduced the number of observations in the analysis to 394 among 335
participants. Table 6.2b summarizes the number of psychosocial visits completed by the 335
participants.
Table 6.2a Summary of the number of completed psychosocial visits in the ELITE cohort
Number of
psychosocial visits
# of participants Cumulative # of
participants
1 46 46
2 47 93
3 49 142
4 75 217
5 53 270
6 75 345
7 67 412
8 36 448

Table 6.2b Summary of the number of completed psychosocial visits in the ELITE cohort when restricted
to cognitive-psychosocial visits completed within 6 months of each other
Number of
psychosocial visits
# of participants Cumulative # of
observations
1 276 276
2 59 394

RESULTS
Study sample
Baseline characteristics for the 335 ELITE women providing cognitive and psychosocial data for
the analyses are summarized in Table 6.3. As with the main ELITE cohort, the majority of participants
were non-Hispanic White (69.3%) and married (57.9%). Study participants included in the analyses had
83

an average (SD) age of 60.9 (7.1) years, and were highly educated (mean = 16.0 (1.9) years of education
correlating with a Bachelor’s degree), had an average annual income of $69,200 ($31,700), and were
overweight (mean BMI = 27.0 (5.5) kg/m
2
). Very few women reported being current smokers (3.0%), and
about half reported consuming alcohol (50.2%). Approximately one third (33.1%) reported current use
of anti-hypertensive medication and a smaller proportion (18.5%) reported current use of cholesterol
lowering medication. The average systolic blood pressure was 115.4 (12.9) mmHg, and the average
diastolic blood pressure was 74.4 (8.4) mmHg. Almost a third of participants (31.4%) were positive for
the ApoE4 genotype. The average CES-Depression score at baseline was low (7.8 out of a possible 60)
and only a small proportion (17.6%) scored 16 or above (the cut off score indicating concern for clinical
depression) (American Psychological Association, 2011).

84

Table 6.3 Baseline characteristics for ELITE women included in the cognitive-psychosocial analysis (n =
335)
Variable Mean ± SD or Number (%)
Age (years) 60.9 ± 7.1
Age ≥65 years at baseline 86 (25.7%)
Age ≥60 years at baseline 168 (50.2%)
Race
Non-Hispanic White 232 (69.3%)
Non-Hispanic Black 25 (7.5%)
Hispanic 43 (12.8%)
Asian or Pacific Islander 35 (10.5%)
Other 0 (0%)
Education (# of years) 16.0 ± 1.9
Annual income $69.2 ± $31.7
1

Marital status
Single, never married 26 (7.8%)
Married 194 (57.9%)
Separated 7 (2.1%)
Divorced 87 (26.0%)
Widowed 21 (6.3%)
Current smoker 10 (3.0%)
Alcohol use 168 (50.2%)
Use of anti-hypertensives at baseline 111 (33.1%)
Use of cholesterol lowering medication at
baseline
62 (18.5%)
BMI (kg/m
2
) 27.0 ± 5.5
1

Blood pressure (mmHg)

Systolic 115.4 ± 12.9
1

Diastolic 74.4 ± 8.4
1

ApoE4+ 104 (31.4%)
1

CES-D score 7.8 ± 7.8
CES-D score ≥ 16
2
59 (17.6%)
Executive function composite score 0.066 ± 1.361
1

Verbal memory composite score 0.085 ± 1.324
1

Visual memory composite score -0.001 ± 1.087
1

Global cognition composite score 0.078 ± 1.892
1

1. Differing sample sizes: income: n = 315, BMI: n = 334, SBP/DBP: n = 333, ApoE4+ status: n = 331, cognitive composite
scores: n = 334.
2. A score of 16 or higher on the CES-D indicates concern for clinical depression.

The average total PSS score for the 394 cognitive-psychosocial paired visits completed within 6
months of each other was 10.78 (6.32). The average total PSS score for ELITE women in the older age
group (≥65 years old) was lower (8.95 (5.65)) compared to that of women in the younger age group (<65
years old) (10.94 (6.60)).
85

The average total number of PERI life events experienced from the 394 cognitive-psychosocial
paired visits was 4.42 (3.16). Participants who reported experiencing any PERI event were more likely to
consider the life events as undesirable (mean number of undesirable events 2.50 (2.38)) than not
undesirable (mean number of not undesirable events 1.92 (1.73)). When examining undesirability by the
different PERI categories, participants were again more likely to rate most of the life events experienced
in the different PERI categories as undesirable (Table 6.4). Life events related to children or
miscellaneous were rated similarly in terms of being undesirable or not undesirable.
Table 6.4 Summary of the average number of PERI events overall and by category that were rated as not
undesirable or undesirable among women who reported an event
Not Undesirable Undesirable N (%)
Reporting Event
PERI Event Categories
School 3.08 ± 2.07 4.20 ± 2.87 50 (12.7%)
Work 2.56 ± 2.00 3.34 ± 2.64 169 (42.9%)
Relationships 2.77 ± 2.36 4.63 ± 3.34 48 (12.2%)
Children 2.82 ± 2.52 2.82 ± 2.09 11 (2.8%)
Family 2.31 ± 1.96 3.45 ± 2.59 143 (36.3%)
Housing 2.84 ± 2.13 3.81 ± 3.39 64 (16.2%)
Legal 2.41 ± 2.19 4.18 ± 3.21 61 (15.5%)
Finances 2.44 ± 1.92 3.54 ± 2.82 153 (38.8%)
Social 2.14 ± 1.77 2.71 ± 2.49 298 (75.6%)
Miscellaneous* 3.27 ± 2.06 3.31 ± 2.36 64 (16.2%)
Health 2.18 ± 1.84 3.20 ± 2.63 174 (44.2%)
Total 1.92 ± 1.73 2.50 ± 2.38 373 (94.7%)
*Entering/leaving the armed services, taking a trip other than a vacation
PSS total was weakly but significantly correlated (Pearson’s r = 0.19, p = 0.0001) with the total
number of PERI events. In univariate mixed effects models, total PSS score was significantly associated
with the total number of PERI life events (β = 0.0924 point increase per PERI life event, 95% CI: 0.0427,
0.1422, p = 0.0003). When the association was examined with PSS total as a categorical variable based
on quintiles, higher levels of stress were found to be associated with a greater number of PERI life
events experienced (Table 6.5).
86

Table 6.5 Associations from univariate mixed effects models between PSS total both as a continuous
variable and as a categorical variable based on quintiles, and the total number of PERI life events for
cognitive-psychosocial paired visits completed within 6 months of each other (n = 394)
β (95% CI) p-value
PSS total (continuous) 0.0924 (0.0427, 0.1422) 0.0003
PSS total (categorical)
1
st
quintile (PSS total: 0-5) Reference group
2
nd
quintile (PSS total: 6-8) -0.0472 (-1.0268, 0.9324) 0.92
3
rd
quintile (PSS total: 9-11) 0.3477 (-0.6076, 1.3030) 0.47
4
th
quintile (PSS total: 12-16) 1.1873 (0.2926, 2.0820) 0.009
5
th
quintile (PSS total: 17-40) 1.1198 (0.1021, 2.1375) 0.031

Associations between PSS total and cognitive function
Associations between PSS total and cognitive outcomes analyzed by linear mixed effects models
are summarized in Table 6.6. In linear mixed effects models where PSS total was analyzed as a
continuous variable, there was a significant positive association between PSS total and visual memory (β
= 0.0313 units per point increase in total PSS score, 95% CI: 0.0083, 0.0543, p = 0.008) after adjusting for
age, CES-D score, current smoking status, blood pressure medications at baseline, lipid lowering
medications at baseline, income, race, and treatment assignment. There were no significant associations
between PSS total and the other cognitive outcomes. Further adjusting the analyses for education level
and marital status did not appreciably alter the estimates of association. Thus, the results were
presented without adjustment for these additional covariates. When PSS total was analyzed as a
categorical variable by PSS quintile, significant associations were found for participants in the highest
quintile (PSS total score: 17-40) with executive function (β = -0.4838 units relative to the first quintile,
95% CI: -0.9335, -0.0340, p = 0.035) and visual memory (β = 0.6614 units relative to the first quintile,
95% CI: 0.2113, 1.1115, p = 0.004). There were no significant interactions by age group (all interaction p-
values ≥0.60) or by ApoE4 +/- genotype (all interaction p-values ≥0.45) for any of the associations
between PSS total and cognitive outcomes.

87

Table 6.6 Associations from linear mixed effects models between PSS total both as a continuous variable
and as a categorical variable based on quintiles, and cognitive outcomes for cognitive-psychosocial
paired visits completed within 6 months of each other (n = 286)
Executive
function
Verbal memory Visual memory Global cognition
PSS total
(continuous)
-0.0158
(-0.0393, 0.0077)
p = 0.19

-0.0119
(-0.0379, 0.0142)
p = 0.37
0.0313
(0.0083, 0.0543)
p = 0.008
-0.0017
(-0.0368, 0.0334)
p = 0.92
PSS total (categorical)
1
st
Quintile
(PSS total: 0-5)

Reference group
2
nd
Quintile
(PSS total: 6-8)
-0.0866
(-0.4677, 0.2946)
p = 0.65

0.0018
(-0.4349, 0.4385)
p = 0.99
0.1188
(-0.2700, 0.5077)
p = 0.55
-0.0599
(-0.6443, 0.5245)
p = 0.84
3
rd
Quintile
(PSS total: 9-11)
-0.0700
(-0.4241, 0.2841)
p = 0.70

-0.1665
(-0.5740, 0.2411)
p = 0.42
0.3396
(-0.0244, 0.7036)
p = 0.07
-0.1883
(-0.7325, 0.3559)
p = 0.50
4
th
Quintile
(PSS total: 12-16)
-0.2216
(-0.5665, 0.1233)
p = 0.21

-0.2224
(-0.6139, 0.1690)
p = 0.26
0.2339
(-0.1140, 0.5819)
p = 0.19
-0.2602
(-0.7852, 0.2647)
p = 0.33
5
th
Quintile
(PSS total: 17-40)
-0.4838
(-0.9335, -0.0340)
p = 0.035

-0.0314
(-0.5395, 0.4768)
p = 0.90
0.6614
(0.2113, 1.1115)
p = 0.004
-0.0045
(-0.6874, 0.6784)
p = 0.99
*Model covariates included age, CES-D score, current smoking status, blood pressure medications at baseline, lipid lowering
medications at baseline, income, race, and treatment assignment.

Associations between PERI life events and cognitive function
In linear mixed effects models, the total number of PERI life events was not significantly
associated with any of the cognitive outcomes (all p-values ≥ 0.18) after adjusting for age, blood
pressure medications at baseline, lipid lowering medications at baseline, CES-D score, education,
income, marital status, race, and treatment assignment (Table 6.7). Looking at the different PERI
categories, the total number of PERI legal events was marginally inversely significantly associated with
global cognition (β = -0.2844 units per PERI legal event, 95% CI: -0.5749, 0.0062, p = 0.055). None of the
PERI life events in any of the other categories was significantly associated with the cognitive outcomes
88

(Table 6.7). There was a borderline significant interaction by ApoE4 +/- genotype for the association
between the total number of housing events and verbal memory (interaction p-value = 0.09). When this
association was examined separately by ApoE4 genotype, there were no significant associations with
verbal memory, but the directions of the associations differed among women who carried an ApoE4
allele (β = -0.4379 units per PERI housing event, 95% CI: -1.0594, 0.1836, p = 0.16) compared to women
who did not (β = 0.1525 units per PERI housing event, 95% CI: -0.1568, 0.4618, p = 0.33). No other
significant interaction by ApoE4 genotype was found between cognitive associations with the total
number of PERI life events or the different categories of PERI life events (all other interaction p-values ≥
0.17). There were borderline significant and significant interactions by age group for the associations
between total number of family events and executive function (p = 0.037), between total number of
finance events and executive function (p = 0.057), and between total number of finance events and
global cognition (p = 0.07). When these associations were examined separately by age group, there were
no significant associations with executive function or global cognition and the directions of the
associations differed by age group.

89

Table 6.7 Associations from linear mixed effects models between the total number of PERI events, the
different PERI categories, and cognitive outcomes for cognitive-psychosocial paired visits completed
within 6 months of each other (n = 286)
Executive
function
Verbal memory Visual memory Global cognition
Total # of PERI
events
0.0143
(-0.0245, 0.0530)
p = 0.47
0.0286
(-0.0134, 0.0705)
p = 0.18

-0.0091
(-0.0464, 0.0282)
p = 0.63
0.0287
(-0.0280, 0.0854)
p = 0.32
Total # of school
events
0.1160
(-0.1231, 0.3550)
p = 0.34

0.1289
(-0.1280, 0.3857)
p = 0.32
-0.0547
(-0.2818, 0.1723)
p = 0.64
0.2604
(-0.086, 0.6074)
p = 0.14
Total # of work
events
0.0829
(-0.0299, 0.1957)
p = 0.15

0.0695
(-0.0535, 0.1926)
p = 0.27
0.0327
(-0.0766, 0.1420)
p = 0.56
0.1094
(-0.0562, 0.2749)
p = 0.19
Total # of
relationship
events
0.0352
(-0.2266, 0.2969)
p = 0.79
0.2663
(-0.0199, 0.5526)
p = 0.07

0.0153
(-0.2409, 0.2715)
p = 0.91
0.2064
(-0.1787, 0.5915)
p = 0.29
Total # of
children events
-0.0466
(-0.7371, 0.6439)
p = 0.89

0.3279
(-0.4529, 1.1087)
p = 0.41
-0.0276
(-0.7276, 0.6725)
p = 0.94
0.0333
(-1.0034, 1.0700)
p = 0.95
Total # of family
events
-0.0358
(-0.7371, 0.6439)
p = 0.89

0.1463
(-0.0201, 0.3127)
p = 0.08
0.1053
(-0.0437, 0.2542)
p = 0.17
0.1407
(-0.0814, 0.3629)
p = 0.21
Total # of housing
events
0.0633
(-0.1711, 0.2978)
p = 0.59

0.0479
(-0.2286, 0.3244)
p = 0.73
0.0401
(-0.2118, 0.2919)
p = 0.75
0.1197
(-0.2410, 0.4803)
p = 0.51
Total # of legal
events
-0.1553
(-0.3523, 0.0416)
p = 0.12

-0.0565
(-0.2748, 0.1617)
p = 0.61
-0.1588
(-0.3522, 0.0346)
p = 0.11
-0.2844
(-0.5749, 0.0062)
p = 0.055
Total # of finance
events
0.0923
(-0.0737, 0.2584)
p = 0.27

0.0681
(-0.1264, 0.2625)
p = 0.49
-0.1088
(-0.2848, 0.0672)
p = 0.22
0.0562
(-0.1995, 0.3118)
p = 0.66
Total # of social
events
0.0253
(-0.0896, 0.1403)
p = 0.66

0.0120
(-0.1121, 0.1362)
p = 0.85
-0.0271
(-0.1368, 0.0826)
p = 0.63
0.0721
(-0.0955, 0.2397)
p = 0.40
90

Total # of
miscellaneous
events
0.2783
(-0.0257, 0.5822)
p = 0.07

-0.0649
(-0.4209, 0.2911)
p = 0.72
0.0557
(-0.2653, 0.3768)
p = 0.73
0.1432
(-0.3261, 0.6126)
p = 0.55
Total # of health
events
-0.0822
(-0.2466, 0.0822)
p = 0.32

-0.0293
(-0.2160, 0.1574)
p = 0.76
-0.1198
(-0.2864, 0.0468)
p = 0.16
-0.1865
(-0.4328, 0.0597)
p = 0.14
*Model covariates included age, marital status, blood pressure medications at baseline, lipid lowering medications at baseline,
CES-D score, education, income, race, and treatment assignment.

In linear mixed effects models, the total number of not undesirable events was significantly
positively associated with executive function (β = 0.1053 units per not undesirable event, 95% CI:
0.0336, 0.1771, p = 0.004) and global cognition (β = 0.1309 units per not undesirable event, 95% CI:
0.0259, 0.2359, p = 0.015) after adjusting for age, blood pressure medications at baseline, lipid lowering
medications at baseline, CES-D score, education, income, marital status, race, and treatment
assignment. There were no other significant associations between the total number of not undesirable
events and verbal memory or visual memory. The total number of undesirable events was not
significantly associated with any of the cognitive outcomes (Table 6.8).
Table 6.8 Associations from linear mixed effects models between the total number of not undesirable
events, the total number of undesirable events, and cognitive outcomes for cognitive-psychosocial
paired visits completed within 6 months of each other (n = 286)
Executive
function
Verbal memory Visual memory Global cognition
Total # of not
undesirable
events
0.1053
(0.0336, 0.1771)
p = 0.004
0.0593
(-0.0189, 0.1376)
p = 0.14
0.0241
(-0.0454, 0.0936)
p = 0.49
0.1309
(0.0259, 0.2359)
p = 0.015

Total # of
undesirable
events
-0.0288
(-0.0786, 0.0210)
p = 0.26
0.0136
(-0.0412, 0.0684)
p = 0.62
-0.0284
(-0.0769, 0.0202)
p = 0.25
-0.0164
(-0.0901, 0.0573)
p = 0.66

*Model covariates included age, marital status, blood pressure medications at baseline, lipid lowering medications at baseline,
CES-D score, education, income, race, and treatment assignment.

DISCUSSION
The aim of this study was to examine the associations between perceived stress, stressful life
events, and cognitive function in a relatively healthy group of middle- to older-aged women. For the
91

most part, previous studies have found inverse associations between global stress and cognitive
function. Some have found no association, depending on the assessment of cognition used. Associations
between life events and cognitive function have also been found to vary, possibly depending on the
level of stress generated by a specific event. High levels of stress have been associated with poorer
cognitive performance while mild and moderate levels have been associated with better cognitive
performance, particularly on tasks that do not require an abundance of cognitive resources (Sandi, 2013;
Yerkes & Dodson, 1908).
The average total PSS score in this study (10.78 (6.32)) was lower than the average of previous
cohorts. For comparison, the average total PSS score for 1,032 women aged 18 years or older who
participated in the 2009 eNation Survey across the United States was 16.14 (7.56) (Cohen & Janicki-
Deverts, 2012). The average total PSS score for 2,463 German men and women with a mean age of 49.4
years surveyed in 2014 was 12.57 (6.42) (Klein et al., 2016). Stress levels have been shown to be higher
among women compared to men and decrease with increasing age (Cohen & Janicki-Deverts, 2012).
Even then, the average total PSS score for ELITE women was still lower than that of similarly aged men
and women in the 2009 eNation Survey: the age group 55-64 had an average PSS of 14.50 (7.20) and the
age group 65 and older had an average PSS of 11.09 (6.77) (Cohen & Janicki-Deverts, 2012). However,
stress levels have also been shown to increase with decreasing education level and income, and the
ELITE women were well educated (mean 16 years of education correlating with a bachelor’s degree) and
had an average annual income of $66,200.
The ELITE women reported an average of 4.42 (3.16) total PERI life events. In comparison, a
2004 study of perimenopausal women aged 44-55 years old reported an average of 9.4 total PERI life
events. Unsurprisingly, the depressed women in the same study reported a higher average of 12.3 total
PERI life events (Schmidt et al., 2004).
92

As expected, the two measures of stress used in this study were significantly correlated with
each other. Women with higher levels of global stress (in the top two quintiles of PSS score) were also
more likely to report experiencing a greater number of life events compared to women with lower levels
of global stress. The significant direct association between PSS total and visual memory was not
unexpected as a previous study found that exposure to acute psychosocial stress enhanced visuospatial
memory performance in healthy males aged 18-23 years (Human et al., 2013).
When looking at the association between categorical PSS total and cognitive function, there was
a trend with increasing levels of global stress associated with worse executive function. Women in the
highest quintile of PSS total (17-40) had significantly worse executive function compared to those in the
lowest quintile of PSS total (0-5). The trend was in the opposite direction for the association between
categorical PSS total and visual memory, where increasing levels of global stress were associated with
better visual memory composite scores. Women in the highest quintile of PSS total had significantly
better performance on the visual memory test compared to those in the first quintile of PSS total. The
findings for categorical PSS total and visual memory echo those for continuous PSS total and visual
memory.
Since stress has been shown to decrease with increasing age, we had expected that the
associations between PSS total and cognitive outcomes would differ by age groups (both <60, ≥60 and
<65, ≥65). Past research has shown that compared to younger adults (aged 18-23), older adults (aged
60-84) are better at emotional regulation and display more resilience in response to stress (Thomas et
al., 2016; Worthy et al., 2011). This finding is supported by MRI studies showing decreased reactivity to
being shown negative information in the amygdala in older adults (aged 70-90) compared to younger
adults (aged 18-29) (Mather et al., 2004). It should be noted that the gap between the age groups being
compared in these previous studies was quite wide. A much younger comparison group was not
93

available for our study and may be a reason why differential impacts of stress on cognitive function by
age were not seen here.
Aside from the borderline significant inverse association between total number of PERI legal
events and global cognition, there were no other significant associations between any of the PERI life
events and cognitive outcomes. This finding is in line with previous research. It is possible that there may
be other sources of stress which are not accounted for in the PERI Life Events Scale. In other words, it
may be that scales like the PERI Life Events that are based on specific lists of events are not able to
adequately capture chronic stress from ongoing life circumstances (Cohen, 1988).
The borderline significant inverse association between the total number of PERI legal events and
global cognition may be in part due to the level of stress associated with legal events. Legal events may
be considered a high stress event in this study population and previous research has shown that higher
stress events, such as being the victim of a crime, are inversely associated with cognitive function
(Rosnick et al., 2007). With aging increasing the risk of chronic diseases, it may be expected that PERI
health and safety events should also have an inverse association with cognitive function. However, this
was a relatively healthy group of middle- to older-aged women given the ELITE study’s exclusion criteria
for history or evidence of cardiovascular disease, diabetes mellitus, and uncontrolled hypertension
among other health conditions. Although the PERI life events questionnaire does not ask participants to
disclose the type of illness or injury experienced, it is possible that any PERI health and safety events
experienced by these relatively healthy participants may not be serious or severe enough to have a
detrimental effect on cognitive function.
When examining the association between total number of housing events and verbal memory
by ApoE4 genotype, the associations, although not statistically significant, were not only in opposite
directions, but were greater in magnitude for women who carried an ApoE4 allele compared to those
who did not. This may be suggestive of this gene playing a role in the association between stress caused
94

by housing and verbal memory, but these results should be interpreted with caution given the small
samples in stratified analysis.
The significant direct associations seen between the total number of PERI life events rated as
not undesirable and executive function and also global cognition support the possibility that these type
of events have a beneficial effect on cognitive function. For example, events such as changing jobs for a
better one, moving to a better residence or neighborhood, taking a vacation or starting a new
recreational activity might have been rated as not undesirable. Engaging in any of these activities would
have been cognitively stimulating and past research has shown that participation in similar “advanced
activities of daily living” may help to protect against cognitive decline (Sposito et al., 2015).
Previous research on psychological stress and cognitive function in older adults has focused on
either stress from life events or global levels of stress as exposures of interest but not typically both in
the same study. Strengths of this study include use of a validated perceived stress instrument capable of
capturing a global level of stress, a comprehensive catalogue of stressful life events, and a thorough
battery of neuropsychological tests. Limitations of this study include the lack of an objective physiologic
measure of stress, such as an assessment of glucocorticoid hormone cortisol levels. In addition, because
of the focus on postmenopausal women, the study sample did not include any individuals middle-aged
(e.g., 40-49 years) and younger. The lack of younger comparison groups may be a reason why we did not
find that the association between stress and cognitive function was modified by age in our study. While
the ELITE study assessed both stress and cognitive function at multiple time points, oftentimes, they
were not assessed at the same time. Participants would complete the psychosocial battery at home
after completing a scheduled ELITE visit. Only 59 subjects had two time points of paired cognitive-
psychosocial data, which restricted our ability to perform longitudinal analysis.

95

CONCLUSION
Results from this study show a significant direct association between global stress and visual
memory in a population of healthy middle- to older-aged women. This association was most apparent in
women who experienced high levels of stress. However, high levels of stress were significantly
associated with reduced performance on executive function tests. Life events that are rated as not
undesirable may have beneficial effects on executive function and global cognition. Further research,
particularly in women, is needed to corroborate these findings. Effective stress management may help
to improve cognitive outcomes in middle- to older-aged adults.
96

REFERENCES

Aartsen, M. J., Van Tilburg, T., Smits, C. H., Comijs, H. C., & Knipscheer, K. C. (2005). Does widowhood
affect memory performance of older persons? Psychol Med, 35(2), 217-226.
https://doi.org/10.1017/s0033291704002831

Aggarwal, N. T., Wilson, R. S., Beck, T. L., Rajan, K. B., Mendes de Leon, C. F., Evans, D. A., & Everson-
Rose, S. A. (2014). Perceived stress and change in cognitive function among adults 65 years and
older. Psychosom Med, 76(1), 80-85. https://doi.org/10.1097/PSY.0000000000000016

American Psychological Association. (2011). Center for Epidemiological Studies - Depression. Retrieved
October, 1 2019 from https://www.apa.org/pi/about/publications/caregivers/practice-
settings/assessment/tools/depression-scale

Amieva, H., Stoykova, R., Matharan, F., Helmer, C., Antonucci, T. C., & Dartigues, J. F. (2010). What
aspects of social network are protective for dementia? Not the quantity but the quality of social
interactions is protective up to 15 years later. Psychosom Med, 72(9), 905-911.
https://doi.org/10.1097/PSY.0b013e3181f5e121

Anstey, K., & Christensen, H. (2000). Education, activity, health, blood pressure and apolipoprotein E as
predictors of cognitive change in old age: a review. Gerontology, 46(3), 163-177.
https://doi.org/10.1159/000022153

Anstey, K. J., Lipnicki, D. M., & Low, L. F. (2008). Cholesterol as a risk factor for dementia and cognitive
decline: a systematic review of prospective studies with meta-analysis. Am J Geriatr Psychiatry,
16(5), 343-354. https://doi.org/10.1097/JGP.0b013e31816b72d4

Appelman, A. P., van der Graaf, Y., Vincken, K. L., Mali, W. P., & Geerlings, M. I. (2010). Combined effect
of cerebral hypoperfusion and white matter lesions on executive functioning - The SMART-MR
study. Dement Geriatr Cogn Disord, 29(3), 240-247. https://doi.org/10.1159/000289813

Arntzen, K. A., Schirmer, H., Johnsen, S. H., Wilsgaard, T., & Mathiesen, E. B. (2012). Carotid
Atherosclerosis Predicts Lower Cognitive Test Results: A 7-Year Follow-Up Study of 4,371 Stroke-
Free Subjects - The Tromso Study. Cerebrovascular Diseases, 33(2), 159-165.
https://doi.org/10.1159/000334182

Baumgart, M., Snyder, H. M., Carrillo, M. C., Fazio, S., Kim, H., & Johns, H. (2015). Summary of the
evidence on modifiable risk factors for cognitive decline and dementia: A population-based
perspective. Alzheimers Dement, 11(6), 718-726. https://doi.org/10.1016/j.jalz.2015.05.016

Beck, I. R., Gagneux-Zurbriggen, A., Berres, M., Taylor, K. I., & Monsch, A. U. (2012). Comparison of
verbal episodic memory measures: consortium to establish a registry for Alzheimer's disease--
Neuropsychological Assessment Battery (CERAD-NAB) versus California Verbal Learning Test
(CVLT). Arch Clin Neuropsychol, 27(5), 510-519. https://doi.org/10.1093/arclin/acs056

Berkman, L. F., Glass, T., Brissette, I., & Seeman, T. E. (2000). From social integration to health: Durkheim
in the new millennium. Soc Sci Med, 51(6), 843-857. https://doi.org/10.1016/s0277-
9536(00)00065-4
97

Bhattacharya, K., Ghosh, A., Monsivais, D., Dunbar, R. I., & Kaski, K. (2016). Sex differences in social focus
across the life cycle in humans. R Soc Open Sci, 3(4), 160097.
https://doi.org/10.1098/rsos.160097

Bo, M., Massaia, M., Speme, S., Cappa, G., Strumia, K., Cerrato, P., . . . Molaschi, M. (2005). Cognitive
function after carotid endarterectomy: greater risk of decline in symptomatic patients with left
internal carotid artery disease. J Stroke Cerebrovasc Dis, 14(5), 221-228.
https://doi.org/10.1016/j.jstrokecerebrovasdis.2005.06.003

Bo, M., Massaia, M., Speme, S., Cappa, G., Strumia, K., Cerrato, P., . . . Poli, L. (2006). Risk of cognitive
decline in older patients after carotid endarterectomy: An observational study. Journal of the
American Geriatrics Society, 54(6), 932-936. https://doi.org/10.1111/j.1532-5415.2006.00787.x

Brosschot, J. F., Gerin, W., & Thayer, J. F. (2006). The perseverative cognition hypothesis: a review of
worry, prolonged stress-related physiological activation, and health. J Psychosom Res, 60(2),
113-124. https://doi.org/10.1016/j.jpsychores.2005.06.074

Brown, S. C., Mason, C. A., Spokane, A. R., Cruza-Guet, M. C., Lopez, B., & Szapocznik, J. (2009). The
relationship of neighborhood climate to perceived social support and mental health in older
Hispanic immigrants in Miami, Florida. J Aging Health, 21(3), 431-459.
https://doi.org/10.1177/0898264308328976

Cai, Z., & Li, H. (2020). An Updated Review: Androgens and Cognitive Impairment in Older Men. Front
Endocrinol (Lausanne), 11, 586909. https://doi.org/10.3389/fendo.2020.586909

Camarda, C., Pipia, C., Azzarello, D., Battaglini, I., Romeo, G., Chiodi, M., & Camarda, R. (2018). Vascular
Risk Factors, Vascular Diseases, and Imaging Findings in a Hospital-based Cohort of Mild
Cognitive Impairment Types. Curr Alzheimer Res, 15(7), 679-690.
https://doi.org/10.2174/1567205015666180119110712

Casado Naranjo, I., Portilla Cuenca, J. C., Duque de San Juan, B., García, A. F., Sevilla, R. R., Serrano
Cabrera, A., . . . Ramírez Moreno, J. M. (2015). Association of vascular factors and amnestic mild
cognitive impairment: a comprehensive approach. J Alzheimers Dis, 44(2), 695-704.
https://doi.org/10.3233/JAD-141770

Chen, Y., Liang, Y., Zhang, W., Crawford, J. C., Sakel, K. L., & Dong, X. (2019). Perceived Stress and
Cognitive Decline in Chinese-American Older Adults. J Am Geriatr Soc, 67(S3), S519-S524.
https://doi.org/10.1111/jgs.15606

Cohen, S. (1988). Perceived stress in a probability sample of the United States. In S. Spacapan & S.
Oskamp (Eds.), The social psychology of health (pp. 31-67). Sage Publications, Inc.

Cohen, S., & Janicki-Deverts, D. (2012). Who's Stressed? Distributions of Psychological Stress in the
United States in Probability Samples from 1983, 2006, and 2009. Journal of Applied Social
Psychology, 42(6), 1320-1334. https://doi.org/10.1111/j.1559-1816.2012.00900.x

98

Cohen, S., & Wills, T. A. (1985). Stress, social support, and the buffering hypothesis. Psychol Bull, 98(2),
310-357.

Comijs, H. C., van den Kommer, T. N., Minnaar, R. W., Penninx, B. W., & Deeg, D. J. (2011). Accumulated
and differential effects of life events on cognitive decline in older persons: depending on
depression, baseline cognition, or ApoE epsilon4 status? J Gerontol B Psychol Sci Soc Sci, 66
Suppl 1, i111-120. https://doi.org/10.1093/geronb/gbr019

DeCarli, C., Miller, B. L., Swan, G. E., Reed, T., Wolf, P. A., & Carmelli, D. (2001). Cerebrovascular and
brain morphologic correlates of mild cognitive impairment in the National Heart, Lung, and
Blood Institute Twin Study. Arch Neurol, 58(4), 643-647.
https://doi.org/10.1001/archneur.58.4.643

Diamond, A. (2013). Executive functions. Annu Rev Psychol, 64, 135-168.
https://doi.org/10.1146/annurev-psych-113011-143750

Dickinson, W. J., Potter, G. G., Hybels, C. F., McQuoid, D. R., & Steffens, D. C. (2011). Change in stress
and social support as predictors of cognitive decline in older adults with and without depression.
Int J Geriatr Psychiatry, 26(12), 1267-1274. https://doi.org/10.1002/gps.2676

Dohrenwend, B. S., Krasnoff, L., Askenasy, A. R., & Dohrenwend, B. P. (1978). Exemplification of a
method for scaling life events: the Peri Life Events Scale. J Health Soc Behav, 19(2), 205-229.

Epel, E. S. (2009). Psychological and metabolic stress: a recipe for accelerated cellular aging? Hormones
(Athens), 8(1), 7-22. https://doi.org/10.14310/horm.2002.1217

Epel, E. S., Blackburn, E. H., Lin, J., Dhabhar, F. S., Adler, N. E., Morrow, J. D., & Cawthon, R. M. (2004).
Accelerated telomere shortening in response to life stress. Proc Natl Acad Sci U S A, 101(49),
17312-17315. https://doi.org/10.1073/pnas.0407162101

Epel, E. S., Crosswell, A. D., Mayer, S. E., Prather, A. A., Slavich, G. M., Puterman, E., & Mendes, W. B.
(2018). More than a feeling: A unified view of stress measurement for population science. Front
Neuroendocrinol, 49, 146-169. https://doi.org/10.1016/j.yfrne.2018.03.001

Fanning, J. P., Wong, A. A., & Fraser, J. F. (2014). The epidemiology of silent brain infarction: a systematic
review of population-based cohorts. Bmc Medicine, 12, Article 119.
https://doi.org/10.1186/s12916-014-0119-0

Farias, S. T., Mungas, D., Reed, B. R., Harvey, D., & DeCarli, C. (2009). Progression of mild cognitive
impairment to dementia in clinic- vs community-based cohorts. Arch Neurol, 66(9), 1151-1157.
https://doi.org/10.1001/archneurol.2009.106

Feinkohl, I., Keller, M., Robertson, C. M., Morling, J. R., Williamson, R. M., Nee, L. D., . . . Edinburgh Type
2 Diabet Study, E. D. (2013). Clinical and Subclinical Macrovascular Disease as Predictors of
Cognitive Decline in Older Patients With Type 2 Diabetes: The Edinburgh Type 2 Diabetes Study.
Diabetes Care, 36(9), 2779-2786. https://doi.org/10.2337/dc12-2241

99

Gallagher, D., Visser, M., Sepúlveda, D., Pierson, R. N., Harris, T., & Heymsfield, S. B. (1996). How useful
is body mass index for comparison of body fatness across age, sex, and ethnic groups? Am J
Epidemiol, 143(3), 228-239. https://doi.org/10.1093/oxfordjournals.aje.a008733

Gardener, H., Caunca, M. R., Dong, C. H., Cheung, Y. K., Elkind, M. S. V., Sacco, R. L., . . . Wright, C. B.
(2017). Ultrasound Markers of Carotid Atherosclerosis and Cognition The Northern Manhattan
Study. Stroke, 48(7), 1855-1861. https://doi.org/10.1161/strokeaha.117.016921

Gatto, N. M., Henderson, V. W., John, J. A. S., McCleary, C., Detrano, R., Hodis, H. N., & Mack, W. J.
(2009). Subclinical atherosclerosis is weakly associated with lower cognitive function in healthy
hyperhomocysteinemic adults without clinical cardiovascular disease. International Journal of
Geriatric Psychiatry, 24(4), 390-399. https://doi.org/10.1002/gps.2134

Gaugler, J., James, B., Johnson, T., Reimer, J., Solis, M., Weuve, J., . . . Hohman, T. J. (2022). 2022
Alzheimer's Disease Facts and Figures (2022 Alzheimer's Disease Facts and Figures, Issue.
https://www.alz.org/media/Documents/alzheimers-facts-and-figures.pdf

Gow, A. J., Corley, J., Starr, J. M., & Deary, I. J. (2013). Which social network or support factors are
associated with cognitive abilities in old age? Gerontology, 59(5), 454-463.
https://doi.org/10.1159/000351265

Gow, A. J., & Mortensen, E. L. (2016). Social resources and cognitive ageing across 30 years: the Glostrup
1914 Cohort. Age Ageing, 45(4), 480-486. https://doi.org/10.1093/ageing/afw070

Gulpers, B., Ramakers, I., Hamel, R., Köhler, S., Oude Voshaar, R., & Verhey, F. (2016). Anxiety as a
Predictor for Cognitive Decline and Dementia: A Systematic Review and Meta-Analysis. Am J
Geriatr Psychiatry, 24(10), 823-842. https://doi.org/10.1016/j.jagp.2016.05.015

Henderson, V. W., John, J. A. S., Hodis, H. N., McCleary, C. A., Stanczyk, F. Z., Shoupe, D., . . . Mack, W. J.
(2016). Cognitive effects of estradiol after menopause A randomized trial of the timing
hypothesis. Neurology, 87(7), 699-708. https://doi.org/10.1212/wnl.0000000000002980

Henderson, V. W., St John, J. A., Hodis, H. N., Kono, N., McCleary, C. A., Franke, A. A., . . . Grp, W. R.
(2012). Long-term soy isoflavone supplementation and cognition in women A randomized,
controlled trial. Neurology, 78(23), 1841-1848.
https://doi.org/10.1212/WNL.0b013e318258f822

Henderson, V. W., St John, J. A., Hodis, H. N., McCleary, C. A., Stanczyk, F. Z., Karim, R., . . . Mack, W. J.
(2013). Cognition, mood, and physiological concentrations of sex hormones in the early and late
postmenopause. Proceedings of the National Academy of Sciences of the United States of
America, 110(50), 20290-20295. https://doi.org/10.1073/pnas.1312353110

Hodis, H. N., Mack, W. J., Dustin, L., Mahrer, P. R., Azen, S. P., Detrano, R., . . . Group, B. R. (2009). High-
dose B vitamin supplementation and progression of subclinical atherosclerosis: a randomized
controlled trial. Stroke, 40(3), 730-736. https://doi.org/10.1161/STROKEAHA.108.526798

100

Hodis, H. N., Mack, W. J., Henderson, V. W., Shoupe, D., Budoff, M. J., Hwang-Levine, J., . . . Group, E. R.
(2016). Vascular Effects of Early versus Late Postmenopausal Treatment with Estradiol. N Engl J
Med, 374(13), 1221-1231. https://doi.org/10.1056/NEJMoa1505241

Hodis, H. N., Mack, W. J., Kono, N., Azen, S. P., Shoupe, D., Hwang-Levine, J., . . . Group, W. s. I. S. H. R.
(2011). Isoflavone soy protein supplementation and atherosclerosis progression in healthy
postmenopausal women: a randomized controlled trial. Stroke, 42(11), 3168-3175.
https://doi.org/10.1161/STROKEAHA.111.620831

Hodis, H. N., Mack, W. J., Lobo, R. A., Shoupe, D., Sevanian, A., Mahrer, P. R., . . . Azen, S. P. (2001).
Estrogen in the prevention of atherosclerosis - A randomized, double-blind, placebo-controlled
trial. Annals of Internal Medicine, 135(11), 939-953. https://doi.org/10.7326/0003-4819-135-11-
200112040-00005

Holtzman, D. M., Herz, J., & Bu, G. (2012). Apolipoprotein E and apolipoprotein E receptors: normal
biology and roles in Alzheimer disease. Cold Spring Harb Perspect Med, 2(3), a006312.
https://doi.org/10.1101/cshperspect.a006312

Holtzman, R. E., Rebok, G. W., Saczynski, J. S., Kouzis, A. C., Wilcox Doyle, K., & Eaton, W. W. (2004).
Social network characteristics and cognition in middle-aged and older adults. J Gerontol B
Psychol Sci Soc Sci, 59(6), P278-284. https://doi.org/10.1093/geronb/59.6.p278

Honig, L. S., Kukull, W., & Mayeux, R. (2005). Atherosclerosis and AD: analysis of data from the US
National Alzheimer's Coordinating Center. Neurology, 64(3), 494-500.
https://doi.org/10.1212/01.WNL.0000150886.50187.30

Human, R., Thomas, K. G. F., Dreyer, A., Amod, A. R., Wolf, P. S. A., & Jacobs, W. J. (2013). Acute
psychosocial stress enhances visuospatial memory in healthy males. South African Journal of
Psychology, 43(3), 300-313. https://doi.org/10.1177/0081246313496913

Johnston, S. C., O'Meara, E. S., Manolio, T. A., Lefkowitz, D., O'Leary, D. H., Goldstein, S., . . . Longstreth,
W. T. (2004). Cognitive impairment and decline are associated with carotid artery disease in
patients without clinically evident cerebrovascular disease. Ann Intern Med, 140(4), 237-247.
https://doi.org/10.7326/0003-4819-140-4-200402170-00005

Kamat, P. K., Kalani, A., Rai, S., Swarnkar, S., Tota, S., Nath, C., & Tyagi, N. (2016). Mechanism of
Oxidative Stress and Synapse Dysfunction in the Pathogenesis of Alzheimer's Disease:
Understanding the Therapeutics Strategies. Mol Neurobiol, 53(1), 648-661.
https://doi.org/10.1007/s12035-014-9053-6

Kats, D., Patel, M. D., Palta, P., Meyer, M. L., Gross, A. L., Whitsel, E. A., . . . Heiss, G. (2016). Social
support and cognition in a community-based cohort: the Atherosclerosis Risk in Communities
(ARIC) study. Age Ageing, 45(4), 475-480. https://doi.org/10.1093/ageing/afw060

Kelly, M. E., Duff, H., Kelly, S., McHugh Power, J. E., Brennan, S., Lawlor, B. A., & Loughrey, D. G. (2017).
The impact of social activities, social networks, social support and social relationships on the
cognitive functioning of healthy older adults: a systematic review. Syst Rev, 6(1), 259.
https://doi.org/10.1186/s13643-017-0632-2
101

Khazaee-Pool, M., Bahrami, M., Luque, J. S., Pashaei, T., Taymoori, P., & Roshani, D. (2018). Validation of
the Farsi version of the medical outcomes study-social support survey for mammography. BMC
Public Health, 18(1), 1280. https://doi.org/10.1186/s12889-018-6174-2

Khuong, L. Q., Vu, T. T., Huynh, V. N., & Thai, T. T. (2018). Psychometric properties of the medical
outcomes study: social support survey among methadone maintenance patients in Ho Chi Minh
City, Vietnam: a validation study. Subst Abuse Treat Prev Policy, 13(1), 8.
https://doi.org/10.1186/s13011-018-0147-4

Klein, E. M., Brähler, E., Dreier, M., Reinecke, L., Müller, K. W., Schmutzer, G., . . . Beutel, M. E. (2016).
The German version of the Perceived Stress Scale - psychometric characteristics in a
representative German community sample. BMC Psychiatry, 16, 159.
https://doi.org/10.1186/s12888-016-0875-9

Korten, N. C., Comijs, H. C., Penninx, B. W., & Deeg, D. J. (2017). Perceived stress and cognitive function
in older adults: which aspect of perceived stress is important? Int J Geriatr Psychiatry, 32(4),
439-445. https://doi.org/10.1002/gps.4486

Kuiper, J. S., Zuidersma, M., Oude Voshaar, R. C., Zuidema, S. U., van den Heuvel, E. R., Stolk, R. P., &
Smidt, N. (2015). Social relationships and risk of dementia: A systematic review and meta-
analysis of longitudinal cohort studies. Ageing Res Rev, 22, 39-57.
https://doi.org/10.1016/j.arr.2015.04.006

Kuiper, J. S., Zuidersma, M., Zuidema, S. U., Burgerhof, J. G., Stolk, R. P., Oude Voshaar, R. C., & Smidt, N.
(2016). Social relationships and cognitive decline: a systematic review and meta-analysis of
longitudinal cohort studies. Int J Epidemiol, 45(4), 1169-1206.
https://doi.org/10.1093/ije/dyw089

Kulstad, J. J., McMillan, P. J., Leverenz, J. B., Cook, D. G., Green, P. S., Peskind, E. R., . . . Craft, S. (2005).
Effects of chronic glucocorticoid administration on insulin-degrading enzyme and amyloid-beta
peptide in the aged macaque. J Neuropathol Exp Neurol, 64(2), 139-146.
https://doi.org/10.1093/jnen/64.2.139

Kumari, M., & Marmot, M. (2005). Diabetes and cognitive function in a middle-aged cohort: findings
from the Whitehall II study. Neurology, 65(10), 1597-1603.
https://doi.org/10.1212/01.wnl.0000184521.80820.e4

La Fleur, C. G., & Salthouse, T. A. (2017). Which Aspects of Social Support Are Associated With Which
Cognitive Abilities for Which People? J Gerontol B Psychol Sci Soc Sci, 72(6), 1006-1016.
https://doi.org/10.1093/geronb/gbv119

Landfield, P. W., Baskin, R. K., & Pitler, T. A. (1981). Brain aging correlates: retardation by hormonal-
pharmacological treatments. Science, 214(4520), 581-584.
https://doi.org/10.1126/science.6270791

102

Langa, K. M., Larson, E. B., Crimmins, E. M., Faul, J. D., Levine, D. A., Kabeto, M. U., & Weir, D. R. (2017).
A Comparison of the Prevalence of Dementia in the United States in 2000 and 2012. JAMA Intern
Med, 177(1), 51-58. https://doi.org/10.1001/jamainternmed.2016.6807

Larson, E. B., & Langa, K. M. (2017). What's the "Take Home" from Research on Dementia Trends? Plos
Medicine, 14(3), Article e1002236. https://doi.org/10.1371/journal.pmed.1002236

Lenze, E. J., & Butters, M. A. (2016). Consequences of Anxiety in Aging and Cognitive Decline. Am J
Geriatr Psychiatry, 24(10), 843-845. https://doi.org/10.1016/j.jagp.2016.07.006

Liao, J., Muniz-Terrera, G., Head, J., & Brunner, E. J. (2018). Dynamic Longitudinal Associations Between
Social Support and Cognitive Function: A Prospective Investigation of the Directionality of
Associations. J Gerontol B Psychol Sci Soc Sci, 73(7), 1233-1243.
https://doi.org/10.1093/geronb/gbw135

Lim, S. L., Gao, Q., Nyunt, M. S. Z., Gong, L. L., Lunaria, J. B., Lim, M. L., . . . Ng, T. P. (2016). Vascular
Health Indices and Cognitive Domain Function: Singapore Longitudinal Ageing Studies. Journal of
Alzheimers Disease, 50(1), 27-40. https://doi.org/10.3233/jad-150516

Lobo, A., Launer, L. J., Fratiglioni, L., Andersen, K., Di Carlo, A., Breteler, M. M., . . . Hofman, A. (2000).
Prevalence of dementia and major subtypes in Europe: A collaborative study of population-
based cohorts. Neurologic Diseases in the Elderly Research Group. Neurology, 54(11 Suppl 5),
S4-9.

Loy, C. T., Schofield, P. R., Turner, A. M., & Kwok, J. B. (2014). Genetics of dementia. Lancet, 383(9919),
828-840. https://doi.org/10.1016/S0140-6736(13)60630-3

Lupien, S. J., de Leon, M., de Santi, S., Convit, A., Tarshish, C., Nair, N. P., . . . Meaney, M. J. (1998).
Cortisol levels during human aging predict hippocampal atrophy and memory deficits. Nat
Neurosci, 1(1), 69-73. https://doi.org/10.1038/271

Lupien, S. J., McEwen, B. S., Gunnar, M. R., & Heim, C. (2009). Effects of stress throughout the lifespan
on the brain, behaviour and cognition. Nat Rev Neurosci, 10(6), 434-445.
https://doi.org/10.1038/nrn2639

Mack, W. J., LaBree, L., Liu, C., Selzer, R. H., & Hodis, H. N. (2000). Correlations between measures of
atherosclerosis change using carotid ultrasonography and coronary angiography.
Atherosclerosis, 150(2), 371-379. https://doi.org/10.1016/s0021-9150(99)00383-4

Manly, J. J., Tang, M. X., Schupf, N., Stern, Y., Vonsattel, J. P., & Mayeux, R. (2008). Frequency and course
of mild cognitive impairment in a multiethnic community. Ann Neurol, 63(4), 494-506.
https://doi.org/10.1002/ana.21326

Mather, M., Canli, T., English, T., Whitfield, S., Wais, P., Ochsner, K., . . . Carstensen, L. L. (2004).
Amygdala responses to emotionally valenced stimuli in older and younger adults. Psychol Sci,
15(4), 259-263. https://doi.org/10.1111/j.0956-7976.2004.00662.x

103

Matsumoto, L., Suzuki, K., Mizuno, Y., Ohike, Y., Ozeki, A., Ono, S., . . . Iwata, A. (2018). Association of
subclinical carotid atherosclerosis with immediate memory and other cognitive functions.
Geriatrics & Gerontology International, 18(1), 65-71. https://doi.org/10.1111/ggi.13142

McEwen, B. S. (2009). The brain is the central organ of stress and adaptation. Neuroimage, 47(3), 911-
913. https://doi.org/10.1016/j.neuroimage.2009.05.071

McEwen, B. S., & Sapolsky, R. M. (1995). Stress and cognitive function. Curr Opin Neurobiol, 5(2), 205-
216. https://doi.org/10.1016/0959-4388(95)80028-x

McLennan, S. N., Ihle, A., Steudte-Schmiedgen, S., Kirschbaum, C., & Kliegel, M. (2016). Hair cortisol and
cognitive performance in working age adults. Psychoneuroendocrinology, 67, 100-103.
https://doi.org/10.1016/j.psyneuen.2016.01.029

MedlinePlus. (2021). APOE gene. National Library of Medicine. Retrieved August 1 from
https://medlineplus.gov/genetics/gene/apoe/#conditions

Michaelson, D. M. (2014). APOE ε4: the most prevalent yet understudied risk factor for Alzheimer's
disease. Alzheimers Dement, 10(6), 861-868. https://doi.org/10.1016/j.jalz.2014.06.015

Michaud, T. L., Siahpush, M., Farazi, P. A., Kim, J., Yu, F., Su, D., & Murman, D. L. (2018). The Association
Between Body Mass Index, and Cognitive, Functional, and Behavioral Declines for Incident
Dementia. J Alzheimers Dis, 66(4), 1507-1517. https://doi.org/10.3233/JAD-180278

Mitchell, A. J., & Shiri-Feshki, M. (2009). Rate of progression of mild cognitive impairment to dementia--
meta-analysis of 41 robust inception cohort studies. Acta Psychiatr Scand, 119(4), 252-265.
https://doi.org/10.1111/j.1600-0447.2008.01326.x

Moceri, V. M., Kukull, W. A., Emanual, I., van Belle, G., Starr, J. R., Schellenberg, G. D., . . . Larson, E. B.
(2001). Using census data and birth certificates to reconstruct the early-life socioeconomic
environment and the relation to the development of Alzheimer's disease. Epidemiology, 12(4),
383-389. https://doi.org/10.1097/00001648-200107000-00007

Moon, J. H., Lim, S., Han, J. W., Kim, K. M., Choi, S. H., Park, K. S., . . . Jang, H. C. (2015). Carotid intima-
media thickness is associated with the progression of cognitive impairment in older adults.
Stroke, 46(4), 1024-1030. https://doi.org/10.1161/STROKEAHA.114.008170

Moreno, G. L., Bruss, J., & Denburg, N. L. (2017). Increased perceived stress is related to decreased
prefrontal cortex volumes among older adults. J Clin Exp Neuropsychol, 39(4), 313-325.
https://doi.org/10.1080/13803395.2016.1225006

Moser, A., Stuck, A. E., Silliman, R. A., Ganz, P. A., & Clough-Gorr, K. M. (2012). The eight-item modified
Medical Outcomes Study Social Support Survey: psychometric evaluation showed excellent
performance. J Clin Epidemiol, 65(10), 1107-1116. https://doi.org/10.1016/j.jclinepi.2012.04.007

Nikzad-Terhune, K. A., Anderson, K. A., Newcomer, R., & Gaugler, J. E. (2010). Do trajectories of at-home
dementia caregiving account for burden after nursing home placement? A growth curve
analysis. Soc Work Health Care, 49(8), 734-752. https://doi.org/10.1080/00981381003635296
104

Noguchi, T., Nojima, I., Inoue-Hirakawa, T., & Sugiura, H. (2019). The Association between Social Support
Sources and Cognitive Function among Community-Dwelling Older Adults: A One-Year
Prospective Study. Int J Environ Res Public Health, 16(21).
https://doi.org/10.3390/ijerph16214228

Ormel, J., Oldehinkel, A. J., & Brilman, E. I. (2001). The interplay and etiological continuity of
neuroticism, difficulties, and life events in the etiology of major and subsyndromal, first and
recurrent depressive episodes in later life. Am J Psychiatry, 158(6), 885-891.
https://doi.org/10.1176/appi.ajp.158.6.885

Oumohand, S. E., Ward, D. D., Boenniger, M. M., Merten, N., Kirschbaum, C., & Breteler, M. M. B.
(2020). Perceived stress but not hair cortisol concentration is related to adult cognitive
performance. Psychoneuroendocrinology, 121, 104810.
https://doi.org/10.1016/j.psyneuen.2020.104810

Park, J., Park, J. H., & Park, H. (2019). Association Between Carotid Artery Intima-Media Thickness and
Combinations of Mild Cognitive Impairment and Pre-Frailty in Older Adults. Int J Environ Res
Public Health, 16(16). https://doi.org/10.3390/ijerph16162978

Peters, J. L., Weisskopf, M. G., Spiro, A., Schwartz, J., Sparrow, D., Nie, H., . . . Wright, R. J. (2010).
Interaction of stress, lead burden, and age on cognition in older men: the VA Normative Aging
Study. Environ Health Perspect, 118(4), 505-510. https://doi.org/10.1289/ehp.0901115

Petersen, R. C., Caracciolo, B., Brayne, C., Gauthier, S., Jelic, V., & Fratiglioni, L. (2014). Mild cognitive
impairment: a concept in evolution. J Intern Med, 275(3), 214-228.
https://doi.org/10.1111/joim.12190

Petsko, G. A. (2006). The next epidemic. Genome Biol, 7(5), 108. https://doi.org/10.1186/gb-2006-7-5-
108

Priede, A., Andreu, Y., Martínez, P., Conchado, A., Ruiz-Torres, M., & González-Blanch, C. (2018). The
factor structure of the Medical Outcomes Study-Social Support Survey: A comparison of
different models in a sample of recently diagnosed cancer patients. J Psychosom Res, 108, 32-38.
https://doi.org/10.1016/j.jpsychores.2018.02.008

Rabin, L. A., Paré, N., Saykin, A. J., Brown, M. J., Wishart, H. A., Flashman, L. A., & Santulli, R. B. (2009).
Differential memory test sensitivity for diagnosing amnestic mild cognitive impairment and
predicting conversion to Alzheimer's disease. Neuropsychol Dev Cogn B Aging Neuropsychol
Cogn, 16(3), 357-376. https://doi.org/10.1080/13825580902825220

Reiman, E. M., Uecker, A., Caselli, R. J., Lewis, S., Bandy, D., de Leon, M. J., . . . Thibodeau, S. N. (1998).
Hippocampal volumes in cognitively normal persons at genetic risk for Alzheimer's disease. Ann
Neurol, 44(2), 288-291. https://doi.org/10.1002/ana.410440226

Reitz, C., Luchsinger, J. A., Tang, M. X., Manly, J., & Mayeux, R. (2006). Stroke and memory performance
in elderly persons without dementia. Archives of Neurology, 63(4), 571-576.
https://doi.org/10.1001/archneur.63.4.571
105

Richardson, T. J., Lee, S. J., Berg-Weger, M., & Grossberg, G. T. (2013). Caregiver health: health of
caregivers of Alzheimer's and other dementia patients. Curr Psychiatry Rep, 15(7), 367.
https://doi.org/10.1007/s11920-013-0367-2

Roberts, R., & Knopman, D. S. (2013). Classification and epidemiology of MCI. Clin Geriatr Med, 29(4),
753-772. https://doi.org/10.1016/j.cger.2013.07.003

Roberts, R. O., Geda, Y. E., Knopman, D. S., Cha, R. H., Pankratz, V. S., Boeve, B. F., . . . Petersen, R. C.
(2012). The incidence of MCI differs by subtype and is higher in men: the Mayo Clinic Study of
Aging. Neurology, 78(5), 342-351. https://doi.org/10.1212/WNL.0b013e3182452862

Roher, A. E., Esh, C., Kokjohn, T. A., Kalback, W., Luehrs, D. C., Seward, J. D., . . . Beach, T. G. (2003).
Circle of willis atherosclerosis is a risk factor for sporadic Alzheimer's disease. Arterioscler
Thromb Vasc Biol, 23(11), 2055-2062. https://doi.org/10.1161/01.ATV.0000095973.42032.44

Romero, J. R., Beiser, A., Seshadri, S., Benjamin, E. J., Polak, J. F., Vasan, R. S., . . . Wolf, P. A. (2009).
Carotid Artery Atherosclerosis, MRI Indices of Brain Ischemia, Aging, and Cognitive Impairment
The Framingham Study. Stroke, 40(5), 1590-1596.
https://doi.org/10.1161/strokeaha.108.535245

Rosnick, C. B., Small, B. J., McEvoy, C. L., Borenstein, A. R., & Mortimer, J. A. (2007). Negative life events
and cognitive performance in a population of older adults. J Aging Health, 19(4), 612-629.
https://doi.org/10.1177/0898264307300975

Ruitenberg, A., den Heijer, T., Bakker, S. L. M., van Swieten, J. C., Koudstaal, P. J., Hoftnan, A., & Breteler,
M. M. B. (2005). Cerebral hypoperfusion and clinical onset of dementia: The Rotterdam study.
Annals of Neurology, 57(6), 789-794. https://doi.org/10.1002/ana.20493

Sachdev, P. S., Blacker, D., Blazer, D. G., Ganguli, M., Jeste, D. V., Paulsen, J. S., & Petersen, R. C. (2014).
Classifying neurocognitive disorders: the DSM-5 approach. Nat Rev Neurol, 10(11), 634-642.
https://doi.org/10.1038/nrneurol.2014.181

Sachdev, P. S., Lipnicki, D. M., Kochan, N. A., Crawford, J. D., Thalamuthu, A., Andrews, G., . . . (COSMIC),
C. S. o. M. i. a. I. C. (2015). The Prevalence of Mild Cognitive Impairment in Diverse Geographical
and Ethnocultural Regions: The COSMIC Collaboration. PLoS One, 10(11), e0142388.
https://doi.org/10.1371/journal.pone.0142388

Sandi, C. (2013). Stress and cognition. Wiley Interdiscip Rev Cogn Sci, 4(3), 245-261.
https://doi.org/10.1002/wcs.1222

Sands, J. D. (1981). The relationship of stressful life events to intellectual functioning in women over 65.
Int J Aging Hum Dev, 14(1), 11-22. https://doi.org/10.2190/7rdc-2ruk-0glr-j2kb

Sapolsky, R. M. (1996). Why stress is bad for your brain. Science, 273(5276), 749-750.
https://doi.org/10.1126/science.273.5276.749

106

Scarmeas, N., & Stern, Y. (2003). Cognitive reserve and lifestyle. J Clin Exp Neuropsychol, 25(5), 625-633.
https://doi.org/10.1076/jcen.25.5.625.14576

Schmidt, P. J., Murphy, J. H., Haq, N., Rubinow, D. R., & Danaceau, M. A. (2004). Stressful life events,
personal losses, and perimenopause-related depression. Arch Womens Ment Health, 7(1), 19-
26. https://doi.org/10.1007/s00737-003-0036-2

Schneider, J. A., Wilson, R. S., Cochran, E. J., Bienias, J. L., Arnold, S. E., Evans, D. A., & Bennett, D. A.
(2003). Relation of cerebral infarctions to dementia and cognitive function in older persons.
Neurology, 60(7), 1082-1088. https://doi.org/10.1212/01.wnl.0000055863.87435.b2

Seeman, T. E., Lusignolo, T. M., Albert, M., & Berkman, L. (2001). Social relationships, social support, and
patterns of cognitive aging in healthy, high-functioning older adults: MacArthur studies of
successful aging. Health Psychol, 20(4), 243-255. https://doi.org/10.1037//0278-6133.20.4.243

Selzer, R. H., Hodis, H. N., Kwongfu, H., Mack, W. J., Lee, P. L., Liu, C. R., & Liu, C. H. (1994). EVALUATION
OF COMPUTERIZED EDGE TRACKING FOR QUANTIFYING INTIMA-MEDIA THICKNESS OF THE
COMMON CAROTID-ARTERY FROM B-MODE ULTRASOUND IMAGES. Atherosclerosis, 111(1), 1-
11. https://doi.org/10.1016/0021-9150(94)90186-4

Selzer, R. H., Mack, W. J., Lee, P. L., Kwong-Fu, H., & Hodis, H. N. (2001). Improved common carotid
elasticity and intima-media thickness measurements from computer analysis of sequential
ultrasound frames. Atherosclerosis, 154(1), 185-193. https://doi.org/10.1016/s0021-
9150(00)00461-5

Shankar, A., Hamer, M., McMunn, A., & Steptoe, A. (2013). Social isolation and loneliness: relationships
with cognitive function during 4 years of follow-up in the English Longitudinal Study of Ageing.
Psychosom Med, 75(2), 161-170. https://doi.org/10.1097/PSY.0b013e31827f09cd
Sherbourne, C. D., & Stewart, A. L. (1991). The MOS social support survey. Soc Sci Med, 32(6), 705-714.
https://doi.org/10.1016/0277-9536(91)90150-b

Sillesen, H. (2014). Carotid intima-media thickness and/or carotid plaque: what is relevant? Eur J Vasc
Endovasc Surg, 48(2), 115-117. https://doi.org/10.1016/j.ejvs.2014.04.026

Sims, R. C., Hosey, M., Levy, S. A., Whitfield, K. E., Katzel, L. I., & Waldstein, S. R. (2014). Distinct
functions of social support and cognitive function among older adults. Exp Aging Res, 40(1), 40-
59. https://doi.org/10.1080/0361073X.2014.857551

Sinoff, G., & Werner, P. (2003). Anxiety disorder and accompanying subjective memory loss in the
elderly as a predictor of future cognitive decline. Int J Geriatr Psychiatry, 18(10), 951-959.
https://doi.org/10.1002/gps.1004

Snyder, B., Duong, P., Trieu, J., & Cunningham, R. L. (2018). Androgens modulate chronic intermittent
hypoxia effects on brain and behavior. Horm Behav, 106, 62-73.
https://doi.org/10.1016/j.yhbeh.2018.09.005

107

Solomon, A., Kivipelto, M., Wolozin, B., Zhou, J., & Whitmer, R. A. (2009). Midlife serum cholesterol and
increased risk of Alzheimer's and vascular dementia three decades later. Dement Geriatr Cogn
Disord, 28(1), 75-80. https://doi.org/10.1159/000231980

Sposito, G., Neri, A. L., & Yassuda, M. S. (2015). Cognitive performance and engagement in physical,
social and intellectual activities in older adults: The FIBRA study. Dement Neuropsychol, 9(3),
270-278. https://doi.org/10.1590/1980-57642015DN93000010

St John, J. A., Henderson, V. W., Hodis, H. N., Kono, N., McCleary, C. A., Franke, A. A., & Mack, W. J.
(2014). Associations between urine excretion of isoflavonoids and cognition in postmenopausal
women in the Women's Isoflavone Soy Health clinical trial. J Am Geriatr Soc, 62(4), 629-635.
https://doi.org/10.1111/jgs.12752

Stein, J. H., Korcarz, C. E., Hurst, R. T., Lonn, E., Kendall, C. B., Mohler, E. R., . . . Post, W. S. (2008). Use of
carotid ultrasound to identify subclinical vascular disease and evaluate cardiovascular disease
risk: A consensus statement from the American society of echocardiography carotid intima-
media thickness task force endorsed by the society for vascular medicine. Journal of the
American Society of Echocardiography, 21(2), 93-111.
https://doi.org/10.1016/j.echo.2007.11.011

Stern, Y. (2002). What is cognitive reserve? Theory and research application of the reserve concept. J Int
Neuropsychol Soc, 8(3), 448-460.

Suemoto, C. K., Santos, I. S., Bittencourt, M. S., Pereira, A. C., Goulart, A. C., Rundek, T., . . . Benseñor, I.
M. (2015). Subclinical carotid artery atherosclerosis and performance on cognitive tests in
middle-aged adults: Baseline results from the ELSA-Brasil. Atherosclerosis, 243(2), 510-515.
https://doi.org/10.1016/j.atherosclerosis.2015.10.008

Tan, S., Sohrabi, H. R., Weinborn, M., Tegg, M., Bucks, R. S., Taddei, K., . . . Martins, R. N. (2019). Effects
of Testosterone Supplementation on Separate Cognitive Domains in Cognitively Healthy Older
Men: A Meta-analysis of Current Randomized Clinical Trials. Am J Geriatr Psychiatry, 27(11),
1232-1246. https://doi.org/10.1016/j.jagp.2019.05.008

Tervo, S., Kivipelto, M., Hänninen, T., Vanhanen, M., Hallikainen, M., Mannermaa, A., & Soininen, H.
(2004). Incidence and risk factors for mild cognitive impairment: a population-based three-year
follow-up study of cognitively healthy elderly subjects. Dement Geriatr Cogn Disord, 17(3), 196-
203. https://doi.org/10.1159/000076356

Thomas, M. L., Kaufmann, C. N., Palmer, B. W., Depp, C. A., Martin, A. S., Glorioso, D. K., . . . Jeste, D. V.
(2016). Paradoxical Trend for Improvement in Mental Health With Aging: A Community-Based
Study of 1,546 Adults Aged 21-100 Years. J Clin Psychiatry, 77(8), e1019-1025.
https://doi.org/10.4088/JCP.16m10671

Thong, J. Y. J., Hilal, S., Wang, Y. B., Soon, H. W., Dong, Y., Collinson, S. L., . . . Qiu, A. Q. (2013).
Association of silent lacunar infarct with brain atrophy and cognitive impairment. Journal of
Neurology Neurosurgery and Psychiatry, 84(11), 1219-1225. https://doi.org/10.1136/jnnp-2013-
305310

108

Tilvis, R. S., Kähönen-Väre, M. H., Jolkkonen, J., Valvanne, J., Pitkala, K. H., & Strandberg, T. E. (2004).
Predictors of cognitive decline and mortality of aged people over a 10-year period. J Gerontol A
Biol Sci Med Sci, 59(3), 268-274. https://doi.org/10.1093/gerona/59.3.m268

Toth, P. P. (2008). Subclinical atherosclerosis: what it is, what it means and what we can do about it. Int J
Clin Pract, 62(8), 1246-1254. https://doi.org/10.1111/j.1742-1241.2008.01804.x

Uchino, B. N. (2009). Understanding the Links Between Social Support and Physical Health: A Life-Span
Perspective With Emphasis on the Separability of Perceived and Received Support. Perspect
Psychol Sci, 4(3), 236-255. https://doi.org/10.1111/j.1745-6924.2009.01122.x

Umberson, D. (1987). Family status and health behaviors: social control as a dimension of social
integration. J Health Soc Behav, 28(3), 306-319.

van den Munckhof, I. C. L., Jones, H., Hopman, M. T. E., de Graaf, J., Nyakayiru, J., van Dijk, B., . . .
Thijssen, D. H. J. (2018). Relation between age and carotid artery intima-medial thickness: a
systematic review. Clinical Cardiology, 41(5), 698-704. https://doi.org/10.1002/clc.22934

Viticchi, G., Falsetti, L., Vernieri, F., Altamura, C., Bartolini, M., Luzzi, S., . . . Silvestrini, M. (2012).
Vascular predictors of cognitive decline in patients with mild cognitive impairment. Neurobiol
Aging, 33(6), 1127.e1121-1129. https://doi.org/10.1016/j.neurobiolaging.2011.11.027

von Zglinicki, T., Serra, V., Lorenz, M., Saretzki, G., Lenzen-Grossimlighaus, R., Gessner, R., . . .
Steinhagen-Thiessen, E. (2000). Short telomeres in patients with vascular dementia: an indicator
of low antioxidative capacity and a possible risk factor? Lab Invest, 80(11), 1739-1747.
https://doi.org/10.1038/labinvest.3780184

VonDras, D. D., Powless, M. R., Olson, A. K., Wheeler, D., & Snudden, A. L. (2005). Differential effects of
everyday stress on the episodic memory test performances of young, mid-life, and older adults.
Aging Ment Health, 9(1), 60-70. https://doi.org/10.1080/13607860412331323782

Wang, L., Muxin, G., Nishida, H., Shirakawa, C., Sato, S., & Konishi, T. (2007). Psychological stress-
induced oxidative stress as a model of sub-healthy condition and the effect of TCM. Evid Based
Complement Alternat Med, 4(2), 195-202. https://doi.org/10.1093/ecam/nel080

Wang, M., Norris, C. M., Graham, M. M., Santana, M., Liang, Z., Awosoga, O., . . . Sajobi, T. T. (2019).
Trajectories of perceived social support in acute coronary syndrome. Qual Life Res, 28(5), 1365-
1376. https://doi.org/10.1007/s11136-018-02095-4

Wendell, C. R., Waldstein, S. R., Evans, M. K., & Zonderman, A. B. (2016). Subclinical carotid
atherosclerosis and neurocognitive function in an urban population. Atherosclerosis, 249, 125-
131. https://doi.org/10.1016/j.atherosclerosis.2016.04.009

Wendell, C. R., Waldstein, S. R., Ferrucci, L., O'Brien, R. J., Strait, J. B., & Zonderman, A. B. (2012). Carotid
Atherosclerosis and Prospective Risk of Dementia. Stroke, 43(12), 3319-3324.
https://doi.org/10.1161/strokeaha.112.672527

109

Wendell, C. R., Zonderman, A. B., Metter, E. J., Najjar, S. S., & Waldstein, S. R. (2009). Carotid Intimal
Medial Thickness Predicts Cognitive Decline Among Adults Without Clinical Vascular Disease.
Stroke, 40(10), 3180-3185. https://doi.org/10.1161/strokeaha.109.557280

Weng, X., George, D. R., Jiang, B., & Wang, L. (2020). Association Between Subjective Cognitive Decline
and Social and Emotional Support in US Adults. Am J Alzheimers Dis Other Demen, 35,
1533317520922392. https://doi.org/10.1177/1533317520922392

Wilson, R. S., Evans, D. A., Bienias, J. L., Mendes de Leon, C. F., Schneider, J. A., & Bennett, D. A. (2003).
Proneness to psychological distress is associated with risk of Alzheimer's disease. Neurology,
61(11), 1479-1485. https://doi.org/10.1212/01.wnl.0000096167.56734.59

Wilson, R. S., Krueger, K. R., Arnold, S. E., Schneider, J. A., Kelly, J. F., Barnes, L. L., . . . Bennett, D. A.
(2007). Loneliness and risk of Alzheimer disease. Arch Gen Psychiatry, 64(2), 234-240.
https://doi.org/10.1001/archpsyc.64.2.234

Worthy, D. A., Gorlick, M. A., Pacheco, J. L., Schnyer, D. M., & Maddox, W. T. (2011). With age comes
wisdom: decision making in younger and older adults. Psychol Sci, 22(11), 1375-1380.
https://doi.org/10.1177/0956797611420301

Wortmann, M. (2012). Dementia: a global health priority - highlights from an ADI and World Health
Organization report. Alzheimers Res Ther, 4(5), 40. https://doi.org/10.1186/alzrt143

Yeh, S. C., & Liu, Y. Y. (2003). Influence of social support on cognitive function in the elderly. BMC Health
Serv Res, 3(1), 9. https://doi.org/10.1186/1472-6963-3-9

Yerkes, R. M., & Dodson, J. D. (1908). The relation of strength of stimulus to rapidity of habit-formation.
Journal of Comparative Neurology and Psychology, 18(5), 459-482.
https://doi.org/10.1002/cne.920180503

Zamora-Macorra, M., de Castro, E. F., Ávila-Funes, J. A., Manrique-Espinoza, B. S., López-Ridaura, R.,
Sosa-Ortiz, A. L., . . . Del Campo, D. S. (2017). The association between social support and
cognitive function in Mexican adults aged 50 and older. Arch Gerontol Geriatr, 68, 113-118.
https://doi.org/10.1016/j.archger.2016.10.005

Zhong, B. L., Chen, S. L., Tu, X., & Conwell, Y. (2017). Loneliness and Cognitive Function in Older Adults:
Findings From the Chinese Longitudinal Healthy Longevity Survey. J Gerontol B Psychol Sci Soc
Sci, 72(1), 120-128. https://doi.org/10.1093/geronb/gbw037

Zhou, Z., Mao, F., Zhang, W., Towne, S. D., Wang, P., & Fang, Y. (2019). The Association Between
Loneliness and Cognitive Impairment among Older Men and Women in China: A Nationwide
Longitudinal Study. Int J Environ Res Public Health, 16(16).
https://doi.org/10.3390/ijerph16162877

Zhu, S., Hu, J., & Efird, J. T. (2012). Role of social support in cognitive function among elders. J Clin Nurs,
21(15-16), 2118-2125. https://doi.org/10.1111/j.1365-2702.2012.04178.x

110

Zucoloto, M. L., Santos, S. F., Terada, N. A. Y., & Martinez, E. Z. (2019). Construct validity of the Brazilian
version of the Medical Outcomes Study Social Support Survey (MOS-SSS) in a sample of elderly
users of the primary healthcare system. Trends Psychiatry Psychother, 41(4), 340-347.
https://doi.org/10.1590/2237-6089-2018-0092

Zuelsdorff, M. L., Engelman, C. D., Friedman, E. M., Koscik, R. L., Jonaitis, E. M., Rue, A. L., & Sager, M. A.
(2013). Stressful events, social support, and cognitive function in middle-aged adults with a
family history of Alzheimer's disease. J Aging Health, 25(6), 944-959.
https://doi.org/10.1177/0898264313498416

111

APPENDICES

112

APPENDIX A:
Medical Outcomes Study-Social Support Survey (MOS-SSS)

People sometimes look to others for companionship, assistance, or other types of support.
How often is each of the following kinds of support available to you if you need it?

None
of the
time
A little
of the
time
Some
of the
time
Most
of the
time
All of
the
time
2. Someone to help you if you were confined to
bed
1 2 3 4 5
3. Someone you can count on to listen to you
when you need to talk
1 2 3 4 5
4. Someone to give you good advice about a crisis 1 2 3 4 5
5. Someone to take you to the doctor if you need
it
1 2 3 4 5
6. Someone who shows you love and affection 1 2 3 4 5
7. Someone to have a good time with 1 2 3 4 5
8. Someone to give you information to help you
understand a situation
1 2 3 4 5
9. Someone to confide in or talk to about yourself
or your problems
1 2 3 4 5
10. Someone who hugs you 1 2 3 4 5
11. Someone to get together with for relaxation 1 2 3 4 5
12. Someone to prepare your meals if you were
unable to do it yourself
1 2 3 4 5
13. Someone whose advice you really want 1 2 3 4 5
14. Someone to do things with to help you get your
mind off things
1 2 3 4 5
15. Someone to help with daily chores if you were
sick
1 2 3 4 5
16. Someone to share your most private worries
and fears with
1 2 3 4 5
17. Someone to turn to for suggestions about how
to deal with a personal problem
1 2 3 4 5
18. Someone to do something enjoyable with 1 2 3 4 5
19. Someone who understands your problems 1 2 3 4 5
20. Someone to love and make you feel wanted 1 2 3 4 5
About how many close friends and close relatives do you have (people who you
feel at ease with and can talk to about what is on your mind)?

113

APPENDIX B:
Perceived Stress Scale (PSS)
The questions in this scale ask you about your feelings and thoughts during the past 2 weeks. For each
item, please answer by circling how often you felt or thought a certain way.

In the past 2 weeks…

None
of the
time
Almost
never
Some
of the
time
Fairly
often
Very
often
1. How often have you been upset because of
something that happened unexpectedly?
1 2 3 4 5
1. How often have you felt that you were unable
to control the important things in your life?
1 2 3 4 5
2. How often have you felt nervous and
“stressed”?
1 2 3 4 5
3. How often have you found that you could not
cope with all the things that you had to do?
1 2 3 4 5
4. How often have you been angered because of
things (that happened) that were outside your
control?
1 2 3 4 5
5. How often have you felt difficulties were piling
up so high that you could not overcome them?
1 2 3 4 5
6. How often have you felt confident (about your
ability to handle your personal problems)?
1 2 3 4 5
7. How often have you felt that things were going
your way?
1 2 3 4 5
8. How often have you felt that you were on top
of things?
1 2 3 4 5
9. How often have you been able to control
irritations in your life?
1 2 3 4 5

114

APPENDIX C:
PERI Events Scale
Listed below are a number of events that sometimes happen in life. Please let us know if any of these
happened to you in the past 12 months.

(1) Check only the item(s) that happened to you in the past 12 months (leave others blank)
(2) When did it happen? (month/year)
(3) Rate how undesirable do you consider the item was to you:

1 = Not undesirable at all
2 = A little undesirable
3 = Somewhat undesirable
4 = Moderately undesirable
5 = Quite undesirable
6 = Very undesirable
7 = Extremely undesirable

Check if
happened
When did it
happen
(month/year)
How
undesir-
able?

In the past 12 months…
About school

1.

Started school or a training program after not
going to school for a long time.

2.

Changed schools or in training program.

3.

Graduated from school or training program.

4. Had problems in school or in training program.

5. Failed school or training program.

6.

Did not graduate from school or training
program.
About work

7. Started work for the first time.

8.
Returned to work after not working for a long
time.

9. Changed jobs for a better one.

10. Changed jobs for a worse one.

115

11.
Changed jobs for one that was no better and no
worse than the last one.

12.

Had trouble with a boss.

13. Demoted at work.

14. Found out not going to be promoted at work.

15.

Conditions at work got worse, other than
demotion or trouble with the boss.

16. Promoted.

17. Had significant success at work.

18.

Conditions at work improved, not counting
promotion or other personal success.

19. Laid off.

20. Fired.

21. Started a business or profession.

22. Expanded business or professional practice.

23. Took on a greatly increased work load.

24. Suffered a business loss or failure.

25. Sharply reduced work load.

26. Retired.

27.

Stopping working, not retirement, for an
extended period.
About relationships or marriage

28. Became engaged.

29. Engagement was broken.

30. Married.

116

31. Started a love affair.

32.

Relationship with partner/significant other
changed for the worse, without separation or
divorce.

33. Couple separated.

34. Termination of love relationship.

35.

Relations with spouse/significant other changed
for the better.

36.

Couple reunited after separation.

37. Infidelity on behalf of spouse/significant other.

38. Trouble with in-laws.

39. Spouse died.

40. Boyfriend/girlfriend died.
About children

41. Became pregnant.

42.

Birth of a first child or grandchild (please circle
one).

43.

Birth of a second or later child or grandchild
(please circle one).

44. Abortion.

45. Miscarriage or stillbirth.

46. Found out that cannot have children.

47. Child died.

48. Adopted a child.

49. Started menopause.

117

About family

50. New person moved into household.

51. Person moved out of the household.

52.
Someone stayed on in the household after he or
she was expected to leave.

53.

Serious family argument other than with spouse.

54.

A change in the frequency of family get-
togethers.

55. Family member other than spouse or child dies:

Mother

Father

Brother or sister

Grandparent

Other (please list):
About housing

56. Moved to a better residence or neighborhood.

57. Moved to a worse residence or neighborhood.

58.

Moved to a residence or neighborhood no better
or no worse than the last one.

59.

Unable to move after expecting to be able to
move on.

60. Built a home or had one built.

61. Remodeled a home.

62. Suffered severe damage to a home.

63. Lost a home through fire, flood or other disaster.

118

About crimes, legal matters

64. Lost property due to theft.

65. Was robbed.

66. Accident in which there were no injuries.

67. Involved in a law suit.

68.

Accused of something for which a person could
be sent to jail.

69. Lost driver’s license.

70. Arrested.

71. Went to jail.

72. Got involved in a court case.

73. Convicted of a crime.

74. Acquitted of a crime.

75. Released from jail.

76. Didn’t get out of jail when expected.
About finances

77. Took out a mortgage.

78.

Started buying a car, furniture, or other large
purchase on the installment plan.

79. Foreclosure of a mortgage or loan.

80.

Repossession of a car, furniture or other items
bought on the installment plan.

81. Took a cut in wage or salary without a demotion.

82.

Suffered a financial loss or loss of property not
related to work.

83. Went on welfare.

119

84. Went off welfare.

85.

Got a substantial increase in wage or salary
without a promotion.

86. Did not get an expected wage or salary increase.

87. Had financial improvement not related to work.
About social activities

88.

Increased church or synagogue, club,
neighborhood, or other organizational activities.

89. Took a vacation.

90. Was not able to take a planned vacation.

91.

Took up a new hobby, sport, craft, or recreational
activity.

92.

Dropped a hobby, sport, craft, or recreational
activity.

93. Acquired a pet.

94. Pet died or was seriously ill.

95. Made new friends.

96. Broke up with a friend.

97. Close friend died.
Miscellaneous

98.

Entered the Armed Services.

99. Left the Armed Services.

100. Took a trip other than a vacation.
About health and safety

101. Physical health improved.

102. Physical illness.

120

103. Injury.

104. Unable to get treatment for an illness or injury.

105.

Serious illness or accident (life-threatening) that
happens to friends/family:

Spouse

Child

Boyfriend/girlfriend

Close friend

Close family member

Distant family member

106.

Sexual assault or forced/pressured sexual contact
by someone other than a spouse or partner.

107.
Sexual assault or forced/pressured sexual contact
by a spouse or partner.

108.
Physical assault or unwanted physical contact
(hitting, kicking, pushing, slapping) by someone
other than a spouse or partner.

109.
Physical assault or unwanted physical contact by
a spouse or partner.

110. Experienced a natural disaster.

Abstract (if available)

Abstract The projected number of people aged 65 years and older with Alzheimer’s disease is expected to increase substantially in the coming decades. This can be attributed to the combined effects of longer life expectancy and the large number of baby boomers reaching old age. Alzheimer’s and associated dementias result in an enormous socioeconomic burden that is also overwhelming on a personal level for caregivers and family members. In this context, the discovery of modifiable risk factors for cognitive impairment and dementia is important.

Previous studies have shown mixed results for associations between subclinical atherosclerosis, psychosocial factors such as perceived social support, perceived stress, stressful life events and reduced cognitive function. The associations between these factors and particular cognitive domains are also not well known. There are several possible reasons for the lack of consistency in previous findings. One is that various instruments used to evaluate the same cognitive domain differ in their sensitivity to the assessment of cognitive decline. Furthermore, some studies used only a single instrument while others employed multi-test batteries to assess cognitive function. Studies employing batteries may be more likely to detect impairments in particular cognitive domains. In addition, a variety of tools have been used to assess stress, most being one-dimensional and only measuring an aspect of stress (e.g., psychological or physiological). As a result, there is wide variation in how different measures of stress are associated with cognitive outcomes.

This dissertation aims to examine the associations between subclinical atherosclerosis as measured by carotid artery intima-media thickness (CIMT) and four cognitive domains: executive function, verbal memory, visual memory, and global cognition in a population of participants enrolled in three clinical trials: the BVAIT, WISH, and ELITE studies (n = 1,495, mean age = 61 years). Cognitive function was assessed by a battery of 14 cognitive and neuropsychological tests that are validated and sensitive to age-associated change in middle- to older-aged adults. This dissertation will also examine the associations between perceived social support, perceived stress, and stressful life events with the same four cognitive domains in a population of women enrolled in the ELITE psychosocial substudy (n = 448, mean age = 61 years). We hypothesize that higher CIMT, high levels of perceived stress and a greater number of stressful life events experienced are associated with poorer cognitive function. We also hypothesize that high levels of perceived social support have a protective effect on cognitive function.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Association of carotid artery stiffness with measures of cognition in older adults

PDF

Cardiovascular disease risk factors and cognitive function

PDF

The associations of inflammatory markers with progression of subclinical atherosclerosis in early and late postmenopausal women

PDF

Subclinical atherosclerosis in overweight Latino youth: influence of cardiometabolic risk factors

PDF

Association of oophorectomy with progression of carotid atherosclerosis

PDF

The effect of vitamin D supplementation on the progression of carotid intima-media thickness and arterial stiffness in elderly African American women: Results of a randomized placebo-controlled trial

PDF

Hormone therapy timing hypothesis and atherosclerosis

PDF

Relationship of blood pressure and antihypertensive medications to cognitive change in the BVAIT, WISH, and ELITE clinical trials

PDF

Dietary fats, atherosclerosis, and cardiovascular disease

PDF

Association of arterial stiffness progression with subclinical atherosclerosis measurements in postmenopausal women

PDF

Bone mineral density is associated with carotid atherosclerosis in healthy postmenopausal women: a longitudinal analysis of randomized clinical trial data

PDF

Association of subclinical atherosclerosis with plasma B-vitamin, cysteine, homocysteine, and cysteinyl glycine in a cardiovascular disease-free population

PDF

Associations between inflammatory markers and change in cognitive endpoints

PDF

Relationship between progression of atherosclerosis and coagulation measures in a randomized-controlled trial

PDF

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

PDF

Cross-sectional association of blood pressure, antihypertensive medications, MRI volumetric measures and cognitive function scores in an aging population

PDF

Association between endogenous sex hormone levels and cognition: the Women’s Isoflavone Soy Health (WISH) trial

PDF

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

PDF

Investigating a physiological pathway for the effect of guided imagery on insulin resistance

PDF

Cerebrovascular mechanisms of cognitive enhancement after periodized resistance training in older adults

Asset Metadata

Creator Lin, Felice (author)

Core Title Subclinical carotid atherosclerosis, psychosocial measures, and cognitive function in middle- to older-aged adults

School Keck School of Medicine

Degree Doctor of Philosophy

Degree Program Epidemiology

Degree Conferral Date 2022-12

Publication Date 09/20/2022

Defense Date 08/18/2022

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

Tag carotid artery intima-media thickness,cognitive function,middle- to older-aged adults,OAI-PMH Harvest,perceived social support,perceived stress,stressful life events,subclinical atherosclerosis

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Mack, Wendy (committee chair), Han, S. Duke (committee member), Hodis, Howard N. (committee member), Karim, Roksana (committee member), Pa, Judy (committee member)

Creator Email felice.lin@gmail.com,flin@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC112013946

Unique identifier UC112013946

Legacy Identifier etd-LinFelice-11236

Document Type Dissertation

Format application/pdf (imt)

Rights Lin, Felice

Type texts

Source 20220921-usctheses-batch-983 (batch), 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 author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

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

Repository Email cisadmin@lib.usc.edu