Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
Impacts of caregiving on wellbeing among older adults and their spousal caregivers in the United States
(USC Thesis Other)
Impacts of caregiving on wellbeing among older adults and their spousal caregivers in the United States
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
IMPACTS OF CAREGIVING ON WELLBEING AMONG OLDER ADULTS AND THEIR
SPOUSAL CAREGIVERS IN THE UNITED STATES
by
RUOTONG (MONA) LIU
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
[SOCIAL WORK]
August 2022
Copyright 2022 Ruotong (Mona) Liu
ii
Dedication
This dissertation is wholeheartedly dedicated to Mom, Dad, and Dandan – thank you for
making this agreeably adventurous journey possible. I love you.
To Meemaw – thank you for being the most caring caregiver anyone could ever ask for;
maybe I could get the chance to take care of you someday. I miss you.
To all the unrepeatable encounters – great to have met you. Au revoir.
iii
Acknowledgements
When I first heard about Max Weber and his “Science as a Vocation” thirteen years ago,
as much as I would love to say it hit me like an epiphany, it was probably just another cached
phrase. While here I am today, wrapping up my dissertation and the eight years in Los Angeles,
ready, or trying to be ready, to embark on the said vocation.
A new journey almost always seems exciting and frightening; nostalgia and sentiment
keep kicking in as I am checking and double checking the packing list, realizing that the suitcase
is already packed, probably not that efficiently, but surely gradually and neatly. I could not be
luckier to have Drs. Iris Chi and Shinyi Wu as my advisors, who have enabled a “search wide,
dig deep” path infused with knowledge, expertise, and all the essentials to be carried forward.
Dr. Chi has helped me find my track, and keeps me on track when my thoughts are floating; her
emphasis on theories and foundations has planted seeds for the sprawling roots that hold me
when I am attracted to, or distracted by, the dazzling methodologies. Without Dr. Wu, I would
not have had the many opportunities to explore multiple fascinating topics, especially those at
the intersection of gerontology and technology. She has set a perfect example in accommodating
multidisciplinary interests, and has granted me the courage to seek more possibilities.
Many thanks also go to Drs. Julie Zissimopoulos and Yuri Jang, who have guided and
accompanied me at many milestones in the journey, witnessing and contributing to the
maturation that is beyond the passage of time. To many other faculty members and staff:
Malinda Sampson, Drs. Michael Hurlburt, Maryalice Jordan-Marsh, Michàlle Mor Barak, Susan
Enguídanos, Min-Kyoung Rhee, Suzanne Wenzel, Lawrence Palinkas, Jordan Davis, and fellow
PhD student colleagues – thank you for having broadened my view, for having created a
supportive space for personal and professional growth, for being yourselves and demonstrating
iv
that there is still meaning and value in the field that is more than Sisyphean and more than
instrumental rationality, despite the disenchanted world and age we are in, and for assuring me
that I am, we are, not alone.
To Mom and Dad. It is a privilege to be able to spend this many years in a city that is
across the ocean, one that I have been starting to call a second home. What was labelled as
“testing the water” back then can probably be renamed as “pursuing the dream” right now.
Thank you for supporting me, unconditionally, allowing me to wander, to sprint, knowing I have
the strongest anchor called family and home; I so want to be your anchor, your rock, and have
the city of stars shining for you, soon. To Meemaw. I do not know if God is watching but I do
hope you are. I am still trying to be the person you are proud of, and want to be ready, always,
anytime, to share with you all the collected moments wherever whenever we meet again. To
Dandan, I am beyond grateful to have you as my best friend since day one. To Kajima (“don’t
go”) and Gomawo (“thank you”), thank you for being fluffy and silly, for being with me.
I did not and still do not know whether I have defended the dissertation, or it has
defended me; as I am not sure whether it is me packing the suitcase for a desired destination, or
the suitcase prepared by the moments has seized me and guided me on the seemingly inevitable
path – science as a vocation. Regardless, it is my pleasure to have connected and be connecting
all the scattered dots in the way it is; it is my honor to have all the crossed- and crossing-paths
shaped and updated me. Thank you thank you thank you.
v
Table of Contents
Dedication ........................................................................................................................................ ii
Acknowledgements ........................................................................................................................ iii
List of Tables ................................................................................................................................. vii
List of Figures ............................................................................................................................... viii
Abstract ........................................................................................................................................... ix
Chapter 1: Introduction ................................................................................................................... 1
Background .............................................................................................................................. 1
Theoretical framework and rationale ....................................................................................... 4
Limitations of existing studies ............................................................................................... 10
Specific aims ......................................................................................................................... 10
Contributions and implications .............................................................................................. 11
Chapter 2: Impacts of Caregiving on Wellbeing among Spousal Caregivers to Older Adults in
the United States: A Coarsened Exact Matching Analysis ........................................................... 14
Introduction ........................................................................................................................... 14
Methods ................................................................................................................................. 17
Results ................................................................................................................................... 22
Discussion .............................................................................................................................. 25
Chapter 3: A Latent Class Analysis of Stressors and Resources among Spousal Caregivers to
Older Adults in the United States and Their Associations with Wellbeing .................................. 37
Introduction ........................................................................................................................... 37
Methods ................................................................................................................................. 43
Results ................................................................................................................................... 49
Discussion .............................................................................................................................. 53
Chapter 4: Predicting Care-Recipients’ Wellbeing Based on Spousal Caregivers’ Co-Occurring
Stressors and Resources ................................................................................................................ 66
Introduction ........................................................................................................................... 66
Methods ................................................................................................................................. 70
Results ................................................................................................................................... 74
Discussion .............................................................................................................................. 76
vi
Chapter 5: Conclusions and Implications ..................................................................................... 85
Summary of major research findings .................................................................................... 87
Implications for future research ............................................................................................. 91
Implications for policy and practice ...................................................................................... 93
References ..................................................................................................................................... 96
Appendix ..................................................................................................................................... 123
vii
List of Tables
Table 2.1. Standardized differences in variables used for CEM before and after matching by
cohort, weight adjusted .................................................................................................................. 31
Table 2.2. Descriptive statistics of matched samples (weight adjusted) ....................................... 34
Table 2.3. Stepwise regression results for health outcomes associated with spousal caregiving .. 35
Table 3.1. Sample characteristics of spousal caregivers (N=793) ................................................. 59
Table 3.2. Model fit statistics for spousal caregiver latent classes with two- to four-class
solutions ......................................................................................................................................... 60
Table 3.3. Spousal caregiver latent classes characterizing stressors and resources (N=793) ....... 61
Table 3.4. Sample characteristics across identified spousal caregiver latent classes (N=793) ..... 62
Table 3.5. Association of spousal caregiver subgroups and wellbeing: Results from linear
regression models (N=793) ........................................................................................................... 63
Table 4.1. Sample characteristics and comparison across three spousal caregiver latent classes . 81
Table 4.2. Generalized estimating equation models of self-rated health over time across spousal
caregiver classes (N=639) ............................................................................................................. 82
Table 4.3. Generalized estimating equation models of depressive symptoms over time across
spousal caregiver classes (N=639) ................................................................................................ 83
viii
List of Figures
Figure 1.1. Conceptual framework for the dissertation ................................................................. 13
Figure 3.1. Latent classes among spousal caregivers to older adults based on the relative
intensities of stressors and resources as informed by the stress process model ............................ 64
Figure 3.2. Spousal caregiver latent classes based on relative intensities of stressors and
resources (N=793) ......................................................................................................................... 65
Figure 4.1. Changes in health across classes ................................................................................. 84
ix
Abstract
Spousal caregivers to older adults are at the intersection of partners and caregivers; they
need to simultaneously fulfil their supportive roles and cope with the stress caused by caring
activities. Compared to other family caregivers such as adult child, spousal caregivers are more
likely to encounter more stressors and poorer physical and mental health, because of the declined
physical condition and increased daily functional needs for both the care-recipients and
themselves. Meanwhile, studies have suggested increasing evidence showing beneficial results
from caregiving such as increased interpersonal relationships, increased quality of life, and lower
mortality. Explanations for the different findings include selection bias: (1) when comparing
caregivers to non-caregivers, studies recruiting participants from convenience samples could
exaggerate the negative effects of caregiving due to socially active individuals’ self-selection
into participation as non-caregivers, and (2) when healthier individuals uptake caregiving role or
remain in caring activities, their self-selection into caregiving because of better initial health
status could confound the findings on health outcomes especially mortalities. The other
explanation concerns the small effect size or insignificant findings when comparing caregivers to
non-caregivers, indicating that caregivers are heterogeneous, and some caregivers, such as
spousal caregivers, could have higher levels of stress than other types of caregivers and
subsequently worse health outcomes.
The dissertation, therefore, chose to focus on a relatively homogenous caregiver
subgroup – spousal caregivers. With the stress process model as the theoretical foundation, the
first study (Chapter 2) examined the whole spousal caregiver group in comparison to their
spousal non-caregiver counterparts, aiming to have an overall understanding of the health
impacts of caregiving that are independent of selection bias. The second (Chapter 3) and third
x
(Chapter 4) studies emphasized the heterogeneity among spousal caregivers and their care-
recipients. The two studies explored spousal caregivers’ latent classes based on the relative
intensities of co-occurring stressors and resources they encounter and possess, and the
association between identified latent classes and spousal caregivers’ profiles as well as care-
recipients’ health outcomes.
Using data from the Health and Retirement Study, the first study compared 2,741 spousal
caregivers with 18,043 spousal non-caregivers using coarsened exact matching. Results indicated
that caregiving was associated with a 0.27-unit increase in depressive symptoms in the
subsequent wave. This highlighted the elevated needs to address mental health among spousal
caregivers.
The second and third studies utilized data from National Health and Aging Trends Study
and National Study of Caregiving. Using latent class analysis, the second study identified three
spousal caregiver latent classes based on the relative intensities of co-occurring stressors and
resources from 793 observations across 639 spousal caregivers. The three latent classes were:
low-stress low-support spousal caregivers, medium-stress high-support caregivers, and high-
stress medium-support caregivers. Compared to low-stress low-support spousal caregivers, high-
stress medium-support caregivers were found to have worse self-rated health and higher levels of
depressive symptoms after controlling for covariates. Based on the identified spousal caregiver
latent classes, the third study further explored how caregivers’ class membership was associated
with their care-recipients’ health at baseline and rates of changes in health. Results indicated that
compared to the counterparts taken care of by low-stress low-support caregivers, recipients taken
care of by medium-stress high-support spousal caregivers had lower levels of self-rated health at
baseline, and those with high-stress medium-support spousal caregivers had both worse self-
xi
rated heath and higher levels of depressive symptoms at baseline. No statistically significant
differences in health changes were detected across the three groups within the one-year time
interval.
The findings indicated that spousal caregiving had a negative impact on caregivers’
mental health. Moreover, results highlighted that the combination of stressors and resources
featured by their co-occurring relative intensities is related to spousal caregivers’ and recipients’
health outcomes: spousal caregivers and the corresponding recipients have the worst health
outcomes when the stressors caregivers encounter are at a higher level relative to the resources
they possess. Social workers and policy makers should, therefore, focus on mental health of
spousal caregivers in general, and pay special attention to the spousal dyads whose level of
stressors exceeds that of the resources they can mobilize.
Chapter 1. Introduction
Background
The aging U.S. population brings an increase in the demand for long-term care (LTC):
7.9 million (15.9%) of older adults aged 65 years or older needed LTC in 2018 (AARP, 2019a),
and the number is anticipated to more than double with about 19 million of older adults needing
LTC by 2050 (Spetz et al., 2015). About two-thirds of LTC is provided by family and other
unpaid caregivers, who contribute an estimated $470 billions of labor force annually (AARP,
2019b). Among family caregivers, spousal caregivers account for 12%, making them the second
largest informal caregiver group behind children (AARP, 2019b). In addition, the prevalence of
spousal caregiving increases with age: it is estimated that 9% 50- to 64-year-old, 24% of 65- to
74-year-old, and 46% of 75+ year-old caregivers are caring for a spouse or partner (AARP,
2015). Moreover, 78% of the spousal caregivers are solo caregivers, especially in care-
recipients’ last years of life (Ornstein et al., 2019); as the population ages, therefore, the demand
for spousal caregiving is expected to grow.
Spousal caregivers are at the intersection of partners and caregivers; they need to
simultaneously fulfill the supportive roles and cope with the stress caused by caring activities
(Ornstein et al., 2019; Schulz & Eden, 2016). Compared to other family caregivers, spousal
caregivers are experiencing higher levels of stressors and poorer physical and mental health,
especially among older spouses, for reasons such as declined physical condition and increased
daily functioning needs for both themselves and the care-recipients(Chen et al., 2020; Khalaila &
Cohen, 2016; Lavela & Ather, 2010; Ribé et al., 2018; Roth et al., 2020; Schulz & Sherwood,
2008). Despite the widely documented negative health outcomes among caregivers, more recent
studies have suggested increasing evidence of beneficial results from caregiving such as
1
increased wellbeing (Roth et al., 2015), increased interpersonal relationships (Yu et al., 2018),
and lower mortality compared to non-caregivers (Roth et al., 2018).
The inconsistent findings can be explained in two ways; one concerns the
comparableness between caregivers and non-caregiver comparisons, and the other focuses on the
heterogeneity among caregivers. For comparableness, researchers have proposed that non-
caregivers are oftentimes convenience sample recruited from revenues such as newsletters or
social groups, which could exaggerate the negative health consequences of caregiving (Roth et
al., 2015), since poorer health among caregivers is equivalent to better health among socially
active individuals (Okun et al., 2013). The healthy caregiver hypothesis, on the other hand,
attempts to address the exaggerated positive impacts of caregiving. It posits that people need to
maintain a minimum level of health to be able to perform caregiving tasks, while individuals
with poor health condition either do not enter or eventually quit caring activities (Fredman et al.,
2015). Therefore, the potential confounding mechanisms in selection bias and endogeneity
would contribute to the lower mortality among caregivers.
The other explanation for the inconclusive findings on health outcomes when comparing
caregivers to non-caregivers is the heterogeneity among caregivers: results from systematic
review and meta-analysis indicated that the differences in health outcomes between caregivers
and non-caregivers were smaller when the caregivers are heterogeneous samples (Pinquart &
Sörensen, 2003b).
The stress process model (Pearlin et al., 1981) has been widely utilized to understand
factors that contribute to caregivers’ health outcome. However, what has received limited
attention so far, is the isolation of stress process – stressors and factors that buffer the effects of
stress, that is directly related to caregiving. One’s own health concerns and the presence of a
2
family member in need, especially when the family member is a spouse/partner, are both
stressors that could have negative impacts on one’s own health or wellbeing (Bobinac et al.,
2010; Bom et al., 2019), regardless of the provision of care. Therefore, the first study aims to
isolate the impacts of spousal caregiving by controlling for stressors that also apply to spousal
non-caregivers.
To better understand the heterogeneous experience and individual differences among
spousal caregivers, the second study zooms in the stress process model to explore how stressors
and resources interactively affect health outcomes, which is also a proposed area of focus
identified in a previous systematic review and meta-analysis (Donnelly et al., 2015). Using an
individual-focused approach, the second study aims to better understand the heterogeneity
among spousal caregivers and the associated health outcomes, through the examination of the
nuances in the relative intensities of the co-occurrence of stressors and resources.
Besides caregivers, spousal care-recipients are affected by caregiving as well, due to their
shared life experience and immediate interpersonal context as couples (Gerstorf, Hoppmann,
Anstey, et al., 2009; Gerstorf, Hoppmann, Kadlec, et al., 2009), and the dyadic nature inherent in
caregiving (T. Pristavec, 2019; Roberto & Jarrott, 2008). Several studies reported that
caregivers’ stress and burden is correlated with recipients’ mortality, hospitalization, and
institutionalization (Kuzuya et al., 2011; Schulz et al., 2020; Stall et al., 2019). Other studies
have separately identified that caregivers’ social support, especially informal support from
family and friends, is positively correlated with care-recipient’s health and quality of life (Kelly
et al., 2017; Leung et al., 2020). The evidence, although limited, provided initial support that
caregivers’ stressors and resources could be related to care-recipients’ wellbeing. However, the
topic still remains less explored in general, and only two studies (T. Pristavec, 2019; Pristavec &
3
Luth, 2020) examined the co-occurrence of positive and negative aspects of caregiving and its
relationship with recipients’ health outcomes. The third study of the dissertation, therefore,
extends the stress process model to explore spousal care-recipients’ health outcomes.
To sum, focusing on a relatively homogeneous caregiver subgroups – spousal caregivers,
and using the stress process model as the theoretical foundation, the dissertation first compares
spousal caregivers with spousal non-caregivers in an attempt to statistically control for selection
bias and stressors that are not uniquely related to the provision of care. Second, the dissertation
pays closer attention to different individual caregiving experience based on the co-occurrence
and relative intensities of stressors and resources. Third, the dissertation explores how the
identified spousal caregiver classes could be potentially related to care-recipients’ health and
changes in health. Below is a review of the stress process model and perspectives that guided the
dissertation, as well as empirical studies relevant to the dissertation.
Theoretical Framework and Rationale
The dissertation is guided by the stress process model. Stressors that are not uniquely
associated with care provision informed the statistical control of selection bias. The nuanced
interactions between stressors and resources, and how their impacts on both caregivers’ and care-
recipients’ wellbeing characterize the new adaptation and extension of the widely utilized model.
The Stress Process Model, Selection Bias, and Comparable Samples
The stress process models have been widely used to understand individual differences in
health outcomes. Pearlin and colleagues (1981) first proposed the stress process model to
examine depression among adults aged 18 to 65 years old. There are three major conceptual
domains in the model: the sources of stress, the moderators and mediators of stress, and the
manifestations of stress. The sources of stress, or stressors, are the core of the stress process
4
model (Pearlin et al., 1990); they can arise from disruptive events, chronic life strains, and
continuous problems. Socioeconomic status and health indicators can be two sources of stressors
that are related to health outcomes (Pearlin, 2010; Pudrovska et al., 2005; Torres & O'Dell,
2016). Irrespective of care provision, the experience of health decline of oneself and/or one’s
spouse/partner can have a negative impact on a person’s health and wellbeing (Amirkhanyan &
Wolf, 2006; Bobinac et al., 2010; Bom et al., 2019).
Failure to control for stressors that are not uniquely associated with the provision of care
could partially explain the inconsistent findings on health outcomes in caregiving literature. An
early study found that compared to non-caregivers, spousal caregivers who reported strains had
significantly higher mortality risk, while there were no statistically significant differences
between caregivers as a whole in comparison to non-caregivers (Schulz & Beach, 1999). Some
subsequent research, in contrast, had the opposite conclusions: although caregivers reported
higher levels of stress than non-caregivers, they had lower mortality risks (Fredman et al., 2010;
Roth et al., 2018). The healthy caregiver hypothesis, which corresponds to the healthy worker
effect in occupational epidemiology, has been proposed to explain the phenomenon (Fredman et
al., 2015). There are two components in the hypothesis: one is that there is a self-selection bias
where healthy family members take on the caregiving role, while those with poor health
conditions either do not enter or eventually quit caring activities; therefore, caregivers have
lower mortality risks compared to non-caregivers (Fredman et al., 2010). The other component is
that factors related to caregiving, such as greater but not excessive physical activities inherent in
performing caregiving tasks, could buffer the negative effects of caregiving, and therefore help
to preserve caregivers’ physical and cognitive health (Bertrand et al., 2012).
5
In addition to the potential self-selection bias among caregivers, there could also be a
selection bias among non-caregivers in studies using non-random samples: Roth and colleagues
(2015) reviewed existing studies on the health correlates of caregiving, and noticed that an
important limitation is the recruitment of non-caregiver controls, who are always from different
sources where socially active individuals volunteer to participate in the research. Consequently,
the health effects of caregiving are confounded by both the health benefits of volunteerism
(Okun et al., 2013), and the self-selection among potentially healthier socially active individuals
(Roth et al., 2015).
Therefore, in order to examine the effects of caregiving, health status prior to becoming a
caregiver and the presence of a partner in need of care, which are two factors associated with
both spousal caregivers and spousal non-caregivers, need to be controlled for to create
comparable samples.
Stressors and Resources in the Stress Process
Pearlin and colleagues (1990) later applied the stress process model to informal
caregiving. The conceptual models of stress process among caregivers to people with
Alzheimer’s include four domains: the background and context of stress, the stressors, the
mediators of stress, and the outcomes or manifestations of stress. Under stressors, the authors
detailed some caregiving-specific stressors, including recipients’ cognitive status, ADL/IADL
dependencies, family conflict, and job-caregiving conflict. The levels of stress differ based on
the sources and intensities of stressors. Empirical studies have shown that the amount of care and
type of care are two sources of caregiving stressors (Kim et al., 2017). In addition, more hours of
care are associated with higher odds of depressive symptoms among caregivers (Hirst, 2005;
Pinquart & Sörensen, 2003a); spousal caregivers who helped with activities of daily living
6
(ADLs) were also found to have more depressive symptoms compared to those helping only with
instrumental activities of daily livings (IADLs; Burton et al., 2003).
Stressors do not uniformly affect wellbeing; resources can buffer the effects of stressors
on health outcomes (Pearlin, 2010; Teja Pristavec, 2019). Social support is a type of resource
that requires not only the involvement in a social network, but also the qualities of the relations
which involve intimate communications and trust. The dissertation focuses on social support
among all types of resources. Cohen defines social support as a social network’s provision of
psychological and instrumental resources that intend to strengthen one’s reaction to stress
(Cohen, 2004).
Another domain is the manifestations of stress. Pearlin and colleagues (1981) emphasized
the multiplicity of stress outcomes and studied depression in their paper. The outcomes of stress
in the subsequent empirical studies have been conceptualized as physical health, mental health,
distress, and giving up the provision of care (Campbell et al., 2008; Walls & Whitbeck, 2012).
Results from meta-analysis have consistently shown that caregiving stressors are associated with
worse physical (Pinquart & Sörensen, 2007; Vitaliano et al., 2003) and mental health (Pinquart
& Sörensen, 2003a, 2003b). Poor physical health is an accumulation of strains and other risk
factors associated with stress over time, and self-rated health is a commonly used independent
measure of survival and physical health among older adults. Depression is a widely discussed
outcome under the realm of mental health (Mitchell & Knowlton, 2012).
For factors under the domain of background and context, the commonly included ones in
existing research are age, gender, race/ethnicity, and socioeconomic status. As caregivers age,
they may encounter more stressors, and have more difficulties providing ADL assistance both
because of the recipients’ needs and caregivers’ own declined health (Wharton & Zivin, 2017;
7
Wolff & Kasper, 2006). For gender, women and men could be at different risks of various types
of stressors and possess different resources, either due to the structural contexts and unequal
distributions of rewards and opportunities they are in (Pearlin et al., 1990), or because women
and men respond to caregiving differently where women feel more obliged and responsible to
care (Calasanti, 2010; Mc Donnell & Ryan, 2011). Caregivers with different racial/ethnic
backgrounds experience and appraise caregiving differently as well, both because of caregiving
stressors themselves, and the different distributions of stressors that are not directly related to
caregiving, such as poverty, and resources such as uneven access to and utilizations of formal
support (Apesoa-Varano et al., 2015; Capistrant, 2016; R. Liu et al., 2021; Pinquart & Sörensen,
2005). For socioeconomic status, income and education are two common proxies. Number of
chronic conditions are also included as multimorbidity is common in old age (Quiñones et al.,
2019).
There are many variations of the stress process model up to date, which is based on
Pearlin’s sociological perspective (Pearlin et al., 1981), and incorporated psychological (Lazarus
& Folkman, 1984) and physiological (McEwen, 1998) perspectives. The model asserts that when
caregiving demands exceed one’s psychological or social resources to cope, immunological and
neuroendocrine disruptions may occur, which increase one’s risk of health decline (Roth et al.,
2015). However, only two studies (Pristavec, 2019; Sung et al., 2021) have investigated the
relative intensities of stressors and resources, and how they interactively affect health outcomes.
Studies including both stressors and resources but examining them in separation do not enable
the exploration of the relative intensities, and can only reveal the relative impact of one aspect
when the other is “controlled for”. The variable-focused method cannot account for the
variations in the combinations of stressors and resources that caregivers are experiencing and
8
possessing. Therefore, a study that explores the nuanced interactions of stressors and resources is
warranted.
Relationship between Care-Recipients’ Wellbeing and the Co-Occurrence of Caregivers’
Stressors and Resources
Besides caregivers, care-recipients could also be impacted by caregiving, since the
caregiving relationship is dyadic by definition (T. Pristavec, 2019; Roberto & Jarrott, 2008).
However, no studies have investigated the relationship between care-recipients’ wellbeing and
the co-occurrent stressors and resources of the caregivers. Recent studies, although relatively
scarce, have examined the impacts of different caregiving facets separately, including caregiving
stressors and caregivers’ resources, on health and wellbeing among care-recipients. Caregivers’
stressors have found to be correlated with care-recipients’ mortality (Schulz et al., 2020),
hospitalization (Kuzuya et al., 2011), and institutionalization (Stall et al., 2019) in longitudinal
studies; cross-sectional studies have also reported the correlation between caregivers’ reported
burden and care-recipients’ depression (Ejem et al., 2018; Ejem et al., 2015) and loneliness
(Iecovich, 2014). Caregivers’ resources have also been found to be indirectly correlated with
care-recipients’ health and quality of life (Gellert et al., 2018; Kelly et al., 2017; Leung et al.,
2020). Only two studies (T. Pristavec, 2019; Pristavec & Luth, 2020) have examined the positive
and negative aspects of caregiving simultaneously instead of in isolation. However, they either
were subject to omitted variable bias (T. Pristavec, 2019) or utilized simplistic binary
classifications of caregiving burden and benefits (Pristavec & Luth, 2020). A study that
examines the relationship between care-recipients’ wellbeing and the co-occurrence of
caregivers’ stressors and resources is brought to the agenda given the existing evidence and
research gaps.
9
Limitations of Existing Studies
The stress process model provides a framework to understand stressors and resources
related to caregiving, and how they synthetically affect the manifestation of health outcomes
across multiple domains. However, there are several limitations in existing studies. First, in
caregiving research, spousal caregivers and other family caregivers are often studied together
and treated as a monolithic sub-category of caregivers, which could have overshadowed the
higher levels of stressors spousal caregivers experience because of their small sample size
relative to adult child caregivers, or when the comparison group is parent caregivers. Second,
there is only a limited number of quantitative studies using a person-centered approach to
identify caregiver subgroups based on the relative intensities of stressors and resources. In
addition, no studies identified have utilized tangible and objective measures of resources that
caregivers possess. Third, few studies have explored the impact of caregiving on care-recipients’
wellbeing. Existing studies also have a heavy reliance on convenience samples, and many were
conducted outside of the United States, the conclusions of which could be less generalizable to
the U.S. population and context.
Specific Aims
The dissertation investigates the health effects of spousal caregiving to address the issue
of selection bias. It takes a person-centered approach to explore spousal caregiver latent classes
based on relative intensities of co-occurrence of stressors and resources – two major domains
outlined in the stress process model. Whether and how spousal caregivers’ latent classes are
related to care-recipients’ health outcomes are also explored. There are three aims for the study:
1. To examine the health effects of spousal caregiving using coarsened exact matching
between spousal caregivers and spousal non-caregivers;
10
2. To identify distinguishable spousal caregiver latent classes based on the co-occurrence
and relative intensities of stressors and resources, the spousal caregiver profiles in each
class, and the related health outcomes; and
3. To investigate how spousal care-recipients’ health and changes in health outcomes are
correlated with spousal caregivers’ latent class memberships, based on the classes
identified in Aim 2.
Figure 1.1 provides an overall preview of the framework guiding the dissertation. Chapter
Two through Chapter Four address each of the three aims correspondingly.
Contributions and Implications
The dissertation has the following significance and contributions: First, it improves the
understanding of health impacts of caregiving on spousal caregivers by addressing selection bias,
which has implications for public health planning at the macro level. Second, it explores another
variation of the stress process model to examine the nuances in the relative intensities and co-
occurrences of stressors and resources. It could further validate the adaptation of the stress
process model, and highlight the interactive process that uniquely characterizes an individual’s
experience. Third, it goes beyond the original stress process model to examine the health impacts
on spousal care-recipients, which can inform specific practice and intervention recommendations
to social workers, service providers, and program officers at the micro- and meso-level. The use
of nationally representative samples could also increase the generalizability of results.
For policy and practice, exiting state and federal policies have addressed some of the
unmet needs among caregivers such as paid and unpaid family leaves (Dawson et al., 2020).
However, most caregivers still remain invisible in the healthcare system despite the growing
11
number and diversity (Reyes et al., 2021). The results of the dissertation could offer insights on
not only which dimensions of health outcomes to be focused on among spousal caregivers, but
also which specific indicators of stressors and/or resources to be targeted at in order to achieve
the desired outcomes. The dissertation also corresponds to National Institute of Aging (NIA)’s
emphasis on caregiving, diversity, and the consequences of aging for society, and addresses one
of the research priorities from the Caregiving Summit to target the heterogeneity and trajectories
of caregiving (Hepburn & Siegel, 2020).
12
Figures
Figure 1.1. Conceptual Framework for the Dissertation
13
Chapter 2. Impacts of Caregiving on Wellbeing among Spousal Caregivers to Older Adults
in the United States: A Coarsened Exact Matching Analysis
Introduction
The aging U.S. population brings an increase in the demand for long-term care (LTC):
7.9 million (15.9%) of older adults aged 65 years or older needed LTC in 2018 (AARP, 2019a),
and the number is anticipated to more than double with about 19 million of older adults needing
LTC by 2050 (Spetz et al., 2015). About two-thirds of LTC is provided by family and other
unpaid caregivers, who contribute an estimated $470 billions of labor force annually (AARP,
2019b). Among family caregivers, spousal caregivers constitute the second largest informal
caregiver group behind children, making up 12% of the caregiver population (AARP, 2019b). In
addition, the prevalence of spousal caregiving increases with age: it is estimated that 11% of 50-
to 64-year-old, 21% of 65- to 74-year-old, and 40% of 75+ year-old caregivers are caring for a
spouse/partner (AARP, 2019b). Moreover, spousal caregivers are oftentimes the solo caregivers,
especially at the last years of life, making them more vulnerable to negative health outcomes due
to the increased caregiving burden and their own advanced age (Chen et al., 2020; Ornstein et al.,
2019). As the population ages, the demand for spousal caregiving is expected to grow;
understanding the health impact of caregiving is pivotal in providing support to spousal
caregivers and in LTC design.
The negative health consequences of spousal caregiving have been widely documented.
An early systematic review and meta-analysis (Pinquart & Sörensen, 2003a) revealed that
compared to adult-child caregivers, spousal caregivers reported higher levels of burden and
worse physical health, even after controlling for care-recipients’ dementia status, disabilities, and
caregivers’ age. The results were confirmed by a meta-analysis conducted later (Pinquart &
14
Sörensen, 2011) comparing spousal caregivers with adult children and adult children-in-laws,
which found that spousal caregivers had higher levels of depressive symptoms and psychological
distress, and more financial and physical burden. Cross-sectional (Dunkle et al., 2014; Penning
& Wu, 2016; Polenick et al., 2020) and longitudinal (Kaufman et al., 2019; C. Liu et al., 2021;
Oldenkamp et al., 2016; Rafnsson et al., 2017) studies conducted recently have also reported
consistent findings in mental health decline associated with spousal caregiving, while
conclusions on spousal caregiving’s relationship with physical health are less conclusive.
Despite the documented negative health effects among spousal caregiver, more recent
studies have suggested increasing evidence of beneficial results from caregiving, such as
increased spiritual health including a sense of rewarding and satisfaction (Roth et al., 2015),
increased interpersonal relationships such as feelings of mutuality in spousal relationships and
increased family cohesion (Yu et al., 2018), and lower mortality risk (Roth et al., 2018). A recent
meta-analysis of longitudinal population-based studies covered research in six countries, and
results indicated that the overall combined effect of caregiving on mortality was 16% lower in
favor of caregivers compared to non-caregivers (Mehri et al., 2021).
Selection bias could be a possible explanation of the inconsistent findings on health
consequences of caregiving. The healthy caregiver hypothesis, which corresponds to the healthy
worker effect in occupational epidemiology, has been proposed to explain the lower mortality
among caregivers (Fredman et al., 2015). The hypothesis addresses the potential confounding
mechanisms in self-selection bias and endogeneity, and posits that people need to maintain a
minimum level of health to be able to perform caregiving tasks, while individuals with poor
health condition either do not enter or eventually quit caring activities (Fredman et al., 2010;
Fredman et al., 2015). Selection biases could also exist among non-caregivers and in non-random
15
sample studies: Roth and colleagues (Roth et al., 2015) reviewed existing studies on the health
correlates of caregiving, and noticed that an important limitation is the recruitment of non-
caregivers controls, who are oftentimes from different sources where socially active individuals
volunteer to participate in the research. Consequently, the health effects of caregiving are
confounded by both the health benefits of volunteerism (Okun et al., 2013), and the potentially
healthier socially active individuals (Roth et al., 2015).
Bom and colleagues (2019) reviewed studies trying to estimate the causal effects of
caregiving, and found evidence of negative health impacts of caregiving on physical and mental
health. However, among the fifteen included studies, only four were conducted in the U.S., none
of which had a specific focus on spousal caregivers to older adults. A longitudinal study
conducted recently (Chen et al., 2020) assessing the causal effects of caregiving found that
spousal caregivers are at higher odds of depressive disorders with the partners’ dementia onset,
but the samples included spousal caregivers to persons with dementia only instead of all spousal
caregivers.
To sum, the impacts of caregiving on health remain inconclusive, and one of the
limitations in existing studies is the lack of specific focus on spousal caregivers. Using a
nationally representative data, the current study aims to control for selection bias in spousal
caregiving to examine the impacts of caregiving on spousal caregivers’ health through coarsened
exact matching (CEM). Similar to a previous study (Schmitz & Westphal, 2015) using
propensity score matching (PSM) to examine the health impacts of informal caregiving among
female caregivers in Germany, the variables selected for matching are classified into three
categories: care needs, the willingness to provide care, and the ability to provide care. Details on
CEM and selected variables are discussed in the methods section.
16
Methods
Data
The data used is from the Health and Retirement Study (HRS). HRS is a nationally
representative longitudinal survey of noninstitutionalized individuals aged 50 and above, and
surveys more than 22,000 people in the U.S. every two years. If the participant (hereafter
partner) is married or living with a spouse/partner, the spouse/partner (hereafter spouse) is
recruited into the study and surveyed as well, regardless of their age. HRS is supported by the
National Institute on Aging (NIA U01AG009740) and the Social Security Administration, and is
undertaken by the University of Michigan. The primary sample is selected through a multistage,
area-clustered probability design with an 80% overall response rate. It oversamples African
Americans, Hispanics, and Floridians.
The main dataset utilized is the RAND HRS Longitudinal File (Bugliari et al., 2021),
which is a cleaned and streamlined data product from Core and Exit Interviews of the HRS that
is easy to use. Since RAND HRS does not provide information on caregiving, raw data from
Section G: Functional Limitations and Helpers of the original HRS files are merged 1:1 on
participant’s ID to get additional information on spousal caregiver- and caregiving- related
variables. For the current study, data from Wave 8 (collected in 2006) to Wave 13 (collected in
2016) are utilized to ensure the consistency in the measurements of caregiving-related variables.
Sample
Sample creation includes three steps. The first step is the identification of spouses.
Spouses are included if (1) they have the same partner for at least two consecutive waves, (2)
both spouses and partners are alive in the two consecutive waves, and (3) the partners are at least
60 years old in the first included wave. Spouses are excluded if they are of the same sex (n=234;
17
0.92%) as the partners, or if spouses are of races/ethnicities other than non-Hispanic white, non-
Hispanic black, or Hispanics, due to the relatively small sample size (n=1,051; 4.12%).
The second step is to identify spouses who are also caregivers (hereafter spousal
caregivers). Spousal caregivers are defined as those who helped with at least one activity of daily
living (Gerstorf, Hoppmann, Kadlec, et al.) or instrumental activity of daily living (IADL) items
to the partners, and are identified by the partner as spouse/ partner. To decrease heterogeneity
among caregivers due to their previous caregiving experience, only new spousal caregivers will
be included. New spousal caregivers are identified by including those who are providing care to
their partners at the current wave but not at any of the previous waves. That is, when there are
repetitive observations across the included waves, the earliest data are utilized.
The third step is to identify spousal non-caregivers for the matching. The selection of
variables used for matching is informed by a previous study (Schmitz & Westphal, 2015), and
there are three areas that affect the transition to caregiving: (1) care needs which indicate the
presence of a family member in need of care, (2) the willingness to provide care including
socioeconomic characteristics that affect the individual’s choice, and (3) the ability to provide
care. Spousal non-caregivers are selected at each wave to serve as matched comparisons. Spousal
non-caregivers are defined as those who never had spousal caregiving experience in any of the
current or previously included waves. That is, spouses who are once a caregiver but become a
non-caregiver in subsequent waves are not eligible for matching. Multiple observations within
the same individual across different waves are allowed for spousal non-caregivers.
Data across waves are then pooled together to generate the final sample. The sample
selection criteria yield a panel dataset with 203,162 person-wave observations from 51,623
unique individuals. Among these, 3,577 unique individuals are new spousal caregivers.
18
Measures
Health outcomes. Health outcomes include physical health, mental health, and cognitive
health. Health outcomes at the second included wave are the outcomes of interests (dependent
variables). Physical health is measured by self-rated health on a 1-5-point scale with 1 indicating
“excellent” and 5 indicating “poor” health. For easier interpretation, the variable is reverse coded
with higher score indicating better health. Mental health is conceptualized as depressive
symptoms, which are measured by Center for Epidemiologic Studies – Depression Scale (CES-D
8), an eight-item scale indicating negative effects and somatic complaints in the preceding week
(Turvey et al., 1999). The scale is an additive score of dichotomous “yes/no” responses ranging
from 0 (no symptoms) to 8 (all symptoms reported). The selection of the eight-item version was
based on factor analyses, and the version has shown high internal consistency (Cronbach’s alpha
is 0.83 for Wave 2 and 0.81 for Wave 3) and validity in HRS (Steffick et al., 2000). A CES-D 8
score of 4 and above indicate high depressive symptomatology (Quiñones et al., 2016; Steffick et
al., 2000; Turvey et al., 2009). The measure of cognitive health is based on the Telephone
Interview for Cognitive Status (TICS), a validated cognitive screening instrument designed to be
used in telephone surveys (Brandt et al., 1988). All self-respondents were administered items in
the immediate and delayed word recall (range: 0-20), a serial 7’s test (range: 0-5), and a
backwards counting test (range: 0-2), summing up to a 27-point scale. According to Langa-Weir
Classifications (Crimmins et al., 2011), the three categories are: normal (12-27), cognitive
impairment no dementia (CIND; 7-11), and dementia (0-6). The cut-off points have been
validated against diagnostic information based on the Aging, Demographics and Memory Study
(ADAMS; Crimmins et al., 2011).
19
Matching variables. Partner’s functional limitations (ADL or IADL) are used as proxies
for care needs (Beach & Schulz, 2017). ADL ranges from 0 to 5 indicating whether the partner
needs help in bathing, eating, dressing, walking across a room, and getting in or out of bed.
IADL ranges from 0 to 3 indicating whether the partner needs help in using a telephone, taking
medication, and handling money. Spouse’s age, gender, education, and income are used to
represent willingness to provide care. Age is the spouse’s age in years at the time of the
interview. Gender is a dichotomous variable indicating male (0) or female (1). Education is a
five-category variable ranging from 1 “less than high school” to 5 “college and above”. Income
is measured by household income over the last calendar year. Income is adjusted for inflation (in
2010 $) based on the medical services portion of the Consumer Price Index (United States
Department of Labor Statistics) for the pooled data, and is transformed into logarithmic form in
regression analyses. Spouse’s health status at baseline, including physical health, mental health,
and cognitive health, is used as proxy for the ability to provide care. Other variables that could
be potentially correlated with the transition into caregiving are controlled for in the regression
analyses.
Control variables. Spouse’s race/ethnicity, partner’s age, both spouse’s and partner’s
number of comorbidities, length of current marriage, and number of living children are included
as control variables as informed by a clinical review on caregiving burden (Adelman et al.,
2014). Race/ethnicity is a three-category variable including non-Hispanic White, non-Hispanic
Black, and Hispanics; non-Hispanic White is the reference group. Partner’s age is measured in
years at the time of the interview. Comorbidities are the count of eight types of doctor-diagnosed
health problems, including high blood pressure, diabetes, cancer, lung disease, heart disease,
stroke, psychiatric problems, and arthritis. Length of current marriage is the length of marriage in
20
years at the time of the interview. Number of living children is a count variable indicating the
number of living children at the time of the interview.
Analysis
To account for observable differences among spousal caregiving dyads and spousal non-
caregiving dyads that could be associated with both the transition into caregiving and caregivers’
health outcomes, coarsened exact matching (CEM; Blackwell et al., 2009) is utilized to match
spousal caregivers and spousal non-caregivers at each wave, allowing for multiple matches per
spousal caregiver.
The basic algorithm of CEM includes three steps. The first is to temporarily coarsen each
variable into groups. Second, exact matching is employed on the coarsened data. Third, the final
matched data are reserved after matching and removing the coarsened data. CEM has two
advantages over other matching methods such as propensity score matching: the first is that there
is no need for ex-post balance checking since the maximum acceptable imbalance is pre-
determined by ex-ante user choice (Iacus et al., 2012); second, CEM is less sensitive to model
specification (King et al., 2011; Stuart, 2010).
For the current study, respondents’ functional limitations (ADL and IADL) are coarsened
into a binary yes (1) versus no (0) variable for matching, with “yes” indicating at least one types
of ADL or IADL limitations. Caregiver’s education is coarsened into two categories indicating
“high school and below” versus “some college and above”. Household income is coarsened into
quintiles. For health outcomes, self-rated health is coarsened into two categories indicating
“poor/fair” versus “good/very good/excellent”, depressive symptoms are coarsened into a binary
yes (1) versus no (0) with those scoring 4 and above in CES-D 8 coded as “yes”, and cognitive
21
functioning is coarsened into three categories indicating “normal” (12-27 in TICS), “CIND” (7-
11 in TICS), and “dementia” (0-6 in TICS).
Standardized differences in each variable used for matching are checked between
matched spousal caregivers and spousal non-caregivers at each wave before and after matching.
Standardized differences that are less than 10% indicate adequate balance (Austin, 2009; Garrido
et al., 2014). Differences between matched and unmatched spousal caregivers are also examined
to check for the representativeness of the matched samples. After checking for balance, total
matched samples are pooled across waves. The average treatment effect on the treated (ATT) is
estimated through ordinary least square regressions. Variables used for matching and all control
variables are included in regression analyses to control for remaining imbalance. All analyses are
adjusted for CEM weights, and standard errors are clustered by spouse’s ID to account for
autocorrelation.
Results
CEM excludes spousal caregivers with missing values in variables used for matching
(n=227; 6.3%). Among the remaining 3,350 spousal caregivers, 2,714 (81.01%) are matched
with spousal non-caregivers. Compared to matched spousal caregivers, unmatched spousal
caregivers are more likely to have worse health status at baseline across all three domains, more
likely to be non-Hispanic Black or Hispanics, have lower levels of education, and lower
household income. Both the unmatched spousal caregivers and their care-recipients are having
more comorbidities and more functional limitations. Counterintuitively, most of them are
experiencing better health outcomes in the subsequent wave, represented by higher levels of self-
rated health, less depressive symptoms, and an increase in cognitive functioning. Detailed
22
comparison between matched and unmatched spousal caregivers by cohort can be found in
appendix.
Comparison between matched spousal caregivers and matched spousal non-caregivers are
examined to check for the quality of matching. Table 2.1 displays the descriptive statistics and
standardized differences between spousal non-caregivers and spousal caregivers for variables
used in CEM matching in the coarsened format. Before CEM matching, spousal non-caregivers
and spousal caregivers differ in both care needs and willingness to provide care: in general,
spousal non-caregivers have partners with fewer functional limitations, they are slightly younger,
more likely to be male, have higher levels of educational attainment, and are in better financial
status. The abilities to provide care differ between spousal caregivers and spousal non-caregivers
as well, but the differences are relatively small compared to the other two domains. After CEM
matching, the variables are adjusted to approximate spousal caregivers’ level with no statistically
significant differences in any of the categories used for matching.
The five waves of matched data are pooled together to create a panel data. Table 2.2
displays the characteristics of matched sample. Compared to spousal non-caregivers, spousal
caregivers have higher levels of depressive symptoms at baseline, higher levels of depressive
symptoms in the subsequent time, and the degree of increase in depressive symptoms are also
larger among spousal caregivers. In addition, spousal caregivers have partners of older age, more
comorbidities, and more functional limitations. All the other characteristics in both health
outcomes and sociodemographic characteristics do not differ statistically significantly between
the two groups.
23
Regression analyses are performed to examine the health outcomes associated with
spousal caregiving. For each health outcome, stepwise regression analyses are performed with
the first set including only “being a caregiver” as the independent variable, and the second set
controlling for the remaining differences between groups. Before regression analyses,
multicollinearity is checked, and the variance inflation factor (VIF) is 1.52, indicating that
multicollinearity would not be an issue. Results are displayed in Table 2.3.
Regression results indicate that being a spousal caregiver is not correlated with self-rated
health in the subsequent time. However, all indicators of the spouses’ health are correlated with
their self-rated health in the subsequent time. A one-unit increase in self-rated health at baseline,
depressive symptoms, and cognitive functioning is associated with a 0.56-unit increase, a 0.03-
unit decrease, and a 0.01-unit increase in self-rated health in the following wave, respectively. In
addition, an additional type of comorbidities (B=-0.09, SE=0.01), an additional type of ADL
(B=-0.09, SE=0.03) and IADL (B=-0.14, SE=0.06) limitations are associated with decreased
self-rated health at the second wave.
For depressive symptoms, compared to spousal non-caregivers, spousal caregivers are
experiencing a 0.27-unit increase on average in the subsequent wave. Spouses’ own health status
at baseline, including better self-rated health (B=-0.20, SE=0.04) and better cognitive
functioning (B=-0.02, SE=0.01) are both protective factors against depressive symptoms,
whereas higher levels of depressive symptoms at baseline (B=0.42, SE=0.04), an additional type
of comorbidities (B=0.10, SE=0.03), and an additional type of ADL limitation (B=0.17;
SE=0.07) are all risk factors of worsening depressive symptoms. In addition, female on average
scores 0.28-unit higher in depressive symptoms compared to their male counterparts. Compared
24
to non-Hispanic Whites, non-Hispanic Black (B=0.30, SE=0.13) and Hispanics (B=0.46,
SE=0.15) are both more likely to experience an increase in depressive symptoms.
Being a spousal caregiver is not correlated with cognitive functioning in the subsequent
time either. Higher levels of self-rated health (B=0.20, SE=0.08) and cognitive functioning
(B=0.48, SE=0.02) at baseline are both associated with better cognitive functioning in the second
wave. Older ages in both spouses themselves (B=-0.06, SE=0.01) and their partners (B=-0.04,
SE=0.01), as well as an additional type of IADL limitations among spouses (B=-1.29, SE=0.37)
are negatively associated with cognitive functioning at the second wave. For sociodemographic
characteristics, being a female (B=0.48, SE=0.02) and having higher educational attainment
(B=0.92, SE=0.14) are associated with better cognition. Compared to non-Hispanic Whites, non-
Hispanic Black (B=-1.38, SE=0.23) and Hispanics (B=-0.81, SE=0.24) are both experiencing
worse cognitive functioning.
In addition, neither partners’ health status, including their number of comorbidities,
functional limitations, nor household-level characteristics, including household income, length of
marriage, and number of living children, are correlated with spouses’ health outcomes in the
subsequent time.
Discussion
Spousal caregivers to older adults are particularly prone to negative health outcomes
compared to other family caregivers, due to the fact that they are also dealing with their own
aging and health issues (Haley et al., 2000; Khalaila & Cohen, 2016; Lavela & Ather, 2010;
Pinquart & Sörensen, 2011; Schulz & Sherwood, 2008). Estimating the health effects of
caregiving is relevant in isolating the health decline that is potentially caused by caregiving,
rather than aging and age-related health issues, which is crucial in both the design of
25
interventions targeted at spousal caregivers, and in computing the cost and benefits of informal
caregiving.
Using longitudinal data from HRS, we partially controlled for endogeneity in self-
selection into caregiving through coarsened exact matching. Results indicate that being a spousal
caregiver is correlated with higher levels of depressive symptoms, but not differences in self-
rated health or cognitive functioning. Previous research has a consensus on the higher propensity
to depressive symptoms among spousal caregivers to persons with dementia (Adams, 2008;
Chen et al., 2020; Joling et al., 2015), cancer (Goldzweig et al., 2019; Li et al., 2013; Rhee et al.,
2008), or ADL needs (Dunkle et al., 2014). The current study defined spousal caregivers as those
providing care for at least one of ADL or IADL needs, and further confirmed the increased
depressive symptoms among spousal caregivers in general. In addition, the increased depressive
symptoms were found to be independent of care-recipients’ numbers of chronic conditions or
level of functional limitations. Only spousal caregivers’ own health at baseline, including their
self-rated health, depressive symptoms, cognitive functioning, number of chronic conditions, and
number of ADL limitations are correlated with depressive symptoms in the subsequent time,
indicating that older adults, regardless of whether they are caregivers or not, are more prone to
depressive symptoms if they are of worse health at baseline. Results also indicate that compared
to male, female spousal caregivers are more likely to have higher depressive symptoms, which is
similar to previous research findings (Pillemer et al., 2018; Pinquart & Sörensen, 2006b; Yee &
Schulz, 2000). One of the explanations is that female and male respond differently to caregiving,
where female feel more obliged to care (Chiao et al., 2015; Xiong et al., 2020), more connected
to care partners (Beeber & Zimmerman, 2012), and are more likely to seek emotional coping
methods (Baker & Robertson, 2008; Robinson et al., 2014; Zodikoff, 2007). In addition, our
26
results indicate that compared to non-Hispanic White, non-Hispanic Black and Hispanics are
more likely to have higher levels of depressive symptoms. A recent systematic review on
dementia caregivers found that compared to non-Hispanic White, Black caregivers reported
better psychological wellbeing (Liu et al., 2020), while another population-based study found no
racial differences in caregivers’ health (Badana et al., 2017). However, neither of the two studies
has controlled for caregivers’ health at baseline. Our results are similar to a study controlling for
selection bias among dementia caregivers, where they found poor mental health among Black
and Hispanic caregivers are more common than White caregivers (Chen et al., 2020). This may
be because that racial/ethnic minorities are less likely to seek mental health assistance compared
to White caregivers (Stockdale et al., 2008), and may echo the point proposed in a recent
research synthesis that racial/ethnic minorities are more likely to suffer from chronic and
prolonged depression compared to Whites (Rawlings & Bains, 2020).
Being a spousal caregiver is not correlated with decreased self-rated health, which is
different from a previous cross-sectional study indicating caregivers report better self-rated
health compared to non-caregivers (Kim et al., 2017). Healthy caregiver hypothesis could be a
plausible explanation, since the above-mentioned cross-sectional study did not control for self-
rated health at baseline. Another possibility is the time interval between waves, which is two
years, might be too short to detect any changes in self-rated health. Similar to the conclusions on
depressive symptoms, spousal caregivers’ own health status rather than care-recipients’ health is
correlated with their subsequent self-rated health.
Nor is being a spousal caregiver correlated with changes in cognitive functioning in the
current study. A previous study comparing spousal caregivers with spousal non-caregivers to
persons with dementia found that spousal caregivers performed worse in memory recovery and
27
information processing speed, but better in cognitive flexibility than non-caregiving controls (de
Vugt et al., 2006). Another study found that spousal caregivers to persons with dementia
performed worse in cognitive processing speed decline than non-caregiving controls, and the
decline is mediated by depression (Vitaliano, 2010). There are four plausible explanations for the
insignificant results. The first explanation is the differences between spousal caregivers in
general versus spousal caregivers to persons with dementia specifically, whose caregiving
activities are more stressful and who often report higher levels of burden and distress (Kasper et
al., 2015). Therefore, the significance in cognitive functioning among spousal caregivers to
persons with dementia may not be as pronounced in the general spousal caregiver population.
The second reason might be the measurement of TICS, which is a global assessment without the
specification of each domain of cognition, and the positive and negative aspect-specific results
may have cancelled out each other when combined. The third point is similar to the point raised
in the previous study (Vitaliano, 2010), that a great portion of cognition decline could be
explained by depression. At last, the two-year time interval may be too short to detect
statistically significant cognition decline among spousal caregivers. Our results also demonstrate
that female versus male, White versus racial/ethnic minorities, those with higher level of
education, of younger age, and fewer IADL limitations are more resilient to cognitive decline,
which is consistent with previous studies (Alley et al., 2007; Jekel et al., 2015; McCarrey et al.,
2016; Weuve et al., 2018).
There are several limitations in the study. First, selection bias is not completely
controlled for. Even if the study partially controlled for endogeneity and addressed the selection
bias in transition into caregiving as proposed by the healthy caregiver hypothesis, spousal
caregivers still have higher levels of depressive symptoms at baseline, and are taking care of
28
partners of older age and with more chronic conditions, and functional limitations. An
assumption is that spousal caregivers and spousal non-caregivers do differ systematically,
especially in terms of care needs, i.e., a partner in need of caregiving. Moreover, selection bias
on unobservable characteristics could not be controlled for and could potentially bias the analysis
results. Second, due to the data collection method, we could only assess spouses who are not
caregivers at baseline but report being a caregiver in the subsequent wave; spousal caregivers
may have experienced transition in and/or out of the caregiving role within the survey periods,
and the lengths of caregiving can vary from months to almost two years, which caused the
heterogeneity in the spousal caregiver sample. Another limitation pertaining to the
inclusion/exclusion criteria is that if spousal caregivers and/or care-recipients are deceased
between waves, the dyads are excluded from the analysis, whereas this subgroup could be those
of worst health outcomes. The third limitation is that two years might be too short to assess
health decline, especially cognitive decline among cognitively intact spousal caregivers. At last,
caregivers with high level of cognitive decline or depressive symptoms are not included in the
analysis due to missing data on these items.
For future research, caregivers transitioning out of caregiving could be included as well
to depict a full picture of the effects of caregiving. In addition, the time intervals could be
extended to evaluate both short-term and long-term health effects of caregiving. At last, as the
current study suggests systematical differences between matched spousal caregivers and
unmatched spousal caregivers, the heterogeneity within spousal caregivers is worth exploring to
identify whether there are subgroups who are experiencing different patterns of changes in
health.
29
The study highlights the importance of addressing mental health among spousal
caregivers, as depressive symptoms are more prevalent among them compared to spousal non-
caregiver counterparts. Recent studies have found that the Affordable Care Act’s Medicaid
expansion (Costa-Font et al., 2021) and the enrollment in Program of Comprehensive Assistance
for Family Caregivers (Smith et al., 2019) are associated with reduced depressive symptoms
among caregivers. As caregivers are the backbone of LTC, further assistance could be provided
to caregivers to improve their mental health outcomes, including interventions directly related to
mental health such as counseling and social support, and services through addressing other
factors that are indirectly related to caregivers’ depressive symptoms, including the provision of
respite care. As caregivers are more likely to delay their own care compared to non-caregivers
(Slaboda et al., 2021), healthcare in general should also be better integrated and more
streamlined to ensure easier access for caregivers in general and spousal caregivers in specific.
30
Tables
Table 2.1. Standardized differences in variables used for CEM before and after matching by
cohort, weight adjusted
SNCGs SCGs
Std.
Diff.
SNCGs SCGs
Std.
Diff.
Cohort
Variables
Mean
(SD)
Mean (SD)
Mean
(SD)
Mean
(SD)
1 Functional
limitations
0.09
(0.29)
0.29 (0.46) 0.54
0.21
(0.40)
0.21
(0.40)
0.00
Spouse’s age
69.14
(8.60)
71.52
(8.20)
0.28
70.89
(7.52)
71.00
(7.59)
0.00
Spouse’s gender
0.52
(0.50)
0.53 (0.50) 0.01
0.54
(0.50)
0.54
(0.50)
0.00
Spouse’s
education
0.45
(0.50)
0.31 (0.46) -0.29
0.32
(0.47)
0.32
(0.47)
0.00
Household
income
3.78
(1.16)
3.36 (1.16) -0.37
3.47
(1.10)
3.47
(1.10)
0.00
Spouse’s SRH
0.77
(0.42)
0.68 (0.47) -0.20
0.74
(0.44)
0.74
(0.44)
0.00
Spouse’s CES-D
8
0.10
(0.30)
0.15 (0.35) 0.15
0.08
(0.27)
0.08
(0.27)
0.00
Spouse’s TICS
2.78
(0.49)
2.71 (0.53) -0.15
2.81
(0.44)
2.81
(0.44)
0.00
2 Functional
limitations
0.08
(0.28)
0.30 (0.46) 0.58
0.21
(0.41)
0.21
(0.41)
0.00
Spouse’s age
69.11
(8.82)
71.77
(8.54)
0.31
71.63
(8.17)
71.83
(8.29)
0.00
Spouse’s gender
0.52
(0.50)
0.59 (0.49) 0.15
0.59
(0.49)
0.59
(0.49)
0.00
Spouse’s
education
0.48
(0.50)
0.34 (0.47) -0.28
0.36
(0.48)
0.36
(0.48)
0.00
Household
income
3.81
(1.14)
3.26 (1.22) -0.46
3.41
(1.16)
3.41
(1.16)
0.00
Spouse’s SRH
0.76
(0.43)
0.73 (0.44) -0.07
0.77
(0.42)
0.77
(0.42)
0.00
Spouse’s CES-D
8
0.09
(0.29)
0.11 (0.31) 0.05
0.08
(0.27)
0.08
(0.27)
0.00
Spouse’s TICS
2.78
(0.50)
2.78 (0.49) -0.02
2.83
(0.43)
2.83
(0.43)
0.00
3 Functional
limitations
0.10
(0.29)
0.30 (0.46) 0.54
0.20
(0.40)
0.20
(0.40)
0.00
Spouse’s age
68.84
(9.29)
70.95
(9.69)
0.22
71.47
(9.06)
71.52
(9.23)
0.00
Spouse’s gender
0.52
(0.50)
0.60 (0.49) 0.16
0.64
(0.48)
0.64
(0.48)
0.00
31
Spouse’s
education
0.50
(0.50)
0.39 (0.49) -0.23
0.40
(0.49)
0.40
(0.49)
0.00
Household
income
3.84
(1.15)
3.46 (1.14) -0.33
3.54
(1.10)
3.54
(1.10)
0.00
Spouse’s SRH
0.78
(0.41)
0.73 (0.44) -0,12
0.80
(0.40)
0.80
(0.40)
0.00
Spouse’s CES-D
8
0.09
(0.29)
0.11 (0.31) 0.05
0.05
(0.22)
0.05
(0.22)
0.00
Spouse’s TICS
2.77
(0.50)
2.77 (0.46) -0.01
2.83
(0.40)
2.83
(0.40)
0.00
4 Functional
limitations
0.08
(0.28)
0.30 (0.46) 0.56
0.18
(0.39)
0.18
(0.39)
0.00
Spouse’s age
68.37
(9.31)
70.31
(10.14)
0.20
70.62
(8.98)
70.58
(9.10)
0.00
Spouse’s gender
0.52
(0.50)
0.54 (0.50) 0.03
0.53
(0.50)
0.53
(0.50)
0.00
Spouse’s
education
0.52
(0.50)
0.40 (0.49) -0.24
0.41
(0.49)
0.41
(0.49)
0.00
Household
income
3.85
(1.15)
3.38 (1.28) -0.39
3.53
(1.21)
3.53
(1.21)
0.00
Spouse’s SRH
0.78
(0.41)
0.75 (0.44) -0.08
0.78
(0.42)
0.78
(0.42)
0.00
Spouse’s CES-D
8
0.09
(0.29)
0.09 (0.28) -0.01
0.05
(0.21)
0.05
(0.21)
0.00
Spouse’s TICS
2.78
(0.50)
2.71 (0.55) -0.12
2.79
(0.47)
2.79
(0.47)
0.00
5 Functional
limitations
0.09
(0.29)
0.28 (0.45) 0.50
0.17
(0.38)
0.17
(0.38)
0.00
Spouse’s age
67.51
(9.28)
70.65
(9.98)
0.33
71.47
(9.23)
71.34
(9.46)
0.00
Spouse’s gender
0.52
(0.50)
0.58 (0.49) 0.10
0.57
(0.49)
0.57
(0.50)
0.00
Spouse’s
education
0.54
(0.50)
0.47 (0.50) -0.13
0.48
(0.50)
0.48
(0.50)
0.00
Household
income
3.85
(1.18)
3.44 (1.25) -0.34
3.62
(1.18)
3.62
(1.18)
0.00
Spouse’s SRH
0.77
(0.42)
0.73 (0.44) -0.10
0.80
(0.40)
0.80
(0.40)
0.00
Spouse’s CES-D
8
0.09
(0.29)
0.10 (0.30) 0.03
0.05
(0.22)
0.05
(0.22)
0.00
Spouse’s TICS
2.80
(0.47)
2.71 (0.52) -0.17
2.78
(0.45)
2.78
(0.45)
0.00
Note: SNCG: spousal non-caregivers; SCG: spousal caregivers; SRH: self-rated health; Std.
Diff.: standardized differences
32
Functional limitations represent partner’s functional limitations >0, spouse’s gender, education,
self-rated health, CES-D 8 are dichotomously coded with 1 representing female, some college
and above, self-rated health ≥ 3, and CES-D 8≥ 4, respectively
33
Table 2.2. Descriptive statistics of matched samples (weight adjusted)
Variable
Total
(N=20,757)
SCGs
(N=2,714)
SNGs
(N=18,043)
Sig.
level
Self-rated health at T1 3.23 (1.00) 3.22 (0.98) 3.23 (1.00)
Self-rated health at T2 3.13 (1.03) 3.10 (1.01) 3.13 (1.03)
Change in self-rated health -0.10 (0.84) -0.12 (0.85) -0.10 (0.84)
Depressive symptoms at
T1
0.95 (1.46) 1.06 (1.58) 0.93 (1.45)
**
Depressive symptoms at
T2
1.12 (1.65) 1.44 (1.91) 1.07 (1.60)
***
Change in depressive
symptoms
0.18 (1.60) 0.38 (1.67) 0.15 (1.59)
***
Cognitive functioning at
T1
15.19 (3.98) 15.14 (3.95) 15.20 (3.99)
Cognitive functioning at
T2
14.73 (4.19) 14.55 (4.15) 14.76 (4.19)
Change in cognitive
functioning
-0.58 (3.59) -0.63 (3.55) -0.57 (3.60)
Spouse’s gender (female) 11,891 (57.29%) 1,561 (57.52%) 10,330 (57.25%)
Spouse’s race
• Non-Hispanic
White
• Non-Hispanic
Black
• Hispanics
16,210 (78.12%)
2,229 (10.74%)
2,310 (11.13%)
2,071 (76.39%)
318 (11.73%)
322 (11.88%)
14,139 (78.38%)
1,911 (10.60%)
1,988 (11.02%)
Spouse’s age at T1 71.22 (8.55) 71.32 (8.67) 71.21 (8.54)
Partner’s age at T1 71.97 (7.82) 73.18 (8.07) 71.78 (7.76)
***
Spouse’s education
(college and above)
8,061 (38.84%) 1,053 (38.80%) 7,008 (38.84%)
Spouse’s comorbidities 2.27 (1.36) 2.33 (1.32) 2.26 (1.37)
Partner’s comorbidities 2.31 (1.33) 2.78 (1.45) 2.24 (1.30)
***
Spouse’s ADL at T1 0.20 (0.63) 0.23 (0.68) 0.20 (0.62)
Spouse’s IADL at T1 0.08 (0.35) 0.07 (0.32) 0.08 (0.35)
Partner’s ADL at T1 0.20 (0.55) 0.25 (0.72) 0.20 (0.51)
**
Partner’s IADL at T1 0.07 (0.29) 0.10 (0.37) 0.06 (0.27)
***
Household income (in
2010$)
60,661.43
(82217.42)
59,937.46
(71771.14)
60,770.33
(83,677.21)
Length of current marriage
at T1
42.17 (15.89) 42.03 (16.81) 42.19 (15.75)
Number of living children
at T1
3.59 (2.05) 3.74 (2.10) 3.56 (2.05)
Note: SNCG: spousal non-caregivers; SCG: spousal caregivers; Sig. level: level of significance
*
p<.05,
**
p<.01,
***
p<.001
34
Table 2.3. Stepwise regression results for health outcomes associated with spousal caregiving
Variables
Self-rated health at T2 Depressive symptoms at T2 Cognitive functioning at T2
B (SE) Sig. B (SE) Sig. B (SE) Sig. B (SE) Sig. B (SE) Sig. B (SE) Sig.
Being a spousal caregiver
-0.04
(0.04)
-0.04
(0.03)
0.37
(0.07)
***
0.27
(0.07)
***
-0.21
(0.16)
-0.05
(0.13)
Self-rated health at T1
0.56
(0.02)
***
-0.20
(0.04)
***
0.20
(0.08)
**
CESD at T1
-0.03
(0.01)
*
0.42
(0.04)
***
-0.05
(0.05)
TICS at T1
0.01
(0.00)
*
-0.02
(0.01)
*
0.48
(0.02)
***
Spouse’s gender (female)
0.00
(0.04)
0.28
(0.07)
***
0.46
(0.14)
**
Spouse’s race (reference:
White)
• Non-Hispanic
Black
• Hispanics
-0.01
(0.06)
0.10
(0.06)
0.30
(0.13)
0.46
(0.15)
*
**
-1.38
(0.23)
-0.81
(0.24)
***
**
Spouse’s age at T1
-0.00
(0.00)
0.00
(0.01)
-0.06
(0.01)
***
Partner’s age at T1
0.00
(0.00)
0.01
(0.01)
-0.04
(0.01)
**
Spouse’s education
(college and above)
0.06
(0.03)
-0.01
(0.07)
0.92
(0.14)
***
Spouse’s comorbidities
-0.09
(0.01)
***
0.10
(0.03)
***
0.01
(0.05)
Partner’s comorbidities
-0.00
(0.01)
-0.02
(0.02)
-0.01
(0.05)
Spouse’s ADL at T1
-0.09
(0.03)
**
0.17
(0.07)
*
-0.15
(0.15)
Spouse’s IADL at T1
-0.14
(0.06)
*
0.20
(0.15)
-1.29
(0.37)
***
35
Partner’s ADL at T1
-0.03
(0.04)
0.04
(0.06)
0.17
(0.17)
Partner’s IADL at T1
0.07
(0.08)
-0.08
(0.10)
-0.32
(0.34)
Household income (2010$)
0.02
(0.02)
-0.02
(0.03)
0.12
(0.08)
Length of current marriage
at T1
0.00
(0.00)
-0.00
(0.00)
0.01
(0.00)
Number of living children
at T1
-0.00
(0.01)
0.00
(0.02)
-0.01
(0.03)
Period (reference: period 1)
• 2
• 3
• 4
• 5
0.17
(0.04)
0.07
(0.04)
0.06
(0.05)
0.10
(0.05)
***
-0.13
(0.09)
-0.07
(0.09)
0.02
(0.10)
-0.02
(0.11)
-0.34
(0.18)
-0.18
(0.21)
0.23
(0.18)
-0.15
(0.19)
Note: Sig.: level of significance
*
p<.05,
**
p<.01,
***
p<.001
36
Chapter 3. A Latent Class Analysis of Stressors and Resources among Spousal Caregivers
to Older Adults in the United States and Their Associations with Wellbeing
Introduction
Twelve percent of informal caregivers are providing care to a spouse or partner, making
them the second largest informal caregiver group behind children (AARP, 2019b). Studies have
indicated that spousal caregivers to older adults are more likely to experience worse physical
health including lower levels of self-rated health and poorer mental health such as higher
depressive symptoms, compared to both adult children and children-in-laws (Abdollahpour et al.,
2014; Kaufman et al., 2019; Oldenkamp et al., 2016; Penning & Wu, 2016; Pinquart & Sörensen,
2007, 2011; Polenick et al., 2020; Rafnsson et al., 2017) and statistically matched spousal non-
caregivers, as demonstrated in the previous chapter and another study (Chen et al., 2020).
Despite the widely documented negative impacts, more recent studies on beneficial
health outcomes associated with spousal caregiving, although relatively scarce, have also
suggested greater reports of happiness (Fredman et al., 2015) and caregiving gains (Polenick &
DePasquale, 2019). Caregiving literature in general has also found greater sense of personal
development (Pendergrass et al., 2019), increased interpersonal relationships such as feelings of
mutuality and increased family cohesion (Yu et al., 2018), and reduced mortality risk (Mehri et
al., 2021; Roth et al., 2015) associated with caregiving. The different findings may suggest
potential heterogeneity among caregivers, and the different health outcomes manifested based on
the differential caregiving experience. In fact, qualitative studies focusing on spousal caregivers
to persons with dementia and/or Parkinson disease (Daley et al., 2017; Davis et al., 2014;
Kaplan, 2001), and geriatric patients without dementia (Gehr et al., 2021) have identified
37
different typologies among spousal caregivers or spousal dyads based on caregiving burden and
adaptation to burden including coping and positive appraisals.
The stress process model, first proposed by Pearlin and colleagues (1981), has been
extensively used to understand individual differences in health outcomes among caregivers.
Results from meta-analysis have consistently shown that caregiving stressors are associated with
higher levels of depression (Pinquart & Sörensen, 2003a, 2003b) and worse physical health
(Pinquart & Sörensen, 2007; Vitaliano et al., 2003). For example, more hours of care are
associated with higher odds of depressive symptoms (Chen et al., 2020; Hirst, 2005; Pinquart &
Sörensen, 2003a) and worse physical health (Chang et al., 2010) among caregivers. Spousal
caregivers who provided help with activities of daily living (ADLs) were also found to have
more depressive symptoms (Burton et al., 2003; Dunkle et al., 2014) and physical health
(Bookwala et al., 2000; Lee et al., 2001) compared to those helping with instrumental activities
of daily living (IADLs). Caregivers to persons with dementia have also been found to have worse
physical and psychological health in meta-analysis (Pinquart & Sörensen, 2003b; Pinquart &
Sörensen, 2007).
Stressors are not the only factors that uniformly influence wellbeing. Resources can
buffer the effects of stress on health outcomes (Pearlin, 2010; Teja Pristavec, 2019). Social
support is a type of resources that is defined as a social network’s provision of psychological and
instrumental resources that intend to strengthen one’s reaction to stress (Cohen, 2004). Meta-
analysis examining social support and subjective burden among caregivers has found that there is
a moderate effect size in the relationship between perceived social support and subjective
burden, but not for received social support (del-Pino-Casado et al., 2018). However, another
study by Melrose and colleagues (2015) found that when the need for support is controlled for,
38
the relationship between received support and mental health would significantly increase. The
results suggest that the effects of received support on wellbeing is conditional on caregivers’
needs for support, which may be derived from the level of stress they are experiencing. Two
studies could partially support the hypothesis: Huang and colleagues (2009) found that social
support could buffer the effects of stressors on depressive symptoms, but only among low-
income caregivers; Li and colleagues (2019) found social support from family and friends is
protective against the negative effect of geographic proximity on caregiver’s depressive
symptoms, but only for long distance rather than short distance caregivers. Therefore, it could be
the relative intensities of stressors and resources, rather than each separate construct alone, that
are correlated with caregivers’ wellbeing.
A qualitative study (Longest & Thoits, 2012) using fuzzy set comparative analysis to
examine the stress process has supported the importance of co-occurrence and relative intensities
of stressors and resources. They studied 532 Indianapolis residents and found that it is the
combination of numerous stressors with low resources that consistently leads to poor health
outcomes, including both physical and mental health. Although the study population is not
spousal caregivers or caregivers in general, it highlighted the value of looking at stressors and
resources in combination. A review by Zarit (2018) also pointed out that one of the limitations in
existing caregiver interventions is the one-size-fits-all approach which assumes everyone is
similar in the experience of stressors and lack of resources and hence target the dimensions,
whereas caregivers differ in “exposure to stressors and access to resources”. The studies further
warrant the needs to examine stressors and resources together, and their interactive effects on
caregivers’ health outcomes.
39
The abovementioned needs, translated into quantitative analysis, call for a person-
centered approach, instead of traditional variable-centered ones where stressors and resources are
treated as distinct constructs and are assumed to have the same impacts on health for everyone
(Collins & Lanza, 2009; Fryer & Shum, 2020). Three studies up to date are identified to have
employed person-centered approaches to distinguish caregiver subgroups (Perlick et al., 2008;
Teja Pristavec, 2019)(Perlick et al., 2008; Pristavec, 2019; Sung et al., 2021). Perlick and
colleagues (2008) utilized k-means cluster analysis to characterize 500 primary caregivers to
persons with bipolar disorder, among whom 145 were spousal caregivers. Criterion variables are
classified into burden, mastery, perceived stigma, avoidance coping, and subjective support.
Three groups were identified: (1) effective caregivers (44.7%), characterized by low stress
appraisal and adoption of adaptive coping, (2) burdened caregivers (26.2%), characterized by
higher levels of burden and avoidance coping, and lower levels of mastery and social support
relative to effective caregivers, and (3) stigmatized caregivers (29.0%)), characterized by higher
levels of perceived stigma than the other two groups. Results showed that burdened caregivers
had worse health outcomes and more health service use than the other two groups, and
stigmatized caregivers reported poorer self-care than effective caregivers. Most included
caregivers are the recipients’ spouse or parents, and the distribution across the three identified
groups did not differ significantly.
Study by Pristavec (2019) used data from Round 5 (conducted in 2015) of National Study
of Caregiving (NSOC) and National Health and Aging Trends Study (NHATS) and examined
how caregiving burden and benefits characterized informal caregiving experience among 2,202
informal caregivers to older adults, 21.29% (469) of whom are spousal caregivers. Twenty-one
items were used to construct latent classes with 15 items measuring perceived burden in
40
emotional, interpersonal, physical, financial, and social domains, and 6 measuring perceived
benefits in emotional, interpersonal, and behavioral domains. Five subgroups were identified:
intensive caregivers (10%), balanced caregivers (18%), dissatisfied caregivers (15%),
relationship caregivers (26%), and satisfied caregivers (32%). No statistically significant
differences were found between spousal caregivers and adult child caregivers in group
classification; however, compared to non-relatives, spousal caregivers had a higher chance of
being dissatisfied versus satisfied caregivers. In addition, compared to White caregivers, Black
caregivers are more likely to be classified in balanced or satisfied caregiver classes. However, it
is unclear whether the five distinguishable subtypes differ in health outcomes since it was not
part of the authors’ investigation.
Study by Sung and colleagues (2021) utilized latent profile analysis to study 278
caregivers taking care of older adults in Singapore, 64 (23.0%) of whom are spousal caregivers.
Aspects used to construct latent profiles include caregiving burden and caregiving benefits.
Burden is characterized by disturbed schedule and poor health, lack of family support, and lack
of finances, whereas benefits are characterized by caregiver esteem, self-affirmation, and outlook
on life. Four caregiver subgroups were identified: balanced (39.9%), satisfied (32.7%), intensive
(17.2%), and dissatisfied caregivers (10.1%). Those with intensive and dissatisfied caregiving
experience reported higher depressive symptoms and lower quality of life compared to those
with satisfied caregiving experience. The distribution of spousal caregivers did not differ across
the four subgroups. Females and Chinese (compared to non-Chinese ethnic minorities) were
more likely to be dissatisfied caregivers, and those with higher levels of household income were
more likely to be satisfied caregivers. Similar to the previous study, one of the limitations, as
41
pointed out by the authors (Teja Pristavec, 2019; Sung et al., 2021), is the heavy focus on
psychological and emotional gains.
To sum, there are three limitations in existing studies. First, there is only a limited
number of quantitative studies using person-centered approaches to distinguish heterogeneous
caregivers based on the relative intensities of stressors they encounter and resources they
possess. Second, the three included studies have a heavy reliance on subjective measures without
considering tangible and objective resources that caregivers can mobilize. Third, none of the
studies focused on spousal caregivers exclusively. Although no statistically significant
differences were detected between spousal caregivers and other types of caregivers in the three
studies, the comparison group (parent caregivers) or small sample size might have overshadowed
the message and further precluded the potential to explore heterogeneity among spousal
caregivers.
Therefore, the goal of the current study is to identify latent classes among spousal
caregivers based on their relative exposure to stressors and possession of objective social
support. Three research questions are proposed:
1. What are the distinguishable spousal caregiver latent classes based on the co-
occurrence and relative intensities of stressors and resources that caregivers
experience and possess?;
2. How are multiple background characteristics associated with the identified spousal
caregiver classes?; and
3. How are health outcomes associated with the identified spousal caregiver classes after
controlling for covariates?
42
The identification of spousal caregiver latent classes based on their stressors and
resources could contribute to the current understanding of spousal caregivers by categorizing
them as persons with similar combinations of interactive characteristics, rather than persons
segregated by dimensions. The study explores another strategy to apply the stress process model,
as the model and the constructs themselves are “inherently and explicitly conditional or
contingent” (Longest & Thoits, 2012). The findings on heterogeneous spousal caregiver classes
could also help with the identification of not only spousal caregivers who are in need, but also
what they need, to inform the spousal caregiving dyads, other family members, social workers,
and policy makers in the provision of services and resources, interventions, and policy
recommendations.
Figure 3.1 displays the conceptual framework of the current paper.
Methods
Data
The data used are from Round 5 and Round 7 (conducted in 2017) of the NSOC, linked
with the corresponding data from NHATS. NHATS is led by the Johns Hopkins University
Bloomberg School of Public Health and is sponsored by the National Institute on Aging
(U01AG032947). The study contains a nationally representative sample of Medicare
beneficiaries aged 65 and older in the United States. The participants are administered in-person
interviews annually to collect information on disablement and its consequences (Montaquila et
al., 2012). NSOC was conducted with family members and others who provided unpaid care to
the NHATS participants. Currently, there are three rounds of NSOC data: Rounds 1 (conducted
in 2011), 5, and 7. Rounds 5 and 7 were pooled together to increase the sample size. Round 1
43
was not included because it lacked some sociodemographic variables, and the time interval was
larger between Round 1 and Round 5 or 7.
Sample
There are a total of 989 spousal caregiver observations in Rounds 5 and 7 of NSOC. Six
(0.6%) caregivers who are taking care of a same-sex spouse, and another 19 (1.9%) caregivers
who are not living with the spousal care-recipients were excluded. The exclusion criteria yield a
sample size of 964 spousal caregiver observations. Latent class analysis, which will be described
in the following sections, is performed on complete cases; the method yields a final sample size
of 793 (82.26%) observations across 639 distinct spousal caregivers.
Measures
Latent Variable: Spousal Caregiver Classes
The latent variable is spousal caregiver classes characterized by the relative intensities of
caregiving stressors and resources. Stressors in the current study include caregiving hours,
whether the spousal caregiver is helping with activities of daily living (Gerstorf, Hoppmann,
Kadlec, et al.) tasks, whether the spousal caregiver is taking care of someone with dementia,
burden in the realm of financial, emotional, physical, family conflict, and role overload.
Resources include instrumental support from friends and family, emotional support from friends
and family, and whether there are other helpers.
Similar to a previous study (Chen et al., 2020) and in accordance with the Social Security
Caregiver Credit Act (2021), providing 80 hours of caregiving or more in the previous month is
classified as high-intensity caregiver. Caregivers are asked how often they provide ADL-related
personal care, such as eating, showering, and bathing, to the recipients in the last month.
Answers are on a 1-5 scale from “every day” to “never”. Caregivers who report “some days” or
44
more are considered ADL caregivers. Care-recipients’ dementia status is generated following
NHATS’ technical paper on the classification of persons by dementia status. NHATS
participants are classified into three groups: probable dementia, possible dementia, and no
dementia. Sensitivity (85.7%) and specificity (87.2%) of the measure have been validated against
the Aging, Demographics, and Memory Study (Adams) (Kasper et al., 2015). Caregiver who is
taking care of one with possible or probable dementia is considered a dementia caregiver in the
current study. In addition, caregivers are asked: “Is helping [care recipient]
financially/emotionally/physically difficult for you?” Similar to previous studies (Chappell et al.,
2014; Lopez-Anuarbe & Kohli, 2019; Pinquart & Sörensen, 2011), the focus is on reported
financial, emotional, and physical difficulty. Answers for each category are binary yes/no
outcomes, as seen in previous studies (Beach & Schulz, 2017; R. Liu et al., 2021; Teja Pristavec,
2019; Skolarus et al., 2016): “yes” is coded as 1 and “no” is coded as 0. Family conflict is
measured by the question: “How much has your family disagreed over the details of [care
recipient’s] care?” Answers included “not so much”, “somewhat”, and “very much”, with “not so
much” coded as 0 and the other two options as 1. The construct has been tested in previous
studies as predictors of caregiver role strains (Polenick & DePasquale, 2019) and unmet needs
for care among older adults (Beach & Schulz, 2017). Role overload is measured by the question
“You have more things to do than you can handle”. Caregivers who report “very much” and
“somewhat” are coded as 1 whereas those reporting “not so much” are coded as 0. The construct
was also examined in previous studies (Ahn & Logan, 2022; Liang et al., 2020; R. Liu et al.,
2021).
Instrumental support is measured by the question: “Do you have friends or family that
help you with your daily activities, such as running errands, or helping you with things around
45
the house?” Answers are coded as 1 for “yes” and 0 for “no”. Similarly, emotional support is
measured by the question: “Do you have friends or family that you talk to about important things
in your life?” Answers are binary coded as well. For other helpers, caregivers are asked: “Do you
have friends of family that help you care for [care recipient]?” and “Have you used any service
that took care of [care recipient] so that you could take some time away from helping?”
Answering “yes” to at least one of the two questions is coded as 1 whereas 0 is for those who
report “no” to both questions. Similar constructs can be found in previous studies examining
social support among caregivers using data from NSOC (Halpern et al., 2017; Leggett et al.,
2022; Liang et al., 2020; Moon & Dilworth-Anderson, 2015).
Outcome Variables
Outcomes of interest include self-rated health and depression. Self-rated health is
measured on a 1 (excellent) to 5 (poor) scale and is reverse coded with higher score indicating
better self-rated health for easier interpretation. Depression is measured by the four-item patient
health questionnaire (PHQ-4) asking how often the caregiver (1) had little interest or pleasure in
doing things, (2) felt down, depressed, or hopeless, (3) felt nervous, anxious, or on edge, and (4)
been unable to stop or control worrying in the last month. The answer to each question ranged
from 1 “not at all” to 4 “nearly every day”. Each score is recoded into 0 to 3, and a sum score is
calculated which ranges from 0 to 12. The reliability of PHQ-4 measured by Cronbach’s alpha in
the sample is 0.73, indicating relatively high reliability.
Background Variables
Caregiver’s and care-recipient’s age is measured in years at the time of the interview.
Caregiver’s gender includes male and female. Caregiver’s race/ethnicity is classified into three
categories: non-Hispanic White, non-Hispanic Black, and Hispanics; an original non-Hispanic
46
others (including American Indian, Asian, and native Hawaiian) category is excluded due to the
small sample size. Caregiver’s education ranges from 1 “no schooling completed” to 9 “master’s,
professional, or doctoral”, and is recoded into a binary variable indicating whether the caregiver
has received education beyond high school. Number of comorbidities is a newly generated count
variable indicating the number of reported nine types of diseases including heart disease, high
blood pressure, arthritis, etc. Since care-recipient’s dementia status is already counted in as a
stressor of caregivers’, care-recipient’s number of comorbidities is calculated by the sum of
chronic conditions reported except for dementia. Household income is measured by the total
income of the caregiver and spouse from work and all other sources in the last year in dollars. To
ensure consistency across years, income is adjusted for inflation (2016$) based on the medical
services portion of the Consumer Price Index (United States Department of Labor Bureau of
Labor Statistics) for the pooled data. Income is transformed into logarithmic form in regression
analysis. The selection of background variables is informed by previous systematic and clinical
reviews (Adelman et al., 2014; Li et al., 2013; Liu et al., 2020) and the stress process model
(Pearlin et al., 1990).
Analysis
Data analysis is conducted in four steps. First, characteristics of the entire sample are
described, including means and standard deviations for continuous variables and frequencies and
percentages for categorical variables. Second, latent class analysis (LCA) is employed to identify
distinguishable spousal caregiver classes. LCA is a person-centered method that can be utilized
to decompose the heterogeneity among samples in which classes are expected to be categorically
distinct (Ruscio & Ruscio, 2008); it is not constrained by the prior specification of group
distributions (Byrd & Carter Andrews, 2016). Compared to cluster analysis, LCA assumes that
47
latent classes exist and explain patterns of observed scores, and is not constrained by the
requirement of continuous indicators (Weller et al., 2020). LCA identifies homogeneous groups
by computing the posterior class probabilities and probabilities of items conditional on class
membership (Garnett et al., 2014). Mutually exclusive and collectively exhaustive latent classes
will be identified so that the homogeneity within and heterogeneity between classes are
maximized. Individual’s probability of class membership in each latent class is calculated and
assigned to the class in which they have the highest posterior probability (Collins & Lanza,
2009). For the current study, two- to four-class solutions are estimated. The decision of the
number of classes is made based on multiple indices, including the Akaike Information Criterion
(AIC), Bayesian Information Criterion (BIC), and entropy. A lower AIC and BIC value indicates
a better fit of the model with BIC considered as the best-performing fit indicator (Nylund et al.,
2007). A higher entropy indicates lower classification error and a more precise classification
(Collins & Lanza, 2009). Model parsimony and interpretability of results are also considered in
model selection. After LCA, each spousal caregiver is assigned to the class with the highest
posterior probability, and the identified class memberships are compared against background
variables using ANOVA test or Chi-square test to further understand the different characteristics
across spousal caregiver classes. At last, multivariate regression models with ordinary least
square is estimated to examine the association between class membership and wellbeing after
controlling for background variables; standard errors are clustered by caregiver ID to account for
autocorrelation. All analyses are conducted using Stata/SE 17.0.
48
Results
Description of Study Sample
Among the 793 included spousal caregivers, the average caregiver is 75.69 (SD=8.76)
years old, taking care of a 79.16-year-old (SD=6.94) care-recipient. Most (58.01%) caregivers
are female, and the majority of them (78.52%) are non-Hispanic White. Looking at the
intersection of gender and race/ethnicity, Non-Hispanic White female (45.06%) and non-
Hispanic White male (33.47%) make up the largest group of spousal caregivers, followed by
non-Hispanic Black female (10.41%) and non-Hispanic Black male (7.77%). Over half (53.47%)
of the spousal caregivers have received education that’s beyond high school. An average
caregiver has 2.69 (SD=1.58) types of chronic conditions, and the care-recipient on average has
2.91 (SD=1.51) types of chronic conditions other than dementia. On average, spousal caregivers’
household income is $71,167.89 (SD=196,273.40) in the past year (in 2016 $). The average self-
rated health is 3.27 (SD=1.05), indicating moderate-high level of self-rated health; the average
score of PHQ-4 is 2.17 (SD=2.39), indicating relatively low levels of depressive symptoms.
For latent class variables, over one-third (36.70%) of caregivers provide more than 80
hours of care in the last month. More than half (58.64%) of them are providing assistance with
ADL activities, and 29.63% of them are caring for a spouse with possible or probable dementia.
In terms of stressors, many (40.23%) report having experienced emotional stress, followed by
physical stress (26.23%) and financial stress (17.53%). In addition, 12.11% have experienced
family conflict, and 39.22% feel that there are more things to do than they can handle. In terms
of resources, 43.63% have received instrumental support from friends or family on daily
activities such as running errands, less than half (48.17%) have other caregivers, either paid or
49
unpaid, help take care of the care-recipients, and over three quarters (78.31%) have received
emotional support from friends or family. Table 3.1 displays the sample characteristics.
Latent Class Analysis Results
Model Selection
Table 3.2 displays the model fit indices used to determine the optimal latent class
solution. The differences in entropy values are minimal, with statistics slightly in favor of models
with three or four classes compared to that with two classes. The lower BIC value with three
classes implies that a three-class model could be the best fit. Taking model interpretability into
consideration, the distribution of spousal caregivers is more balanced in the three-class model
with each class having above 25% of the observations, whereas the four-class model generates
two groups with relatively small portions of observations (10-15%). Therefore, the three-class
model is selected in the final analysis.
Profiles of Spousal Caregiver Latent Classes
Figure 3.2 displays the three spousal caregiver classes based on their stressors and
resources. Table 3.3 shows the distribution of selected attributes across identified classes.
Class 1 (low-stress low-support spousal caregivers; 39%). This class of spousal
caregivers is exposed to the lowest level of stressors, featured by fewer caregiving hours, less
chance of being either an ADL or a dementia caregiver, and lower odds of having financial and
physical stresses. They also experience less family conflict, although the level of emotional
stress and role overload is similar or slightly higher than the second class of caregivers.
50
Simultaneously, or subsequently, they are receiving less instrumental support from friends,
family, and other paid or unpaid caregivers. Although experiencing some level of emotional
stress and role overload, they are receiving less emotional support compared to the other two
groups.
Class 2 (medium-stress high-support spousal caregivers; 36%). This class of spousal
caregivers is characterized by the high levels of resources they possess. Compared to the low-
stress low-support class, they are more likely to be providing help with ADLs, slightly more
likely to be providing more hours of care and taking care of someone with dementia. Although
they have higher odds of experiencing financial stress and family disagreement over care-
recipient’s care plans, the odds of having emotional stress are as low as those with low stress,
and they have the lowest odds of feeling role overload. The very high levels of emotional and
instrumental support they receive may help explain the relatively lower odds of emotional stress
and role overload.
Class 3 (high-stress medium-support spousal caregivers; 25%). This class of spousal
caregivers is simultaneously exposed to high levels of stressors and in possession of medium-
high levels of resources. Compared to the other two groups, spousal caregivers in this class are
more likely to provide longer hours of care, take care someone with ADL needs or with
dementia. Except for family conflict, they are experiencing high stress across multiple
dimensions – financially, physically, emotionally; they also tend to be at higher chances of
feeling role overload. Although they are receiving moderate-high levels of emotional support,
compared to those in medium-stress high-support group, they are at lower chance of having
someone assist with daily errands or taking care of the care-recipients.
51
Features Associated with Spousal Caregiver Classes
Table 3.4 displays the results from bivariate analysis on the associations between
background variables and latent class memberships. Results indicate that the average age among
medium-stress high-support caregiver class is higher (Mean=77.29, SD=8.37), whereas the
average age among the high-stress medium-support caregiver class is the youngest (Mean=73.40,
SD=9.16). Compared to the other two groups, high-stress medium-support spousal caregivers
include far more female than male with nearly three quarters of subsamples identified as female.
Compared to low-stress low-support and high-stress medium-support classes, medium-stress
high-support class has greater portion of non-Hispanic Black individuals. Caregivers’ number of
comorbidities are also relatively small in low-stress low-support class compared to the other two.
No statistically significant differences in other social determinants of health are identified,
including household income and education.
Spousal Caregiver Classes and Associated Wellbeing
Multivariate linear regression analyses are performed to examine the relationship
between spousal caregiver classes and wellbeing measures including caregivers’ self-rated health
and depressive symptoms. Table 3.5 displays the finding.
Results indicate that compared to low-stress low-support spousal caregivers, high-stress
medium-support spousal caregivers on average score 0.42-unit lower on self-rated health,
indicating they have worse self-rated health. In addition, compared to non-Hispanic White
spousal caregivers, non-Hispanic Black spousal caregivers have reported lower levels of self-
rated health. In addition, caregivers’ increased number of chronic conditions is negatively
52
associated with better self-rated health. Spousal caregivers who are more educated and who have
more household income also report higher levels of self-rated health.
For depressive symptoms, high-stress medium-support spousal caregivers on average
score 1.87-unit higher on PHQ-4 compared to low-stress low-support spousal caregiver
counterparts, indicating they have higher levels of depressive symptoms. Different from the
results on self-rated health, no racial/ethnic differences are found in depressive symptoms, nor is
household income correlated with spousal caregivers’ level of depression. Higher levels of
education and fewer chronic conditions among spousal caregivers are still positively associated
with lower levels of depressive symptoms at statistically significant levels.
Discussion
Using a nationally representative data on community dwelling older adults and their
informal caregivers, the current study has identified three latent classes of spousal caregivers
based on the relative intensities of stressors and resources they are exposed to and in possession
of. The associations between class memberships and background factors as well as health
outcomes including self-rated health and depressive symptoms are also explored.
Although traditional variable-centered approaches could identify the relationships
between constructs of interests and the dependent outcomes, individual’s experience within one
category is largely shaped by or correlated with the membership to other categories (Garnett et
al., 2014; Rosenfield, 2012)\. Our results indicate that spousal caregivers are at different
intersections of stressors and resources, both of which are represented by a series of indicators.
In addition, the level of stressors a spousal caregiver is exposed to is not correlated with the
resources they possess in a linear way, either positive or negative. For example, although low-
53
stress spousal caregivers also have the lowest level of resources, spousal caregivers with medium
levels of stressors demonstrated the highest level of resources received, and the ones with the
highest level of stressors are receiving a medium-level of resources.
Results on the non-linear interactive relationships between stressors and resources are
similar to findings from previous studies examining perceived caregiving burden and benefits
(Perlick et al., 2008; Teja Pristavec, 2019; Sung et al., 2021). The current study identifies fewer
(three) classes compared to the previous two using LCA (five classes) and LPA (four classes).
Since the current study focuses on spousal caregivers only, it is plausible that the relatively more
homogeneous subpopulation will produce fewer latent classes compared to studies focusing on
informal caregivers in general. After combining the more nuanced degrees of separations in the
previous two LCA/LPA studies, for example, combining intensive and dissatisfied caregivers
(25%), as well as balanced and relationship caregivers (43%) in the study of Pristavec (2019),
and intensive and dissatisfied caregivers (27%) in the study of Sung et al. (2021), the current
study generates similar proportions of spousal caregivers across three classes, which to some
extent validates the exploratory results of the analysis.
For the high-stress medium-support spousal caregiver class, a substantial larger
proportion is female, and they are expressing extensively high levels of emotional stress and role
overload, while at the same time more likely to be taking care of a spouse with ADL needs or
dementia. They have also expressed high levels of emotional stress, and are less likely to have
received instrumental support from friends, family, or paid caregivers to help with their daily
errands or take care of the care-recipients. The finding is similar to the that from Sung and
colleagues (2021) where female is more likely to be dissatisfied caregivers. It is also consistent
with previous studies indicating that female caregivers experience higher levels of emotional
54
stress (Chiao et al., 2015; Pillemer et al., 2018; Pinquart & Sörensen, 2006a; Xiong et al., 2020),
either because of the different structural contexts and unequal distributions of rewards and
opportunities they are in (Pearlin et al., 1990), or because female feel more obliged and
responsible to care than their male counterparts (Calasanti, 2010; Mc Donnell & Ryan, 2011;
Swinkels et al., 2019). The high levels of financial stress among the subgroup of high-stress
medium-support spousal caregivers, and the fact that most of them are female, are also consistent
with previous findings that female caregivers are more likely to report financial burden because
that they are likely to lose more wages and social security benefits compared to male
counterparts, due to quitting workforce to fulfill caregiving tasks (Lee et al., 2015). The higher
level of emotional and financial burden could also be correlated with the fact that they are also
more likely to be taking care of a partner with dementia and ADL needs.
Results also indicate that there is a much higher proportion of non-Hispanic Black
caregivers in the medium-stress high-support group compared to the other two groups, which is
similar to the previous study suggesting Black caregivers are more likely to belong to balanced
and satisfied caregivers (Teja Pristavec, 2019). Results are also consistent with findings from a
qualitative study (Epps et al., 2021) where African American caregivers are more likely to
receive support from a network of family members that extend beyond the primary caregiver and
care-recipient dyads. Another study by Dilworth-Anderson and colleagues (2005) also found that
African American caregivers have higher scores on the Cultural Justification for Caregiving
Scale compared to White counterparts, suggesting that they are more reliant on the
interdependence of family and community members, and reciprocity is expected among family
members. Previous quantitative studies and systematic review have largely focused on appraisal
of caregiving in the examination of racial/ethnic differences, and concluded that African
55
Americans, compared to White counterparts, have more positive perceptions towards caregiving,
hence lower levels of burden (Dilworth-Anderson et al., 2002; C. Liu et al., 2021). The current
study highlights that beyond internal positive appraisal, stress and burden associated with
caregiving could also be buffered the external and quantifiable resources and social support
spousal caregivers receive.
No statistically significant differences have been found in both self-rated health and
depressive symptoms between low-stress low-support spousal caregivers and medium-stress
high-support spousal caregivers. However, high-stress spousal caregivers report both worse self-
rated health and higher levels of depressive symptoms compared to the low-stress low-support
counterparts, albeit possessing a medium-level of support. The findings are consistent with that
from cluster analysis (Perlick et al., 2008) and latent profile analysis (Sung et al., 2021). The
worse health outcomes among high-stress medium-support spousal caregivers have highlighted
the importance and practicality of looking at stressors and resources together: the medium level
of resources that spousal caregivers possess may not be sufficient relative to the high-level of
stressors they are exposed to. It is the imbalance between stressors and resources, rather than
stressors or resources alone, that is associated with spousal caregivers’ health outcomes. If
stressors and resources were considered separately, the medium level of resources that the third
class of spousal caregivers have may obscure the high level of exposure to stressors, while the
high levels of stressor and the relative insufficiency of resources make this class the most
vulnerable. Similarly, low-stress low-support spousal caregivers may have been identified as in
need due to the low level of resources, whereas the low levels of stressors they face may not
necessarily require high levels of resources. Likewise, targeting stressors without considering
resources may put medium-stress high-support spousal caregivers into agenda given the medium
56
level of stressor exposure; however, the resources they possess from revenues outside of
potential interventions, such as family, friends, and paid caregivers, could have already buffered
part of the negative impacts from stressors.
Results have also confirmed the importance of social determinants of health, especially
education and income. Although the distribution of educational attainment and household
income do not statistically differ across groups, both are statistically significantly associated with
at least one of the health outcomes examined, where people with higher levels of education and
more income are better off compared to their counterparts. That is, despite the heterogeneity
among spousal caregivers, social determinants are still driving factors contributing to health
disparities and warrant further examination, as suggested by Young and colleagues (2020).
There are several limitations in the current study. First, the cross-sectional design
precludes the temporal association between caregiver classes and health outcomes. Associations
between class membership and health outcomes could not discern whether high-stress medium-
support caregivers, for example, already experienced worse self-rated health and mental health
prior to assuming caregiving role. Future research could include subsequent waves and conduct
latent transition analysis to explore how spousal caregivers’ class memberships change, and how
such changes affect the subsequent health outcomes. The small sample sizes of Hispanic
participants have also limited the potential racial/ethnic differences between Hispanic and non-
Hispanic White in group membership distribution, health outcomes, and other background
characteristics. Future studies are recommended to include racial/ethnic minority populations and
those who are not Medicare beneficiaries, when data allows. At last, due to the exploratory
nature, sampling weights are not included in the current study to increase the internal validity.
57
The tradeoff is that the results will lose some external validity and may not be nationally
representative.
The limitations notwithstanding, this study has several contributions and implications.
First, it extends the stress process model to characterize spousal caregivers based on the
simultaneous exposures to and possessions of stressors and resources. Instead of looking at them
as separate constructs, using indicators of stressors and resources altogether created three
distinguishable spousal caregiver classes that are homogeneous in the levels of (im)balances in
the two aspects. Therefore, policies, social service programs, and interventions should not only
target those either in low resources or high stressors, but also the subpopulations facing a gap in
the resources they possess that could be mobilized to buffer the stressors they face. Especially,
when spousal caregivers have to fulfill their caring obligations and bear with the inevitable
stressors, increasing the levels of resources they possess, to the extent to be able to buffer the
detrimental impacts of stressors, could reduce the odds of suffering from negative health
outcomes. Second, social support is a multi-dimensional concept. The current study utilizes
tangible social support that is more objective than subjective appraisals of resources, which could
be more straightforward to policy makers, social service providers, and social workers in the
identification of groups in need. At last, the study focuses exclusively on spousal caregivers to
older adults, instead of lumping them with other family caregivers. Spousal caregivers are more
subject to negative health outcomes and are more likely to be taking care of a partner alone in the
last years of life (Ornstein et al., 2019). In addition, spousal dyads could influence each other’s
health (Lu & Shelley, 2019; Meyler et al., 2007). Identifying spousal caregiver latent classes
could contribute to the understanding of this especially vulnerable population, which could also
potentially inform the exploration of differential health impacts on those of the care-recipients.
58
Tables
Table 3.1. Sample characteristics of spousal caregivers (N=793)
Variable Mean (SD) or N (%)
Latent Class Variables
Stressors
Caregiving hours ≥ 80 hours/month 291 (36.70%)
ADL caregiver 465 (58.64%)
Dementia caregiver 235 (29.63%)
Financial stress 139 (17.53%)
Emotional stress 319 (40.23%)
Physical stress 208 (26.23%)
Family conflict 96 (12.11%)
Role overload 311 (39.22%)
Resources
Instrumental support 346 (43.63%)
Emotional support 621 (78.31%)
Other caregivers 382 (48.17%)
Outcome Variables
Self-rated health 3.27 (1.05)
Depressive symptoms (PHQ-4) 2.17 (2.39)
Background Variables
Caregiver’s age 75.69 (8.76)
Care-recipient’s age 79.16 (6.94)
Caregiver’s gender (female) 460 (58.01%)
Caregiver’s race/ethnicity
• Non-Hispanic White
• Non-Hispanic Black
• Hispanic
596 (78.52%)
138 (18.18%)
25 (3.29%)
Caregiver’s education (above high school) 424 (53.47%)
Caregiver’s number of comorbidities 2.69 (1.58)
Care-recipient’s number of comorbidities (other than dementia) 2.91 (1.51)
Household income (2016$) 71167.89 (196273.40)
59
Table 3.2. Model fit statistics for spousal caregiver latent classes with two- to four-class
solutions
Classes LL df AIC BIC Entropy
2 -4981.75 23 10009.49 10117.03 0.94
3 -4830.96 35 9731.92 9895.57 0.95
4 -4798.10 47 9690.20 9909.96 0.95
Note: LL=Log likelihood; df=Degrees of freedom; AIC=Akaike information criterion;
BIC=Bayesian information criterion
60
Table 3.3. Spousal caregiver latent classes characterizing stressors and resources (N=793)
Spousal Caregiver Classes
Class 1
Low-Stress
Low-support
SCGs
Class 2
Medium-Stress
High-Support
SCGs
Class 3
High-Stress
Medium-Support
SCGs
N (%) 313 (39.47%) 282 (35.56%) 198 (24.97%)
Items Probability (%) of endorsing item
Stressors
Caregiving hours ≥ 80 hours/month 25 30 66
ADL caregiver 42 62 81
Dementia caregiver 22 31 40
Financial stress 4 11 49
Emotional stress 24 24 90
Physical stress 9 12 75
Family conflict 8 14 15
Role overload 26 23 82
Resources
Instrumental support 8 85 44
Emotional support 67 92 78
Other caregivers 8 89 57
Note: SCG=Spousal caregiver
61
Table 3.4. Sample characteristics across identified spousal caregiver latent classes (N=793)
Variables Spousal Caregiver Classes Sig.
level 1 2 3
Caregiver’s age 75.70 (8.56) 77.29 (8.37) 73.40 (9.16)
***
Care-recipient’s age 78.70 (6.71) 79.98 (7.18) 78.72 (6.88)
*
Caregiver’s gender (female) 171 (54.63%) 145 (51.42%) 144 (72.73%)
***
Caregiver’s race/ethnicity
• Non-Hispanic White
• Non-Hispanic Black
• Hispanic
255 (84.44%)
38 (12.58%)
9 (2.98%)
189 (70.79%)
70 (26.22%)
8 (3.00%)
152 (80.00%)
30 (15.79%)
8 (4.21%)
**
Caregiver’s education (above high school) 182 (58.15%) 141 (50.00%) 101 (51.01%)
Caregiver’s number of comorbidities 2.47 (1.51) 2.81 (1.62) 2.87 (1.62)
**
Care-recipient’s number of comorbidities 2.84 (1.52) 2.89 (1.47) 3.03 (1.54)
Household income (2016$) 92278.34
(303351.9)
57012.07
(51206.63)
57957.66
(64990.53)
Note: mean and standard deviations for continuous variables; frequencies and percentages for categorical variables; Class 1=low-
stress low-support spousal caregivers, Class 2=medium-stress high-support spousal caregivers, Class 3=high-stress medium-
support spousal caregivers
Sig. level=level of significance;
*
p<.05,
**
p<.01,
***
p<.001
62
Table 3.5. Association of spousal caregiver subgroups and wellbeing: Results from linear regression models (N=793)
Self-rated health Depressive symptoms
Spousal caregiver subgroups (ref: low-stress low-support SCGs)
• Medium-stress high-support SCGs
• High-stress medium-support SCGs
-0.08 (-0.24, 0.09)
-0.42 (-0.59, -0.25)
0.01 (-0.33, 0.36)
1.87 (1.42, 2.31)
Caregiver’s age -0.01 (-0.01, 0.00) -0.00 (-0.02, 0.01)
Care-recipient’s age 0.00 (-0.01, 0.01) -0.00 (-0.03, 0.02)
Caregiver’s gender (female) 0.02 (-0.14, 0.17) 0.32 (-0.01, 0.66)
Caregiver’s race/ethnicity (ref: non-Hispanic White)
• Non-Hispanic Black
• Hispanics
-0.21 (-0.41, -0.01)
-0.26 (-0.62, 0.11)
-0.18 (-0.60, 0.24)
0.04 (-1.08, 1.15)
Caregiver’s education (above high school) 0.27 (0.13, 0.41) -0.50 (-0.84, -0.16)
Caregiver’s comorbidities -0.27 (-0.31, -0.22) 0.26 (0.16, 0.37)
Care-recipient’s comorbidities -0.03 (-0.07, 0.11) 0.05 (-0.05, 0.15)
Household income 0.05 (0.00, 0.11) 0.02 (-0.08, 0.12)
Note: SCGs=Spousal caregivers. 95% confidence interval in parentheses. Bolded values indicate statistically significant results based
on Wald test (p<.05)
63
Figures
Figure 3.1. Latent classes among spousal caregivers to older adults based on the relative
intensities of stressors and resources as informed by the stress process model
64
Note: For the figure on the top, sections with grey background color represent stressors; sections with white
background color represent resources
Figure 3.2. Spousal caregiver latent classes based on relative intensities of stressors and
resources (N=793)
65
Chapter 4. Predicting Care-Recipients’ Wellbeing Based on Spousal Caregivers’ Co-
Occurring Stressors and Resources
Introduction
Research on consequences of caregiving on caregivers’ wellbeing has different findings.
The negative consequences have been widely documented: caregivers are experiencing worse
health both psychologically and physically (Feast et al., 2016; Ornstein & Gaugler, 2012;
Pinquart & Sörensen, 2003b; Vaingankar et al., 2016). Simultaneously, there are beneficial
outcomes associated with caregiving among spousal caregivers (Fredman et al., 2015; Polenick
& DePasquale, 2019) and informal caregivers in general (Mehri et al., 2021; Pendergrass et al.,
2019; Roth et al., 2015; Yu et al., 2018), including higher levels of happiness, greater sense of
personal development, increased interpersonal relationship, and reduced mortality risk. The
findings suggest that caregiving is a multifaceted activity, and the different aspects inherent in
caregiving could contribute to health outcomes in different directions.
Besides caregivers, care-recipients could also be impacted by caregiving, since the
caregiving relationship is dyadic by definition (T. Pristavec, 2019; Roberto & Jarrott, 2008).
Recent studies, although relatively scarce, have examined the impacts of different caregiving
facets, including caregiving stressors and caregivers’ resources, on health and wellbeing among
care-recipients. For example, Schulz and colleagues (2021) utilized data from National Health
and Aging Trends Study (NHATS) and the linked data from National Study of Caregivers
(NSOC), and found that caregiving burden is an independent predictor of care-recipients’
mortality. Kuzuya and colleagues (2011) conducted a longitudinal study among 1,067 pairs of
community-dwelling older adults aged 65 and above and their informal caregivers in Japan. They
found that heavy caregiver burden, measured by Zarit Burden Interview, is associated with care-
66
recipients’ mortality and hospitalization, even after adjusting for confounders. In a systematic
review examining the association of caregiver stress and dementia care-recipients’ health
outcomes, Stall and colleagues (2019) report that informal caregivers’ distress is commonly
associated with care-recipients’ institutionalization, worsening behavioral and psychological
symptoms, and experience of elder abuse. Other cross-sectional studies also reported that
caregiver-reported burden is statistically significantly correlated with care-recipients’ level of
depressive symptoms (Ejem et al., 2018; Ejem et al., 2015) and loneliness (Iecovich, 2014).
Three studies (Gellert et al., 2018; Kelly et al., 2017; Leung et al., 2020) examined the
relationship between caregiver’s resources and care-recipients’ health outcomes. Leung and
colleagues (2020) studied 225 patient-caregiver dyads recruited from three hospitals in Hong
Kong, and found that family support had a significant positive indirect effect on recipients’
quality of life through caregivers’ self-efficacy, and friend support had a significant positive
direct effect on caregivers’ burden but a minimal effect on recipients’ quality of life. Kelly and
colleagues (2017), utilizing data from NHATS and NSOC, found that caregivers’ social
engagement was positively correlated with care-recipients’ health, whereas the use of
organizational support was negatively correlated with care-recipient’s health. Gellert and
colleagues (2018) studied 108 individuals with early-state dementia and their spousal caregivers
at baseline and one month apart. They found that the relationship between caregivers’ stress and
care-recipients’ quality of life was moderated by social support. The study also examined stress
and quality of life separately for caregiver and care-recipients, and no partner effect (the impact
of one partner to his or her counterpart) was identified.
Only two studies (T. Pristavec, 2019; Pristavec & Luth, 2020) examined the positive and
negative aspects of caregiving together instead of in isolation. Pristavec (2019) utilized
67
longitudinal data from NHATS and the linked data from NSOC, and examined how caregiver’s
experience, featured by the simultaneous experience of caregiving burden and perceived
benefits, prospectively affected care-recipients’ mental health two years later among 781 older
adults and their informal caregiver dyads. The five included caregiver categories were (1)
intensive caregivers who reported high caregiving burden and moderate caregiving benefits
across all physical, personal, social, and interpersonal domains, (2) balanced caregivers who
reported moderate burden except for social burden, and benefits, (3) dissatisfied caregivers who
reported burden only, (4) relationship caregivers who reported low interpersonal burden only,
and interpersonal benefits, and (5) satisfied caregivers who only reported benefits. Using binary
logistic regressions, she found that compared to recipients who had dissatisfied caregivers, those
whose caregivers were satisfied were less likely to report depression, and those who were taken
care of by caregivers who belonged to balanced, relationship, and satisfied groups were less
likely to report anxiety. This study implies that caregiving experience, shaped by caregiving
burden and benefits, could impact care-recipients’ health outcomes. However, the study did not
control for care-recipients’ level of depression or anxiety at baseline, which could induce omitted
variable bias, and overestimate the correlations of the independent and control variables,
including the statistical significance of spousal caregiver categories. A later study by Pristavec
and Luth (2020) compared mortality risks among care-recipients over a six-year follow-up. They
created binary variables for caregiving burden and caregiving benefits separately and created an
interaction term of the two. Results indicated that for those whose caregivers reported burden
only, and caregivers who reported burden alongside benefits, the mortality risks were 38% and
5% higher than those whose caregivers reported neither burden nor benefits, respectively. The
findings also indicated potential differences in care-recipients’ health. However, the study did
68
not account for nuances in the relative intensities of caregiving burden and benefits that have
previously been identified by the authors (Teja Pristavec, 2019). Nevertheless, there is an
implication that when caregivers’ relative intensities of caregiving stressors and resources are
taken together, there are heterogeneous caregiving experiences, and the uniquely characterized
experiences could be differentially associated with care-recipients’ health outcomes.
To sum, there is evidence, albeit limited, from existing literature that indicates (1)
caregiving stress is negatively associated with care-recipients’ health outcomes, (2) caregivers’
resources, especially support from family and friends, are positively associated with care-
recipients’ health outcomes, and (3) caregivers’ characteristics, featured by the co-occurrence
and relative intensities of caregiving stressors and resources, could be differentially associated
with care-recipients’ health outcomes. However, limitations include that (1) many studies,
especially those focusing on care-recipients’ outcomes, are conducted outside of the United
States and using convenience samples, (2) studies either focus on older adults and their informal
caregivers in general, or a subpopulation with a particular condition (dementia, cancer, or
receiving palliative care); no studies have a specific focus on spousal caregivers, (3) most studies
have examined a single aspect of caregiving without taking into account the multi-faceted nature,
and the two studies (T. Pristavec, 2019; Pristavec & Luth, 2020) that considered the
dimensionality of caregiving could either be subject to omitted variable bias or did not fully
utilized the nuanced caregiving experience identified previously.
In the previous chapter, three spousal caregiver latent classes are identified – low-stress
low-support spousal caregivers (Class 1), medium-stress high-support spousal caregivers (Class
2), and high-stress medium-support spousal caregivers (Class 3) – featured by the differences in
relative intensities of stressors and resources they encounter and possess. High-stress medium-
69
support caregivers are found to have worse health outcomes compared to their low-stress low-
support counterparts. Following the inquiry, the current paper seeks to explore, whether
wellbeing among care-recipients differs across the three classes. Specifically, we are interested in
exploring the differences in both health status at baseline, and the levels of changes across
groups.
Methods
Data
The data used are from Round 5 (conducted in 2015) and Round 7 (conducted in 2017) of
NSOC, linked with the corresponding data of NHATS from Round 5 (conducted in 2015) to
Round 8 (conducted in 2018). NHATS is led by the Johns Hopkins University Bloomberg
School of Public Health and is sponsored by the National Institute on Aging (U01AG032947).
The study contains a nationally representative sample of Medicare beneficiaries aged 65 and
older in the United States living in residential care or communities. The participants are
administered in-person interviews annually to collect information on disablement and its
consequences (Montaquila et al., 2012). NHATS oversamples Black older adults and those aged
85 and older. The baseline response rate is 71% in 2011. NSOC is conducted with family
members and others who provide unpaid care to the NHATS participants on self-care, mobility,
or household activities. Each NHATS participant can list up to five caregivers. The NSOC
response rates were 67.2% for Round 5 and 92.4% for Round 7 (Freedman et al., 2019).
Sample
The identification and inclusion/exclusion criteria for the 793 observations across 639
distinct spousal caregivers were described in the previous chapter (Chapter 3). To get additional
information on care-recipients, the pooled spousal caregiver data were split into two files by
70
study rounds. Data originally from Round 5 of NSOC/NHATS was merged 1:1 with Round 6
(conducted in 2016) of NHATS based on care-recipient’s identifier (spid), and data from Round
7 was merged with that from Round 8 of NHATS in the same fashion. The two merged files
were then appended to generate the new sample.
Same as the previous study, the new sample contains 793 care-recipient observations
across 639 unique individuals. Since the current paper contains two waves of data for each
observation, for individuals with repetitive observations (n=154; 19.43%), only the first two
waves of data were included in the current study. Therefore, the final sample includes 639
distinct spousal care-recipients with observations in two consecutive waves.
Measures
Outcome Variables
Outcomes of interest include care-recipients’ self-rated health and depressive symptoms
at both time points. Self-rated health is measured on a 1 (excellent) to 5 (poor) scale and is
reverse coded with higher score indicating better self-rated health for easier interpretation.
Depressive symptoms are measured by the four-item patient health questionnaire (PHQ-4)
asking how often the respondent (1) had little interest or pleasure in doing things, (2) felt down,
depressed, or hopeless, (3) felt nervous, anxious, or on edge, and (4) been unable to stop or
control worrying in the last month. The answer to each question ranged from 1 “not at all” to 4
“nearly every day”. Each score is recoded into 0 to 3, and a sum score is calculated which ranges
from 0 to 12. The reliability (Cronbach’s alpha) in the sample of PHQ-4 is 0.75 and 0.77, for
both time points, respectively, indicating relatively high reliability.
71
Spousal Caregiver Classes
Spousal caregiver classes include three categories, as identified and described in the
previous chapter: low-stress low-support spousal caregivers, medium-stress high-support spousal
caregivers, and high-stress medium-support spousal caregivers. Low-stress low-support spousal
caregivers are chosen as the reference group.
Control Variables
Care-recipient’s age is measured in years at the time of the interview. Care-recipient’s
gender includes male and female. Race/ethnicity is classified into three categories: non-Hispanic
White, non-Hispanic Black, and others; the “other” group includes participants who self-reported
to be Hispanic/Latino, American Indian, Asian, and native Hawaiian. Due to small sample sizes
in Hispanic/Latino (n=13) and other (n=14), they are combined into one group. Education ranges
from 1 “no schooling completed” to 9 “master’s, professional, or doctoral”, and is recoded into a
binary variable indicating whether the care-recipient has received education beyond high school.
Number of comorbidities is a newly generated count variable indicating the reported eight types
of disease including heart disease, high blood pressure, arthritis, etc. Dementia status is not
counted as part of comorbidities as it is already taken into account as an indicator constructing
the spousal caregiver latent classes. Household income is measured by total income of the care-
recipient and spousal caregiver from work and all other sources in the last year in dollars. To
ensure consistency across years, income is adjusted for inflation (in 2016$) based on the medical
services portion of the Consumer Price Index (United States Department of Labor Bureau of
Labor Statistics) for the pooled data. Income is transformed into logarithmic form in regression
analysis. The selection of background variables is informed by previous studies (e.g. Pristavec,
2019; Pristavec & Luth, 2020).
72
For control variables, gender, race/ethnicity, and education are considered time-invariant;
age, number of comorbidities, and household income are time-variant. Since NHATS does not
provide information on household income in Round 6 and 8, and care-recipients number of
comorbidities have not changed much over the two included years, only baseline information of
time-variant variables is included.
Analysis
Data analysis is conducted in three steps. First, missing values for health outcomes at the
subsequent wave (T2) are imputed using multiple imputation (MI) with 50 iterations. MI uses a
regression-based approach to create a set of values for missing observations, and is considered to
be the best practice to address missing data in statistical analysis (Kenward & Carpenter, 2007).
Second, sample characteristics are described, with frequencies and percentages for categorical
variables and means and standard deviations for continuous variables. Differences in sample
characteristics across three groups are compared using ANOVA or Chi-square tests. In order to
assess differences in baseline (intercepts) and rates of changes (slopes) across three classes, a
series of models are estimated in the third step using generalized estimating equations (GEEs).
GEE is a marginal model for longitudinal data; compared to mixed-effect models which take an
individual-level approach, GEE uses a population-level approach based on quasi-likelihood
functions and provides the population-average estimates of parameters (Wang, 2014). GEE
relaxes the assumption of traditional regression models, such as normal distribution, and only
requires the correct specification of marginal mean, variance, and link function (Diggle et al.,
2002; Liang & Zeger, 1986; Wang, 2014; Zeger & Liang, 1986). Quiñones and colleagues
(2019) have adopted the same method examining changes in the accumulation of chronic
conditions across different racial/ethnic groups. In the current paper, a linear specification and
73
first-order autoregressive covariance structure are adopted to account for the non-independence
of observations for each individual over time. Model 1 displays the unconditional GEE models
and reports results for the whole population. Model 2 includes coefficients for spousal caregiver
latent classes, and the interaction between time and classess. The coefficients for classes indicate
the differences in baseline characteristics (intercepts) compared to the reference group (low-
stress low-support spousal caregivers), and the coefficients for the interaction term estimate the
differences in rates of change (slopes) compared to the reference group. Model 3 expands on
Model 2 to include covariates.
To check for the robustness of results, the same analyses are performed on the complete
sample (N=509; 79.66%). All analyses were performed using STATA/SE 17.0, and figures are
generated using Microsoft Excel.
Results
Table 4.1 displays the sample characteristics. Among the 639 included spousal care-
recipients, the average self-rated health at baseline is 2.73 (SD=1.02), and decreases to 2.67
(SD=0.93) one year later. The baseline depressive symptoms are 2.65 (SD=2.76), with a 0.18-
unit increase on average over the year. Care-recipients are 79.23 (SD=6.45) years old on average,
most (58.37%) of them are male, and the majority (79.40%) are identified as non-Hispanic
White. Over half (53.03%) of them have received education that is beyond high school. At
baseline, they on average have 2.86 (SD=1.51) types of chronic conditions other than dementia,
and the average household income (in 2016$) is $74,181.69 (SD=216,816.20). For comparison
across groups, care-recipients whose spousal caregivers belong to the high-stress medium-
support caregiver class have lower levels of self-rated health and higher levels of depressive
symptoms both at baseline and one year later. The rate of increase in depressive symptoms is
74
also higher among recipients taken care of by high-stress medium-support spousal caregivers,
compared to their counterparts in the other two groups. In addition, there are much fewer female
care-recipients (26.32%) taken care of by high-stress medium-support spousal caregivers. The
proportion of non-Hispanic Black care-recipients is higher in the group taken care of by medium-
stress high-support spousal caregivers, and the level of education among care-recipients taken
care of by low-stress low-support spousal caregivers is higher than those in the other two groups.
GEE regression results can be found in Table 4. 2 and Table 4.3. For self-rated health
across the whole sample, Model 1 indicates that its decrease over time (β=-0.06, 95% CI=[-0.13,
0.01]) is not statistically significant. Model 2 and Model 3 have similar results indicating that the
rates of changes in health are not statistically significant over time, and the rates do not differ
statistically significantly across groups. In the unadjusted model (Model 2), the baseline level of
self-rated health is 0.25-unit and 0.44-unit lower for those taken care of by spousal caregivers
who are in medium-stress high-support and high-stress medium-suupport classes, respectively,
compared to their counterparts taken care of by low-stress low-support spousal caregivers. The
differences decrease to 0.20- and 0.35-unit in the covariate-adjusted model (Model 3). Self-rated
health at baseline is statistically significantly associated with self-rated health one-year later
across all models. In addition, being a minority (other than non-Hispanic White) and having
more types of chronic conditions are negatively associated with self-rated health at T2, whereas
older ages and having education attainment that’s beyond high school are protective factors.
For depressive symptoms, the rate of changes is statistically significant (a 0.3-unit
increase) in Model 2, but the significance is no longer evident after controlling for other
75
covariates, as seen in Model 3. In addition, only recipients taken care of by high-stress medium-
support spousal caregivers are found to have higher levels of depressive symptoms at baseline
(β=1.21, 95% CI=[0.68, 1.74]), while the differences between caregivers taken care of by
medium-stress high-support spousal caregivers and low-stress low-support spousal caregivers are
not statistically significant in the covariate-adjusted model (Model 3). For covariates, older age
and higher levels of education are protective factors of depressive symptoms, whereas being of
racial/ethnic minority that is not non-Hispanic Black and increased number of comorbidities are
associated with higher levels of depression. Figure 4.1 presents findings from covariate-adjusted
models to visualize the differences and changes in self-rated health and depressive symptoms
across groups.
GEE performed on complete cases have similar results, indicating the robustness of our
findings. The comparison between complete cases and missing cases in baseline characteristics,
and results for regression analyses performed on complete cases can be found in appendix.
Discussion
The current study explores the relationship between spousal caregiver classes, featured by
the relative intensities of stressors and resources they have, and care-recipients’ health outcomes
in both baseline status and rates of change. We find that care-recipients taken care of by high-
stress medium-support spousal caregivers have worse health at baseline, in both self-rated health
and depressive symptoms. However, no differences in rates of changes are found to be
statistically significant across groups, nor is the rate significant overall. In addition to health
76
status at baseline, health outcome at T2 is mostly correlated with other health indicators such as
chronic conditions and sociodemographic characteristics, especially race and education.
Results on baseline health status are expected. The construct of spousal caregiver
subgroups, especially stressors, is based partially on care-recipients’ needs, such as functional
limitations, dementia status, and amount of care needed. As described in the previous chapter,
medium-stress high-support spousal caregivers are more likely to be taking care of a partner with
higher activity of daily living (Gerstorf, Hoppmann, Kadlec, et al.) needs, and the care-
recipients, correspondingly, could report worse self-rated health at baseline, while the differences
in depressive symptoms do not differ much. High-stress medium-support spousal caregivers, on
the other hand, are taking care of partners with much more ADL needs, more likely to be living
with dementia, and who require more hours of care, which indicate the care-recipients have
lower levels of physical health compared to their counterparts taken care of by low-stress low-
support spouses. The high level of financial and emotional stress among high-stress medium-
support spousal caregivers could also partially explain the higher levels of depressive symptoms
among care-recipients, since cohabitating older adults couples often share financial resources
(Bisdee et al., 2013; Halliday Hardie & Lucas, 2010), and financial strain is found to be
associated with worse mental health outcomes (Jones et al., 2019; Samuel et al., 2022). In
addition, previous research has also indicated that caregivers’ level of emotional stress is
negatively correlated with care-recipients’ mental health outcomes (Buck et al., 2015; Ejem et
al., 2015).
The rate of change over time is not significant for the whole sample, as indicated in
Model 1s. The one-year gap across observation waves might be too short to detect any
significant changes over time. Our sample includes community-dwelling older adults who are
77
taking care of by a spousal caregiver, with an overall moderate-high level of self-rated health,
and relatively low level of depressive symptoms at baseline. It is reasonable that there would not
be drastic changes in either domain over a year. The differences in rates of changes in health
across groups are not significant either, and there are four possible ways to explain the results,
which can be classified into two directions. The first interpretation is that the non-significance is
due to the short study period while there could be long-term differences across groups. Figure
4.1 captures some differences in magnitude and directions of health changes across groups: for
self-rated health, recipients taken care of by high-stress medium-support spousal caregivers
could experience a faster decline in self-rated health versus the other two groups; for depressive
symptoms, recipients taken care of by high-stress medium-support spousal caregivers could
accumulate more depressive symptoms at a higher rate, while those taken care of by medium-
stress high-support caregivers remain stable, or could even see a decrease in depressive
symptoms. The second direction to interpret the results is that even in the long-term, the
differences in rates of health changes do not differ across groups. First, similar to the findings
from the study conducted by Gellert and colleagues (2018), while the level of stress is negatively
correlated with quality of life in both dementia patients and their spousal caregivers, the
magnitude of the relation is much smaller in care-recipients. That is, although caregivers and
spousal care-recipients share some of the stressors and resources, the level of stress is not
pronounced in care-recipients’ appraisal of health or decline in health in the way it is in
caregivers’ health. Second, as shown in actor-partner interdependence models among caregiver-
recipient dyads, partner effect (the impact of one partner to his or her counterpart) is not
supported (Ayotte et al., 2010; Gellert et al., 2018). That is, care-recipients’ changes in health are
more closely related to the stressors and resources of their own rather than their spousal
78
caregivers’. Although no relevant literature has been identified, the third explanation under this
direction is that, care-recipients taken care of by high-stress medium-support caregivers could
have seen a faster decline in health; however, spousal caregivers have borne with the common
stressors they face, and have utilized their resources and social support to ensure the
corresponding care-recipients’ health is not declining faster than their counterparts taken care of
by other types of spousal caregivers. Future studies could increase the time interval between
study periods, or include more waves of data to further examine the differences in trajectories of
health changes across groups. In order to examine actor-partner effects, and the differences in
degrees of change associated with spousal caregiver classes, at least three waves of data for both
spousal caregivers and care-recipients are required, to further explore the spiral, interactive
changes in both parties’ health outcomes and their mutual impacts on each other.
Findings on sociodemographic factors are consistent with existing literature: compared to
non-Hispanic White, non-Hispanic Black are less likely to have depressive symptoms due to
more positive perception towards caregiving (Dilworth-Anderson et al., 2002; R. Liu et al.,
2021) and higher levels of social support from family and friends (Epps et al., 2021; Teja
Pristavec, 2019). In addition, educational attainment is still a driving factor contributing to health
disparities (Young et al., 2020). Care-recipients’ chronic conditions are also positively associated
with level of depression (Fiest et al., 2011) and negatively associated with self-rated health
(Molarius & Janson, 2002).
There are several limitations in the current study. First, outcome measures are self-
reported and subject to reporting errors. Second, although longitudinal data for care-recipient is
available, NSOC only surveys caregivers at one time point, and it is hard to decide whether
spousal caregivers continue caregiving tasks, and whether their class memberships have changed
79
over time. Therefore, although the current study have utilized two waves of data for each
participant, the association between care-recipients’ health and caregiving needs, hence caregiver
class membership, still cannot be controlled for. Third, relationship quality between spousal
caregivers and care-recipients cannot be controlled for, while the quality could be a third factor
that is associated with both spousal caregivers’ class membership, and care-recipients’ health and
health change. At last, sampling weights are not included in the analysis due to the exploratory
nature and the intent to ensure the internal validity of the results. However, the tradeoff is the
limited representativeness and generalizability of results.
The limitations notwithstanding, the current study is among the first to explore the
relationship between spousal caregivers’ co-occurring stressors and resources, and their care-
recipients’ health and health changes. Using a combination of relatively objective indicators to
separate spousal populations could benefit social workers and service providers in the
identification of peoples that are in need. The findings from this study, if confirmed by future
studies, highlight the importance of studying dyads that are in both spousal and caregiving
relationships. Spousal caregivers and their care-recipients could be simultaneously affected by
caregivers’ stressors and resources, and for care-recipients who have lower levels of health at
baseline but are not experiencing higher rates of decline, their spousal caregivers could be the
one sacrificing and are in need of more support to maintain both parties’ health.
80
Tables
Table 4.1. Sample characteristics and comparison across three spousal caregiver latent classes
Variables All
(n=639)
Class 1
(n=260;
40.69%)
Class 2
(n=226;
35.37%)
Class 3
(n=153;
23.94%)
Sig.
level
Outcomes
Self-rated health at T1 2.73 (1.02) 2.93 (1.02) 2.68 (0.97) 2.48 (1.01)
***
Self-rated health at T2 2.67 (0.93) 2.88 (0.98) 2.67 (0.82) 2.31 (0.91)
***
Changes in self-rated health -0.06 (0.87) -0.04 (0.77) -0.01 (0.89) -0.17 (0.98)
Depressive symptoms at T1 2.65 (2.76) 2.15 (2.40) 2.64 (2.65) 3.49 (3.27)
***
Depressive symptoms at T2 2.82 (2.67) 2.45 (0.42) 2.56 (2.45) 3.83 (3.12)
***
Changes in depressive symptoms 0.18 (2.45) 0.30 (2.03) -0.08 (2.34) 0.34 (3.14)
Control variables
Care-recipient’s age at T1 79.23 (6.45) 78.66 (6.74) 80.00 (6.34) 79.08 (7.04)
Care-recipient’s gender (female) 266 (41.63%) 120 (46.15%) 106 (46.90%) 40 (26.14%)
***
Care-recipient’s race/ethnicity
• non-Hispanic White
• non-Hispanic Black
• others
501 (79.40%)
103 (16.32%)
27 (4.28%)
216 (83.72%)
33 (12.79%)
9 (3.49%)
164 (73.54%)
52 (23.32%)
7 (3.14%)
121 (80.67%)
18 (12.00%)
11 (7.33%)
**
Care-recipient’s education (above high school) 333 (53.03%) 155 (60.08%) 102 (46.15%) 76 (51.01%)
**
Care-recipient’s number of comorbidities
(other than dementia) at T1
2.86 (1.51) 2.79 (1.53) 2.84 (1.45) 2.99 (1.51)
Household income at T1 (in 2016$) 74181.69
(216816.20)
97206.68
(331633.90)
57942.08
(50639.67)
59042.18
(67732.09)
Note: Class 1=low-stress low-support spousal caregivers; Class 2=medium-stress high-support spousal caregivers; Class 3=high-stress
medium-support spousal caregivers
Sig. level=level of significance;
*
p<.05,
**
p<.01,
***
p<.001
81
Table 4.2. Generalized estimating equation models of self-rated health over time across spousal
caregiver classes (N=639)
Self-rated health
Model 1 Model 2 Model 3
Baseline
health
status
Overall 2.73 (2.66, 2.81) 2.93 (2.81, 3.04) 2.04 (1.08, 2.30)
Class 1 Ref Ref
Class 2 -0.25 (-0.42, -0.08) -0.20 (-0.36, -0.04)
Class 3 -0.44 (-0.63, -0.25) -0.35 (-0.53, -0.17)
Rate of
change
Time -0.06 (-0.13, 0.01) -0.04 (-0.15, 0.06) -0.04 (-0.14, 0.07)
Class 1 * Time Ref Ref
Class 2 * Time 0.03 (-0.12, 0.19) 0.05 (-0.11, 0.20)
Class 3 * Time -0.12 (-0.30, 0.05) -0.11 (-0.29, 0.06)
Covariates Age 0.01 (0.00, 0.02)
Female 0.12 (-0.01, 0.24)
Race (ref: White)
• Black
• others
-0.24 (-0.41, -0.07)
-0.50 (-0.80, -0.20)
Education 0.23 (0.10, 0.35)
Comorbidities -0.17 (-0.21, -0.13)
Income 0.04 (-0.02, 0.09)
Note: Class 1=low-stress low-support spousal caregivers; Class 2=medium-stress high-support
spousal caregivers; Class 3=high-stress medium-support spousal caregivers
95% confidence interval in parenthesis; bolded values indicate statistical significance at p<.05
level based on Wald test
82
Table 4.3. Generalized estimating equation models of depressive symptoms over time across
spousal caregiver classes (N=639)
Depressive symptoms
Model 1 Model 2 Model 3
Baseline
health
status
Overall 2.65 (2.44, 2.86) 2.15 (1.83, 2.48) 5.43 (2.60, 8.26)
Class 1 Ref Ref
Class 2 0.49 (0.02, 0.97) 0.43 (-0.04, 0.91)
Class 3 1.34 (0.81, 1.87) 1.21 (0.68, 1.74)
Rate of
change
Time 0.18 (-0.01, 0.37) 0.30 (0.00, 0.60) 0.28 (-0.02, 0.58)
Class 1 * Time Ref Ref
Class 2 * Time -0.38 (-0.82, 0.05) -0.39 (-0.82, 0.05)
Class 3 * Time 0.04 (-0.45, 0.53) 0.02 (-0.47, 0.51)
Covariates Age -0.03 (-0.05, -0.00)
Female 0.09 (-0.28, 0.46)
Race (ref: White)
• Black
• others
-0.11 (-0.62, 0.39)
1.48 (0.59, 2.37)
Education -0.50 (-0.87, -0.12)
Comorbidities 0.27 (0.14, 0.39)
Income -0.15 (-0.32, 0.01)
Note: Class 1=low-stress low-support spousal caregivers; Class 2=medium-stress high-support
spousal caregivers; Class 3=high-stress medium-support spousal caregivers
95% confidence interval in parenthesis; bolded values indicate statistical significance at p<.05
level based on Wald test
83
Figures
Note: Class 1=low-stress low-support spousal caregivers; Class 2=medium-stress high-support
spousal caregivers; Class 3=high-stress medium-support spousal caregivers
Figure 4.1. Changes in health across classes
84
Chapter 5. Conclusions and Implications
Each individual has multiple identities and different experiences associated with the
identity. Intersectional perspective emphasizes the multidimensionality of lived experiences and
the interconnections of lived experiences (Crenshaw, 1989; Dilworth-Anderson et al., 2020; R.
Liu et al., 2021). Inspired by the perspective but with a less focus on the exercise of power based
on social identifies, the dissertation focuses on spousal caregivers, their caregiving experience,
and the inevitably related spousal care-recipients. Both spousal caregivers and spousal non-
caregivers are going through stresses associated with the decline of health in themselves and
their partners. The dissertation tries to disentangle the stress uniquely associated with the
provision of care, which characterizes the intersectional identity of spousal caregivers, who are
in both spousal and caregiving relationships. Within spousal caregivers, the dissertation
investigates their latent classes featured by the interactive stressors and resources, rather than
either of the dimensions in isolation, and further explores the associations with spousal
caregivers’ and care-recipients’ health outcomes.
This dissertation addresses three research questions regarding spousal caregivers: (1)
What are the health effects of spousal caregiving?; (2) Are there distinguishable spousal
caregiver latent classes based on the co-occurrence and relative intensities of stressors and
resources? If so, what are their associations with spousal caregivers’ health outcomes?; and (3)
Whether spousal care-recipients’ health and changes in health outcomes are associated with
spousal caregivers’ latent class memberships?
Featuring three unique but interrelated studies, the dissertation adapts and extends the
stress process model (Pearlin et al., 1981), and utilizes two nationally representative datasets.
The first study provides empirical evidence to understand the differences in health outcomes
85
between spousal caregivers and spousal non-caregivers, and the second study explores the
differences among spousal caregivers. Instead of applying existing models, the first study
dissects stressors, trying to isolate those applicable to both spousal caregivers and spousal non-
caregivers from stressors directly related to spousal caregiving. As such, spouses’ health
conditions prior to the uptake of caregiving are controlled for, partially addresses the selection
bias where healthier individuals enter or remain in caring activities, contributing to lower
mortalities among caregivers (Fredman et al., 2015). The control of the presence of a
spouse/partner in need of care also partially addresses the common stressor arisen from concerns
over a partner in declined health among both spousal caregivers and spousal non-caregivers,
regardless of care provision.
The second study zooms in the stress process model to explore the heterogeneity among
spousal caregivers based on how their stressors and resources interact in the process, which has
received little attention in existing studies (Donnelly et al., 2015). Focusing on individual
experience and the grouping of spousal caregivers, instead of the relative impact on health
outcomes of distinct constructs/variables, the study takes a person-centered approach to identify
spousal caregiver latent classes based on the relative intensities of several indicators representing
caregiving stressors and resources.
After the identification of spousal caregiver latent class memberships, the third study
considers the dynamic nature inherent in caregiving (T. Pristavec, 2019; Roberto & Jarrott, 2008)
and expands the stress process model to incorporate care-recipients’ health outcomes. It
examines how the identified spousal caregiver latent classes are correlated with care-recipients’
health outcomes. Specifically, the study utilizes a marginal model to explore differences in
health outcomes across groups both at baseline and rates of changes.
86
Echoing limitations in existing studies identified in Chapter 1, the dissertation addresses
the gaps in current research by (1) focusing exclusively on spousal caregivers, (2) utilizing a
person-centered approach to characterize subgroups, and (3) extending the population to include
care-recipients. Theoretically, the dissertation first tries to comprehend the stress process model
from a bird’s-eye view to isolate caregiving-related stressors from chronic stressors and stressors
not uniquely related to caregiving. After this, the dissertation zooms in the model with a focus on
the interaction between two domains – stressors and resources, and explores how the relative
intensities of co-occurrence characterize spousal caregiver sub-classes. Extending the model, the
dissertation also includes care-recipients’ health outcomes to contextualize the dyadic nature in
both spousal and caregiving relationships.
Summary of Major Research Findings
The first study focuses on the health impacts of spousal caregiving by partially
controlling for selection bias. Using pooled panel data from Wave 8 to Wave 13 in the Health
and Retirement Study, health outcomes between spousal caregivers and spousal non-caregivers
are compared under coarsened exact matching and regression analysis. Variables used for
matching are classified into three categories: care needs, the willingness to provide care, and the
ability to provide care. Health outcomes assessed include self-rated health, depressive symptoms,
and cognitive functioning. Through the comparison between 2,741 spousal caregivers and 18,043
matched spousal non-caregiver observations, spousal caregiving is found to be associated with
increased depressive symptoms after the common stressors applied to both spousal caregivers
and non-caregivers are controlled for. The results are consistent with previous studies addressing
mental health needs among spousal caregivers to persons with dementia (Adams, 2008; Joling et
al., 2015; Chen et al., 2020), cancer (Goldzweig et al., 2019; Li et al., 2013; Rhee et al., 2008), or
87
activities of daily living needs (Dunkle et al., 2014). Instead of focusing on a subgroup of
spousal caregivers to recipients with a certain type of conditions, the current study expands the
population to include the general spousal caregivers; instead of contrasting spousal caregivers to
socially active spousal non-caregiver counterparts who might experience fewer stressor, the
current study benefits from the nationally representative data and advanced statistical methods in
the generation of comparable controls. Results further confirms the elevated needs to address
mental health among spousal caregivers.
The second study of the dissertation aims to understand the diverse caregiving experience
within a relatively homogeneous group – spousal caregivers, by identifying latent classes
characterized by the relative intensities of co-occurring stressors and resources. Using data from
Round 5 and Round 7 of National Study of Caregiving (NSOC), and the linked data from the
National Health and Aging Trends Study (NHATS), latent class analyses are performed on 793
observations across 639 distinct spousal caregivers. With eight indicators of caregiving stressors
and three indicators of caregiving resources, the study identifies three spousal caregiver latent
classes: (1) low-stress low-support spousal caregivers, featured by exposure to the lowest level of
stressors including fewer caregiving hours, less chance of taking care of a partner with either
ADL or dementia, and lower odds of having financial and physical stresses; (2) medium-stress
high-support spousal caregivers, characterized by the high levels of resources they possess,
including very high levels of emotional and instrumental support received from family and
friends; and (3) high-stress medium-support spousal caregivers who are simultaneously exposed
to high levels of stressors and in possession of medium-high levels of resources. Although they
are receiving moderate-high levels of emotional support from family and friends, the odds of
having instrumental support are lower compared to medium-stress high-support caregivers. In
88
addition, compared to low-stress low-support spousal caregivers, high-stress medium-support
spousal caregivers have worse self-rated health and higher levels of depressive symptoms. The
overall spousal caregiver profiles and distribution across classes are similar to previous studies
using person-centered approaches (Perlick et al., 2008; Teja Pristavec, 2019; Sung et al., 2021),
which have partially validated the results. The current study generates fewer latent classes
compared to the previous ones. One possible explanation is the focus on spousal caregivers
rather than informal caregivers in general, sicne there is less variation among the relatively
homogeneous spousal caregiver subpopulation. The other explanation is the constructs of latent
class indicators: the previous two studies focus on negative and positive aspects of caregiving
using mostly subjective indicators, while the current study has more emphasis on tangible and
objective measures. The inclusion of resources corresponds to the stress process model as they
are proposed as buffers to stressors; the focus on objective measures could be more
operationalizable than subjective ones for social workers and service providers to identify the
group of people in need. In addition, the study further confirms the utilities of person-centered
approaches in the examination and exploration of stress process models, and highlights the
interactive nature of the stress process. Findings from the study could inform social workers and
policy makers to target spousal caregivers whose mobilizable resources are relatively low in
contrast to the stressors they face in the provision of services and policy recommendations.
Instead of focusing on a single aspect, such as low resources or high stressors, and consequently
subpopulations characterized by the single aspects, spousal caregivers who are experiencing gaps
in resources to cope with stressors are the most in need, and the gaps can be experienced across
people of all sociodemographic characteristics, regardless of their gender, race/ethnicity,
education, and income levels.
89
Few studies have examined the relationship between caregivers’ interactive stress process
and care-recipients’ health outcome. Building on the results from the second study, the third
study explores whether and how spousal caregivers’ latent classes are associated with care-
recipients health and changes in health. Another two rounds (Round 6 and Round 8) of data from
NHATS are pooled with the data included in the second study to allow for longitudinal
observations among care-recipients. 639 spousal care-recipients are included with data points at
both waves after multiple imputation. Using stepwise generalized estimating equations, spousal
care-recipients taken care of by medium-stress high-support caregivers are found to have lower
levels of self-rated health at baseline compared to those taken care of by low-stress caregivers;
recipients with high-stress medium-support caregivers have both worse self-rated health and
higher depressive symptoms at baseline compared to counterparts taken care of by low-stress
low-support caregivers. The findings on different levels of health at baseline among care-
recipients have highlighted the importance of studying the spousal dyads. Rates of health
changes are neither statistically significant among the whole sample, nor statistically
significantly different across groups. The short time interval (one-year) could partially explain
the insignificant changes in health. Differences across groups in directions and rates of health
changes are visually identified but not statistically confirmed. We also contemplate a plausible
explanation for care-recipients who have lower levels of health at baseline but are not
experiencing higher rates of decline. We conjecture that their spousal caregivers could be the
ones sacrificing and are in need of more support to maintain both parties’ health. Future research
is needed to further explore the rates of change in health and health trajectories, and the
potentially different rates of changes in health outcomes among care-recipients when they are
taken care of by spousal caregivers versus other family caregivers.
90
Implications for Future Research
There are two main challenges while trying to estimate the health effects of caregiving:
one is potential endogeneity and selection bias where healthier individuals are able to provide
care, the other is the isolation of the caregiving effects from family effects, which refer to the
impact of caring about someone, even in absence of care provision (Bom et al., 2019). The first
study in the dissertation compares spousal caregivers to spousal non-caregivers by partially
controlling for the common stressors and selection bias, and has found spousal caregiving is
associated with higher levels of depressive symptoms. The relative levels of association with
health decline between family effects and care provision could be further distinguished. Future
studies could compare the two effects to better understand the spillover health effects of having a
spouse/partner in need of care versus providing care to a spouse/partner.
The second study utilizes a person-centered approach to investigate spousal caregivers’
diverse experience characterized by their interactive stressors and resources, rather than either of
the dimensions in isolation. The results that spousal caregivers who are in possession of medium-
high levels of resources are simultaneously exposed to the highest level of stressors and
experiencing worse health outcomes have highlighted the practicality of a person-centered
intersectional approach to produce more inclusive findings.
Relatively few studies have used person-centered approaches that are beyond the
interaction of two or more variables to understand the heterogeneity among caregivers; studies
extending results on caregiver class memberships to include care-recipients are even fewer. The
dissertation is among the first that utilizes a person-centered latent class approach, and the first to
focus exclusively on spousal caregivers and their recipients. The findings have highlighted the
importance to adopt a dyadic approach to examine both parties in a dynamic relationship. The
91
examination of dyads could be particularly crucial for spousal caregivers and their recipients,
who are in both spousal and caring relationships, both of which are dyadic in nature.
Theoretically, the study explored another variation and another way to adapt Pearlin’s
stress process model (1981), which further confirmed the flexibility and adaptiveness of the
model. The four main domains in the stress process model are intercorrelated, and each of the
domain has various constructs. The dissertation acknowledges the multidimensionality of both
stressors and resources, and the interactive nature between them. Future studies could further
explore the relationship between identified spousal caregiver latent classes and health outcomes
in other domains, and use person-centered approaches to identify latent classes among other
types of caregivers, as well as the association between class membership and health or wellbeing.
Because of limitation in data availability, which includes the cross-sectional data for
spousal caregivers in the current study and the two waves of data which are one year apart for
care-recipients, the topics that can be explored are limited in the dissertation. Future studies
could explore the transition of spousal caregivers’ classes over time, and how it is related to both
spousal caregivers’ and care-recipients’ health outcomes, using either latent transition analysis,
or parallel process models. The relationship between spousal caregivers’ classes, or the transition
in classes, and both caregivers’ and care-recipients’ health trajectories could also be explored,
with the incorporation of trajectory modeling analyses. Using an actor-partner interdependence
model and comparing spousal caregiving dyads and other caregiving dyads (for example, care-
recipients and other family caregivers), future studies could further distinguish the dyadic
intercorrelations related to spousal dyads and caregiving dyads. Incorporating more waves of
data could also enable the identification of whether the negative effects of caregiving, as implied
in the dissertation, persist, attenuate, or exaggerate over time.
92
Due to the limitation in data, health concordance among spouses, and more specifically,
spouses in a caregiving relationship, is not incorporated in the current study. If given the
availabilities of data, future studies could further explore how health concordance, and the rate of
changes in levels of concordance, differ based on the classification of spousal caregivers. The
differences could further be explored by gender given implications from existing literature that
husbands and wives differ in the level and domain of partner effects.
Implications for Policies and Practice
Spousal caregiving is prevalent, and the prevalence increases with age (Ornstein et al.,
2019). As the population ages, the demand for informal spousal caregiving is expected to grow,
bringing both spousal caregivers and their care-recipients into the spotlight. The needs to better
support family caregivers have been recognized in the past two decades (Dawson et al., 2020);
the dissertation further confirms the necessity to provide support to family caregivers in general
and spousal caregivers in specific, and could inform supportive practice and policies.
Findings from the first study highlights the vulnerability of spousal caregivers as a group,
especially in the domain of mental health. Social workers and service providers could target
spousal caregivers with an emphasis to address their mental health needs. The National Family
Caregiver Support Program has included individual counseling and organization of support
groups in the service, increasing spousal caregivers’ awareness of, access to, and utilization of
such services could potentially decrease the detrimental effects of caregiving on their mental
health, which takes an initial step in the recognition and fulfilment of family caregivers’ needs.
The Caregiver Advise, Record, Enable (CARE) acts also formally document caregivers in
patients’ record and require hospitals to provide instructions for care after discharge (Dawson et
al., 2020; Reyes et al., 2021). Continuous programs are needed to broaden the availability and
93
diversity of supports the caregivers have access to, in order to directly or indirectly improve their
mental health outcomes. In addition, spousal caregivers and other family caregivers are more
likely to prioritize the care-recipients’ needs and delay their own health needs; interventions and
support programs could highlight the importance of self-care among caregivers by providing
educational materials and programs. Due to the limited spare time caregivers have, the design
and provision of both interventions and services should accommodate the constraints to be
engaging, time-saving, streamlined, and integrated.
The second study highlights the interactive health impact of caregivers’ relative
intensities of stressors and resources. The Recognize, Assist, Include, Support, and Engage
(RAISE) Family Caregivers Act addresses the need to recognize and support family caregivers in
a way that reflects their diverse needs. Findings from the study could inform social workers and
service providers that one way to address the diverse needs among spousal and other family
caregivers is to routinely assess their stressors and resources, so as to determine their ability to
continually provide care, and the health impact of care provision. Currently, caregiver
assessment is not well integrated in the health care delivery setting, the study findings have
further emphasized the importance to assess and address their unique needs that are beyond
single dimensions. At the micro-level, if the accumulation of stressors, partially featured by care-
recipients’ deteriorating health conditions, is inevitable, increasing the amount of resources that
spousal caregivers have access to, through the provision of emotional support such as peer
support groups, instrumental support such as the respite services, and financial support through
Medicaid Expansion, could potentially buffer the negative impacts on caregivers’ physical and
mental health.
94
The third study implies that caregiving could have an impact on care-recipients’ health,
and that caregivers’ and care-recipients’ profiles could potentially shape those of each other’s.
Most older adults prefer to “age in place” – to stay in their own homes as they get older and to
delay the relocation to long-term care settings (AARP, 2019; Bigonnesse & Chaudhury, 2020).
In order to accommodate the preference for both older adults and their spousal caregivers, Area
Agency on Aging and other social and community service providers should continue the
provision of personal care, assistance in household chores, and other health care services to older
adults and spousal caregivers as resources to alleviate caregiving stressors. Long-term care
insurance programs could also be considered to improve access to long-term services and
supports to relieve the burden on spousal caregivers to older adults (Reyes et al., 2021).
Interventions especially those adopting new technological innovations, healthcare providers, and
long-term care providers should also take a dyadic and family-centered perspective, to transit the
focus of care from either care-recipients or caregivers to the family, and to expand access to and
utilization of formal resources among spousal caregivers to older adults and the care-recipients.
Spousal caregiving has a negative impact on caregivers’ mental health. Moreover, it is
the combination of stressors and resources, rather than each construct alone, that is related to
spousal caregivers’ and recipients’ health outcomes: spousal caregivers and the corresponding
recipients have the worst health outcomes when stressors caregivers encounter are at a higher
level relative to the resources they possess. Social workers and policy makers should, therefore,
focus on mental health of spousal caregivers in general, and pay special attention to the spousal
dyads whose level of stressors encountered exceeds that of resources they can mobilize.
95
References
AARP. (2015). Caregiving in the U.S. AARP Public Policy Institute. Retrieved from
https://www.caregiving.org/wp-
content/uploads/2020/05/2015_CaregivingintheUS_Final-Report-June-4_WEB.pdf
AARP. (2019a). Long-term service and supports. AARP Public Policy Institute. Retrieved from
http://www.advancingstates.org/sites/nasuad/files/LTSS%20Fact%20Sheet%202019.pdf
AARP. (2019b). Valuing the invaluable: 2019 update. AARP Public Policy Institute. Retrieved
from https://www.aarp.org/content/dam/aarp/ppi/2019/11/valuing-the-invaluable-2019-
update-charting-a-path-forward.doi.10.26419-2Fppi.00082.001.pdf
Abdollahpour, I., Nedjat, S., Noroozian, M., Salimi, Y., & Majdzadeh, R. (2014, 2014/09/01).
Caregiver Burden: The Strongest Predictor of Self-Rated Health in Caregivers of Patients
With Dementia. Journal of Geriatric Psychiatry and Neurology, 27(3), 172-180.
https://doi.org/10.1177/0891988714524627
Adams, K. B. (2008, Jun). Specific effects of caring for a spouse with dementia: differences in
depressive symptoms between caregiver and non-caregiver spouses. Int Psychogeriatr,
20(3), 508-520. https://doi.org/10.1017/s1041610207006278
Adelman, R. D., Tmanova, L. L., Delgado, D., Dion, S., & Lachs, M. S. (2014). Caregiver
Burden: A Clinical Review. Jama, 311(10), 1052-1060.
https://doi.org/10.1001/jama.2014.304
Ahn, S., & Logan, J. G. (2022, 2022/01/01/). Perceived role overload and physical symptom
experience among caregivers of older adults: The moderating effect of social support.
Geriatric Nursing, 43, 197-205.
https://doi.org/https://doi.org/10.1016/j.gerinurse.2021.11.021
96
Alley, D., Suthers, K., & Crimmins, E. (2007). Education and Cognitive Decline in Older
Americans: Results From the AHEAD Sample. Research on Aging, 29(1), 73-94.
https://doi.org/10.1177/0164027506294245
Amirkhanyan, A. A., & Wolf, D. A. (2006). Parent Care and the Stress Process: Findings From
Panel Data. The Journals of Gerontology: Series B, 61(5), S248-S255.
https://doi.org/10.1093/geronb/61.5.S248
Apesoa-Varano, E. C., Tang-Feldman, Y., Reinhard, S. C., Choula, R., & Young, H. M. (2015).
Multi-cultural caregiving and caregiver interventions: A look back and a call for future
action. Generations, 39(4), 39-48.
Austin, P. C. (2009). Balance diagnostics for comparing the distribution of baseline covariates
between treatment groups in propensity‐score matched samples. Statistics in medicine,
28(25), 3083-3107.
Ayotte, B. J., Margrett, J. A., & Hicks-Patrick, J. (2010, 2010/03/01). Physical Activity in
Middle-aged and Young-old Adults: The Roles of Self-efficacy, Barriers, Outcome
Expectancies, Self-regulatory Behaviors and Social Support. Journal of Health
Psychology, 15(2), 173-185. https://doi.org/10.1177/1359105309342283
Badana, A. N. S., Marino, V., & Haley, W. E. (2017, 2019/07/01). Racial Differences in
Caregiving: Variation by Relationship Type and Dementia Care Status. Journal of aging
and health, 31(6), 925-946. https://doi.org/10.1177/0898264317743611
Baker, K. L., & Robertson, N. (2008, 2008/07/01). Coping with caring for someone with
dementia: Reviewing the literature about men. Aging & Mental Health, 12(4), 413-422.
https://doi.org/10.1080/13607860802224250
97
Beach, S. R., & Schulz, R. (2017, 2017/03/01). Family Caregiver Factors Associated with Unmet
Needs for Care of Older Adults [https://doi.org/10.1111/jgs.14547]. Journal of the
American Geriatrics Society, 65(3), 560-566.
https://doi.org/https://doi.org/10.1111/jgs.14547
Beeber, A. S., & Zimmerman, S. (2012, 2012/02/01). Adapting the Family Management Style
Framework for Families Caring for Older Adults With Dementia. Journal of Family
Nursing, 18(1), 123-145. https://doi.org/10.1177/1074840711427144
Bertrand, R. M., Saczynski, J. S., Mezzacappa, C., Hulse, M., Ensrud, K., & Fredman, L. (2012).
Caregiving and cognitive function in older women: Evidence for the healthy caregiver
hypothesis. Journal of aging and health, 24(1), 48-66.
Bisdee, D., Daly, T., & Price, D. (2013). Behind Closed Doors: Older Couples and the Gendered
Management of Household Money. Social Policy and Society, 12(1), 163-174.
https://doi.org/10.1017/S147474641200053X
Bobinac, A., van Exel, N. J. A., Rutten, F. F. H., & Brouwer, W. B. F. (2010, 2010/07/01/).
Caring for and caring about: Disentangling the caregiver effect and the family effect.
Journal of health economics, 29(4), 549-556.
https://doi.org/https://doi.org/10.1016/j.jhealeco.2010.05.003
Bom, J., Bakx, P., Schut, F., & van Doorslaer, E. (2019, 2019/01/01/). Health effects of caring
for and about parents and spouses. The Journal of the Economics of Ageing, 14, 100196.
https://doi.org/https://doi.org/10.1016/j.jeoa.2019.100196
Bookwala, J., Yee, J. L., & Schulz, R. (2000). Caregiving and Detrimental Mental and Physical
Health Outcomes. In G. M. Williamson, D. R. Shaffer, & P. A. Parmelee (Eds.), Physical
98
Illness and Depression in Older Adults: A Handbook of Theory, Research, and Practice
(pp. 93-131). Springer US. https://doi.org/10.1007/0-306-47178-7_6
Brandt, J., Spencer, M., & Folstein, M. (1988). The telephone interview for cognitive status.
Neuropsychiatry Neuropsychol Behav Neurol, 1(2), 111-117.
Buck, H. G., Mogle, J., Riegel, B., McMillan, S., & Bakitas, M. (2015, Dec). Exploring the
Relationship of Patient and Informal Caregiver Characteristics with Heart Failure Self-
Care Using the Actor-Partner Interdependence Model: Implications for Outpatient
Palliative Care. J Palliat Med, 18(12), 1026-1032. https://doi.org/10.1089/jpm.2015.0086
Bugliari, D., Carroll, J., Hayden, O., Hayes, J., Hurd, M., Karabatakis, A., Main, R., Marks, J.,
McCullough, C., & Meijer, E. (2021). RAND HRS Longitudinal File 2018 (V1)
Documentation.
Burton, L. C., Zdaniuk, B., Schulz, R., Jackson, S., & Hirsch, C. (2003). Transitions in spousal
caregiving. The Gerontologist, 43(2), 230-241.
Byrd, C. M., & Carter Andrews, D. J. (2016). Variations in students' perceived reasons for,
sources of, and forms of in-school discrimination: A latent class analysis. Journal of
School Psychology, 57, 1-14. https://doi.org/10.1016/j.jsp.2016.05.001
Calasanti, T. (2010). Gender Relations and Applied Research on Aging. The Gerontologist,
50(6), 720-734. https://doi.org/10.1093/geront/gnq085
Campbell, P., Wright, J., Oyebode, J., Job, D., Crome, P., Bentham, P., Jones, L., & Lendon, C.
(2008, Oct). Determinants of burden in those who care for someone with dementia. Int J
Geriatr Psychiatry, 23(10), 1078-1085. https://doi.org/10.1002/gps.2071
Capistrant, B. D. (2016). Caregiving for older adults and the caregivers’ health: An
epidemiologic review. Current Epidemiology Reports, 3(1), 72-80.
99
Chang, H.-Y., Chiou, C.-J., & Chen, N.-S. (2010, May-Jun). Impact of mental health and
caregiver burden on family caregivers' physical health. Archives of gerontology and
geriatrics, 50(3), 267-271. https://doi.org/10.1016/j.archger.2009.04.006
Chappell, N. L., Dujela, C., & Smith, A. (2014, 2015/08/01). Caregiver Well-Being:
Intersections of Relationship and Gender. Research on Aging, 37(6), 623-645.
https://doi.org/10.1177/0164027514549258
Chen, C., Thunell, J., & Zissimopoulos, J. (2020). Changes in physical and mental health of
Black, Hispanic, and White caregivers and non-caregivers associated with onset of
spousal dementia. Alzheimers Dement (N Y), 6(1), e12082.
https://doi.org/10.1002/trc2.12082
Chiao, C. Y., Wu, H. S., & Hsiao, C. Y. (2015). Caregiver burden for informal caregivers of
patients with dementia: A systematic review. International nursing review, 62(3), 340-
350.
Cohen, S. (2004). Social Relationships and Health [doi:10.1037/0003-066X.59.8.676]. American
Psychological Association.
Collins, L. M., & Lanza, S. T. (2009). Latent class and latent transition analysis: With
applications in the social, behavioral, and health sciences (Vol. 718). John Wiley &
Sons.
Costa-Font, J., Raut, N., & Van Houtven, C. (2021). Medicaid Expansion and the Mental Health
of Spousal Caregivers.
100
Crenshaw, K. (1989). Demarginalizing the Intersection of Race and Sex: A Black Feminist
Critique of Antidiscrimination Doctrine, Feminist Theory and Antiracist Politics. The
University of Chicago Legal Forum, 140, 139-167.
Crimmins, E. M., Kim, J. K., Langa, K. M., & Weir, D. R. (2011). Assessment of cognition
using surveys and neuropsychological assessment: the Health and Retirement Study and
the Aging, Demographics, and Memory Study. Journals of Gerontology Series B:
Psychological Sciences and Social Sciences, 66(suppl_1), i162-i171.
Daley, R. T., O'Connor, M. K., Shirk, S. D., & Beard, R. L. (2017, 2017/01/01/). ‘In this
together’ or ‘Going it alone’: Spousal dyad approaches to Alzheimer's. Journal of Aging
Studies, 40, 57-63. https://doi.org/https://doi.org/10.1016/j.jaging.2017.01.003
Davis, L. L., Chestnutt, D., Molloy, M., Deshefy-Longhi, T., Shim, B., & Gilliss, C. L. (2014,
2014/11/01). Adapters, Strugglers, and Case Managers: A Typology of Spouse
Caregivers. Qualitative Health Research, 24(11), 1492-1500.
https://doi.org/10.1177/1049732314548879
Dawson, W. D., Bangerter, L. R., & Splaine, M. (2020). The Politics of Caregiving: Taking
Stock of State-Level Policies to Support Family Caregivers. Public Policy & Aging
Report, 30(2), 62-66. https://doi.org/10.1093/ppar/praa005
de Vugt, M. E., Riedijk, S. R., Aalten, P., Tibben, A., van Swieten, J. C., & Verhey, F. R. (2006).
Impact of behavioural problems on spousal caregivers: a comparison between
Alzheimer's disease and frontotemporal dementia. Dement Geriatr Cogn Disord, 22(1),
35-41. https://doi.org/10.1159/000093102
101
del-Pino-Casado, R., Frías-Osuna, A., Palomino-Moral, P. A., Ruzafa-Martínez, M., & Ramos-
Morcillo, A. J. (2018). Social support and subjective burden in caregivers of adults and
older adults: A meta-analysis. PLoS One, 13(1), e0189874.
Diggle, P., Diggle, P. J., Heagerty, P., Liang, K.-Y., & Zeger, S. (2002). Analysis of longitudinal
data. Oxford university press.
Dilworth-Anderson, P., Brummett, B. H., Goodwin, P., Williams, S. W., Williams, R. B., &
Siegler, I. C. (2005). Effect of Race on Cultural Justifications for Caregiving. The
Journals of Gerontology: Series B, 60(5), S257-S262.
https://doi.org/10.1093/geronb/60.5.S257
Dilworth-Anderson, P., Moon, H., & Aranda, M. P. (2020). Dementia Caregiving Research:
Expanding and Reframing the Lens of Diversity, Inclusivity, and Intersectionality. The
Gerontologist, 60(5), 797-805. https://doi.org/10.1093/geront/gnaa050
Dilworth-Anderson, P., Williams, I. C., & Gibson, B. E. (2002, Apr). Issues of race, ethnicity,
and culture in caregiving research: a 20-year review (1980-2000). Gerontologist, 42(2),
237-272. https://doi.org/10.1093/geront/42.2.237
Donnelly, N. A., Hickey, A., Burns, A., Murphy, P., & Doyle, F. (2015). Systematic review and
meta-analysis of the impact of carer stress on subsequent institutionalisation of
community-dwelling older people. PLoS One, 10(6), e0128213.
https://doi.org/10.1371/journal.pone.0128213
Dunkle, R. E., Feld, S., Lehning, A. J., Kim, H., Shen, H.-W., & Kim, M. H. (2014). Does
becoming an ADL spousal caregiver increase the caregiver’s depressive symptoms?
Research on Aging, 36(6), 655-682.
102
Ejem, D., Bauldry, S., Bakitas, M., & Drentea, P. (2018). Caregiver burden, care recipient
depressive symptomology, and social exchange: Does race matter? Journal of palliative
care, 33(2), 100-108.
Ejem, D. B., Drentea, P., & Clay, O. J. (2015). The effects of caregiver emotional stress on the
depressive symptomatology of the care recipient. Aging & Mental Health, 19(1), 55-62.
Epps, F., Foster, K., Alexander, K., Brewster, G., Chester, M., Thornton, J., & Aycock, D. (2021,
2021/11/01). Perceptions and Attitudes Toward Dementia in Predominantly African
American Congregants. Journal of Applied Gerontology, 40(11), 1511-1516.
https://doi.org/10.1177/0733464820987350
Feast, A., Moniz-Cook, E., Stoner, C., Charlesworth, G., & Orrell, M. (2016, Nov). A systematic
review of the relationship between behavioral and psychological symptoms (BPSD) and
caregiver well-being. Int Psychogeriatr, 28(11), 1761-1774.
https://doi.org/10.1017/s1041610216000922
Fiest, K. M., Currie, S. R., Williams, J. V. A., & Wang, J. (2011, 2011/06/01/). Chronic
conditions and major depression in community-dwelling older adults. Journal of Affective
Disorders, 131(1), 172-178. https://doi.org/https://doi.org/10.1016/j.jad.2010.11.028
Fredman, L., Cauley, J. A., Hochberg, M., Ensrud, K. E., & Doros, G. (2010, May). Mortality
associated with caregiving, general stress, and caregiving-related stress in elderly women:
results of caregiver-study of osteoporotic fractures. J Am Geriatr Soc, 58(5), 937-943.
https://doi.org/10.1111/j.1532-5415.2010.02808.x
https://www.aarp.org/content/dam/aarp/ppi/2019/11/valuing-the-invaluable-2019-update-
charting-a-path-forward.doi.10.26419-2Fppi.00082.001.pdf
103
Fredman, L., Lyons, J. G., Cauley, J. A., Hochberg, M., & Applebaum, K. M. (2015, Sep). The
Relationship Between Caregiving and Mortality After Accounting for Time-Varying
Caregiver Status and Addressing the Healthy Caregiver Hypothesis. J Gerontol A Biol
Sci Med Sci, 70(9), 1163-1168. https://doi.org/10.1093/gerona/glv009
Freedman, V. A., DeMatteis, J., & Kasper, J. (2019). National Study of Caregiving (NSOC) I–III
Weighting Guide. NHATS technical paper, 24.
Fryer, L. K., & Shum, A. (2020). Person-centered approaches to explaining students’ cognitive
processing strategies. In Handbook of Strategies and Strategic Processing (pp. 361-372).
Routledge.
Garnett, B. R., Masyn, K. E., Austin, S. B., Miller, M., Williams, D. R., & Viswanath, K. (2014,
Aug). The intersectionality of discrimination attributes and bullying among youth: an
applied latent class analysis. J Youth Adolesc, 43(8), 1225-1239.
https://doi.org/10.1007/s10964-013-0073-8
Garrido, M. M., Kelley, A. S., Paris, J., Roza, K., Meier, D. E., Morrison, R. S., & Aldridge, M.
D. (2014). Methods for constructing and assessing propensity scores. Health services
research, 49(5), 1701-1720.
Gehr, T. J., Freiberger, E., Sieber, C. C., & Engel, S. A. (2021, 2021/09/06). A typology of
caregiving spouses of geriatric patients without dementia: caring, worried, desperate.
BMC geriatrics, 21(1), 483. https://doi.org/10.1186/s12877-021-02425-1
Gellert, P., Häusler, A., Suhr, R., Gholami, M., Rapp, M., Kuhlmey, A., & Nordheim, J. (2018).
Testing the stress-buffering hypothesis of social support in couples coping with early-
stage dementia. PLoS One, 13(1), e0189849-e0189849.
https://doi.org/10.1371/journal.pone.0189849
104
Gerstorf, D., Hoppmann, C. A., Anstey, K. J., & Luszcz, M. A. (2009). Dynamic links of
cognitive functioning among married couples: Longitudinal evidence from the Australian
Longitudinal Study of Ageing. Psychology and Aging, 24(2), 296-309.
https://doi.org/10.1037/a0015069
Gerstorf, D., Hoppmann, C. A., Kadlec, K. M., & McArdle, J. J. (2009). Memory and depressive
symptoms are dynamically linked among married couples: longitudinal evidence from the
AHEAD study. Developmental psychology, 45(6), 1595-1610.
https://doi.org/10.1037/a0016346
Goldzweig, G., Schapira, L., Baider, L., Jacobs, J., Andritsch, E., & Rottenberg, Y. (2019). Who
will care for the caregiver? Distress and depression among spousal caregivers of older
patients undergoing treatment for cancer. Supportive Care in Cancer, 27(11), 4221-4227.
Haley, W. E., LaMonde, L. A., Han, B., Narramore, S., & Schonwetter, R. (2000). Family
caregiving in hospice: effects on psychological and health functioning among spousal
caregivers of hospice patients with lung cancer or dementia. The Hospice Journal, 15(4),
1-18.
Halliday Hardie, J., & Lucas, A. (2010, 2010/10/01). Economic Factors and Relationship Quality
Among Young Couples: Comparing Cohabitation and Marriage
[https://doi.org/10.1111/j.1741-3737.2010.00755.x]. Journal of Marriage and Family,
72(5), 1141-1154. https://doi.org/https://doi.org/10.1111/j.1741-3737.2010.00755.x
Halpern, M. T., Fiero, M. H., & Bell, M. L. (2017). Impact of caregiver activities and social
supports on multidimensional caregiver burden: analyses from nationally-representative
surveys of cancer patients and their caregivers. Quality of Life Research, 26(6), 1587-
1595. http://www.jstor.org/stable/44853503
105
Hirst, M. (2005, Aug). Carer distress: a prospective, population-based study. Soc Sci Med, 61(3),
697-708. https://doi.org/10.1016/j.socscimed.2005.01.001
Huang, C. Y., Sousa, V. D., Perng, S. J., Hwang, M. Y., Tsai, C. C., Huang, M. H., & Yao, S. Y.
(2009, Feb). Stressors, social support, depressive symptoms and general health status of
Taiwanese caregivers of persons with stroke or Alzheimer's disease. J Clin Nurs, 18(4),
502-511. https://doi.org/10.1111/j.1365-2702.2008.02443.x
Iacus, S. M., King, G., & Porro, G. (2012). Causal inference without balance checking:
Coarsened exact matching. Political analysis, 20(1), 1-24.
Iecovich, E. (2014). Aging in place: From theory to practice. Anthropological notebooks, 20(1).
Jekel, K., Damian, M., Wattmo, C., Hausner, L., Bullock, R., Connelly, P. J., Dubois, B.,
Eriksdotter, M., Ewers, M., Graessel, E., Kramberger, M. G., Law, E., Mecocci, P.,
Molinuevo, J. L., Nygård, L., Olde-Rikkert, M. G., Orgogozo, J. M., Pasquier, F., Peres,
K., Salmon, E., Sikkes, S. A., Sobow, T., Spiegel, R., Tsolaki, M., Winblad, B., &
Frölich, L. (2015). Mild cognitive impairment and deficits in instrumental activities of
daily living: a systematic review. Alzheimers Res Ther, 7(1), 17.
https://doi.org/10.1186/s13195-015-0099-0
Joling, K. J., van Marwijk, H. W., Veldhuijzen, A. E., van der Horst, H. E., Scheltens, P., Smit,
F., & van Hout, H. P. (2015). The two-year incidence of depression and anxiety disorders
in spousal caregivers of persons with dementia: who is at the greatest risk? The American
Journal of Geriatric Psychiatry, 23(3), 293-303.
Jones, S. M. W., Nguyen, T., & Chennupati, S. (2019, 2020/06/01). Association of Financial
Burden With Self-Rated and Mental Health in Older Adults With Cancer. Journal of
aging and health, 32(5-6), 394-400. https://doi.org/10.1177/0898264319826428
106
Kaplan, L. (2001, 2001/01/01). A Couplehood Typology for Spouses of Institutionalized Persons
With Alzheimer's Disease: Perceptions of “We”–“I”* [https://doi.org/10.1111/j.1741-
3729.2001.00087.x]. Family Relations, 50(1), 87-98.
https://doi.org/https://doi.org/10.1111/j.1741-3729.2001.00087.x
Kasper, J. D., Freedman, V. A., Spillman, B. C., & Wolff, J. L. (2015, Oct). The
Disproportionate Impact Of Dementia On Family And Unpaid Caregiving To Older
Adults. Health Aff (Millwood), 34(10), 1642-1649.
https://doi.org/10.1377/hlthaff.2015.0536
Kaufman, J. E., Lee, Y., Vaughon, W., Unuigbe, A., & Gallo, W. T. (2019). Depression
associated with transitions into and out of spousal caregiving. The International Journal
of Aging and Human Development, 88(2), 127-149.
Kelly, M. E., Duff, H., Kelly, S., McHugh Power, J. E., Brennan, S., Lawlor, B. A., & Loughrey,
D. G. (2017, Dec 19). 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
Kenward, M. G., & Carpenter, J. (2007, 2007/06/01). Multiple imputation: current perspectives.
Statistical Methods in Medical Research, 16(3), 199-218.
https://doi.org/10.1177/0962280206075304
Khalaila, R., & Cohen, M. (2016). Emotional suppression, caregiving burden, mastery, coping
strategies and mental health in spousal caregivers. Aging & Mental Health, 20(9), 908-
917. https://doi.org/10.1080/13607863.2015.1055551
107
Kim, Y., Carver, C. S., Spiegel, D., Mitchell, H. R., & Cannady, R. S. (2017). Role of family
caregivers' self‐perceived preparedness for the death of the cancer patient in long‐term
adjustment to bereavement. Psycho‐oncology, 26(4), 484-492.
King, G., Nielsen, R., Coberley, C., Pope, J. E., & Wells, A. (2011). Comparative effectiveness
of matching methods for causal inference. Unpublished manuscript, Institute for
Quantitative Social Science, Harvard University, Cambridge, MA.
Kuzuya, M., Enoki, H., Hasegawa, J., Izawa, S., Hirakawa, Y., Shimokata, H., & Akihisa, I.
(2011, Apr). Impact of caregiver burden on adverse health outcomes in community-
dwelling dependent older care recipients. Am J Geriatr Psychiatry, 19(4), 382-391.
https://doi.org/10.1097/JGP.0b013e3181e9b98d
Lavela, S. L., & Ather, N. (2010, Mar). Psychological health in older adult spousal caregivers of
older adults. Chronic Illn, 6(1), 67-80. https://doi.org/10.1177/1742395309356943
Lazarus, R. S., & Folkman, S. (1984). Stress, appraisal, and coping. Springer publishing
company.
Lee, H. S., Brennan, P. F., & Daly, B. J. (2001, 2001/02/01). Relationship of empathy to
appraisal, depression, life satisfaction, and physical health in informal caregivers of older
adults [https://doi.org/10.1002/1098-240X(200102)24:1<44::AID-NUR1006>3.0.CO;2-
S]. Research in Nursing & Health, 24(1), 44-56.
https://doi.org/https://doi.org/10.1002/1098-240X(200102)24:1<44::AID-
NUR1006>3.0.CO;2-S
Lee, Y., Tang, F., Kim, K. H., & Albert, S. M. (2015). The Vicious Cycle of Parental Caregiving
and Financial Well-being: A Longitudinal Study of Women. The Journals of
Gerontology: Series B, 70(3), 425-431. https://doi.org/10.1093/geronb/gbu001
108
Leggett, A. N., Strominger, J., Robinson-Lane, S. G., & Maust, D. T. (2022, 2022/04/01).
Disparities in Health Care Task Participation and Provider Communication by Family
Caregiver Race. Journal of General Internal Medicine, 37(5), 1321-1324.
https://doi.org/10.1007/s11606-021-06766-w
Leung, D. Y. P., Chan, H. Y. L., Chiu, P. K. C., Lo, R. S. K., & Lee, L. L. Y. (2020). Source of
Social Support and Caregiving Self-Efficacy on Caregiver Burden and Patient's Quality
of Life: A Path Analysis on Patients with Palliative Care Needs and Their Caregivers.
International journal of environmental research and public health, 17(15), 5457.
https://doi.org/10.3390/ijerph17155457
Li, M., Mao, W., Chi, I., & Lou, V. W. Q. (2019, Feb). Geographical proximity and depressive
symptoms among adult child caregivers: social support as a moderator. Aging Ment
Health, 23(2), 205-213. https://doi.org/10.1080/13607863.2017.1399349
Li, Q., Mak, Y. W., & Loke, A. (2013). Spouses' experience of caregiving for cancer patients: a
literature review. International nursing review, 60(2), 178-187.
Liang, J., Aranda, M. P., & Lloyd, D. A. (2020, 2020/12/01). Association between Role
Overload and Sleep Disturbance among Dementia Caregivers: The Impact of Social
Support and Social Engagement. Journal of aging and health, 32(10), 1345-1354.
https://doi.org/10.1177/0898264320926062
Liang, K.-Y., & Zeger, S. L. (1986). Longitudinal data analysis using generalized linear models.
Biometrika, 73(1), 13-22. https://doi.org/10.1093/biomet/73.1.13
Liu, C., Marino, V. R., Howard, V. J., Haley, W. E., & Roth, D. L. (2021). Positive aspects of
caregiving in incident and long-term caregivers: Role of social engagement and
distress.Aging & Mental Health, 1-7.
109
Liu, R., Chi, I., & Wu, S. (2021, Oct 4). Caregiving Burden among Caregivers of People with
Dementia through the Lens of Intersectionality. Gerontologist.
https://doi.org/10.1093/geront/gnab146
Liu, Z., Heffernan, C., & Tan, J. (2020, 2020/10/10/). Caregiver burden: A concept analysis.
International Journal of Nursing Sciences, 7(4), 438-445.
https://doi.org/https://doi.org/10.1016/j.ijnss.2020.07.012
Longest, K. C., & Thoits, P. A. (2012, 2012/11/01). Gender, the Stress Process, and Health: A
Configurational Approach. Society and Mental Health, 2(3), 187-206.
https://doi.org/10.1177/2156869312451151
Lopez-Anuarbe, M., & Kohli, P. (2019, May 22). Understanding Male Caregivers' Emotional,
Financial, and Physical Burden in the United States. Healthcare (Basel), 7(2).
https://doi.org/10.3390/healthcare7020072
Lu, P., & Shelley, M. (2019, 2019/10/01). Why spouses depress each other: A cross-national
study to test the shared resource hypothesis in depressive symptom concordance within
older adult couples [https://doi.org/10.1111/aswp.12183]. Asian Social Work and Policy
Review, 13(3), 307-319. https://doi.org/https://doi.org/10.1111/aswp.12183
Mc Donnell, E., & Ryan, A. (2011, 2013/03/01). Male caregiving in dementia: A review and
commentary. Dementia, 12(2), 238-250. https://doi.org/10.1177/1471301211421235
McCarrey, A. C., An, Y., Kitner-Triolo, M. H., Ferrucci, L., & Resnick, S. M. (2016, Mar). Sex
differences in cognitive trajectories in clinically normal older adults. Psychol Aging,
31(2), 166-175. https://doi.org/10.1037/pag0000070
McEwen, B. S. (1998). Stress, adaptation, and disease: Allostasis and allostatic load. Annals of
the New York academy of sciences, 840(1), 33-44.
110
Mehri, N., Kinney, J., Brown, S., & Rajabi Rostami, M. (2021, Jul). Informal Caregiving and
All-Cause Mortality: A Meta-Analysis of Longitudinal Population-Based Studies. J Appl
Gerontol, 40(7), 781-791. https://doi.org/10.1177/0733464819893603
Melrose, K. L., Brown, G. D. A., & Wood, A. M. (2015, 2015/04/01/). When is received social
support related to perceived support and well-being? When it is needed. Personality and
Individual Differences, 77, 97-105.
https://doi.org/https://doi.org/10.1016/j.paid.2014.12.047
Meyler, D., Stimpson, J. P., & Peek, M. K. (2007, 2007/06/01/). Health concordance within
couples: A systematic review. Social Science & Medicine, 64(11), 2297-2310.
https://doi.org/https://doi.org/10.1016/j.socscimed.2007.02.007
Mitchell, M. M., & Knowlton, A. (2012). Caregiver role overload and network support in a
sample of predominantly low-income, African-American caregivers of persons living
with HIV/AIDS: a structural equation modeling analysis. AIDS and Behavior, 16(2), 278-
287.
Molarius, A., & Janson, S. (2002, Apr). Self-rated health, chronic diseases, and symptoms
among middle-aged and elderly men and women. J Clin Epidemiol, 55(4), 364-370.
https://doi.org/10.1016/s0895-4356(01)00491-7
Moon, H., & Dilworth-Anderson, P. (2015, Mar). Baby boomer caregiver and dementia
caregiving: findings from the National Study of Caregiving. Age Ageing, 44(2), 300-306.
https://doi.org/10.1093/ageing/afu119
Nylund, K. L., Asparouhov, T., & Muthén, B. O. (2007, 2007/10/23). Deciding on the Number
of Classes in Latent Class Analysis and Growth Mixture Modeling: A Monte Carlo
111
Simulation Study. Structural Equation Modeling: A Multidisciplinary Journal, 14(4),
535-569. https://doi.org/10.1080/10705510701575396
Okun, M. A., Yeung, E. W., & Brown, S. (2013, Jun). Volunteering by older adults and risk of
mortality: a meta-analysis. Psychol Aging, 28(2), 564-577.
https://doi.org/10.1037/a0031519
Oldenkamp, M., Hagedoorn, M., Slaets, J., Stolk, R., Wittek, R., & Smidt, N. (2016). Subjective
burden among spousal and adult-child informal caregivers of older adults: results from a
longitudinal cohort study. BMC geriatrics, 16(1), 208-208.
https://doi.org/10.1186/s12877-016-0387-y
Ornstein, K., & Gaugler, J. E. (2012, Oct). The problem with "problem behaviors": a systematic
review of the association between individual patient behavioral and psychological
symptoms and caregiver depression and burden within the dementia patient-caregiver
dyad. Int Psychogeriatr, 24(10), 1536-1552. https://doi.org/10.1017/s1041610212000737
Ornstein, K. A., Wolff, J. L., Bollens-Lund, E., Rahman, O.-K., & Kelley, A. S. (2019). Spousal
Caregivers Are Caregiving Alone In The Last Years Of Life. Health affairs (Project
Hope), 38(6), 964-972. https://doi.org/10.1377/hlthaff.2019.00087
Pearlin, L. I. (2010). The life course and the stress process: some conceptual comparisons. The
journals of gerontology. Series B, Psychological sciences and social sciences, 65B(2),
207-215. https://doi.org/10.1093/geronb/gbp106
Pearlin, L. I., Menaghan, E. G., Lieberman, M. A., & Mullan, J. T. (1981). The Stress Process.
Journal of Health and Social Behavior, 22(4), 337-356. https://doi.org/10.2307/2136676
Pearlin, L. I., Mullan, J. T., Semple, S. J., & Skaff, M. M. (1990). Caregiving and the stress
process: An overview of concepts and their measures. The Gerontologist, 30(5), 583-594.
112
Pendergrass, A., Mittelman, M., Graessel, E., Özbe, D., & Karg, N. (2019, 2019/11/02).
Predictors of the personal benefits and positive aspects of informal caregiving. Aging &
Mental Health, 23(11), 1533-1538. https://doi.org/10.1080/13607863.2018.1501662
Penning, M. J., & Wu, Z. (2016). Caregiver stress and mental health: Impact of caregiving
relationship and gender. The Gerontologist, 56(6), 1102-1113.
Perlick, D. A., Rosenheck, R. A., Miklowitz, D. J., Kaczynski, R., Link, B., Ketter, T.,
Wisniewski, S., Wolff, N., & Sachs, G. (2008, Jun). Caregiver burden and health in
bipolar disorder: a cluster analytic approach. J Nerv Ment Dis, 196(6), 484-491.
https://doi.org/10.1097/NMD.0b013e3181773927
Pillemer, S., Davis, J., & Tremont, G. (2018). Gender effects on components of burden and
depression among dementia caregivers. Aging & Mental Health, 22(9), 1162-1167.
Pinquart, M., & Sörensen, S. (2003a). Associations of stressors and uplifts of caregiving with
caregiver burden and depressive mood: a meta-analysis. The Journals of Gerontology
Series B: Psychological Sciences and Social Sciences, 58(2), P112-P128.
Pinquart, M., & Sörensen, S. (2003b). Differences between caregivers and noncaregivers in
psychological health and physical health: A meta-analysis [doi:10.1037/0882-
7974.18.2.250].
Pinquart, M., & Sörensen, S. (2005, Feb). Ethnic differences in stressors, resources, and
psychological outcomes of family caregiving: a meta-analysis. Gerontologist, 45(1), 90-
106. https://doi.org/10.1093/geront/45.1.90
Pinquart, M., & Sörensen, S. (2006a). Gender differences in caregiver stressors, social resources,
and health: An updated meta-analysis. The Journals of Gerontology Series B:
Psychological Sciences and Social Sciences, 61(1), P33-P45.
113
Pinquart, M., & Sörensen, S. (2006b). Helping caregivers of persons with dementia: which
interventions work and how large are their effects? International psychogeriatrics, 18(4),
577-595.
Pinquart, M., & Sörensen, S. (2007, Mar). Correlates of physical health of informal caregivers: a
meta-analysis. J Gerontol B Psychol Sci Soc Sci, 62(2), P126-137.
https://doi.org/10.1093/geronb/62.2.p126
Pinquart, M., & Sörensen, S. (2011, Mar). Spouses, adult children, and children-in-law as
caregivers of older adults: a meta-analytic comparison. Psychol Aging, 26(1), 1-14.
https://doi.org/10.1037/a0021863
Polenick, C. A., & DePasquale, N. (2019, May 17). Predictors of Secondary Role Strains Among
Spousal Caregivers of Older Adults With Functional Disability. Gerontologist, 59(3),
486-498. https://doi.org/10.1093/geront/gnx204
Polenick, C. A., Leggett, A. N., Webster, N. J., Han, B. H., Zarit, S. H., & Piette, J. D. (2020).
Multiple chronic conditions in spousal caregivers of older adults with functional
disability: Associations with caregiving difficulties and gains. The Journals of
Gerontology: Series B, 75(1), 160-172.
Pristavec, T. (2019). The Burden and Benefits of Caregiving: A Latent Class Analysis. The
Gerontologist, 59(6), 1078-1091. https://doi.org/10.1093/geront/gny022
Pristavec, T. (2019, May-Jun). The caregiving dyad: Do caregivers' appraisals of caregiving
matter for care recipients' health? Arch Gerontol Geriatr, 82, 50-60.
https://doi.org/10.1016/j.archger.2019.01.020
114
Pristavec, T., & Luth, E. A. (2020, Nov 13). Informal Caregiver Burden, Benefits, and Older
Adult Mortality: A Survival Analysis. J Gerontol B Psychol Sci Soc Sci, 75(10), 2193-
2206. https://doi.org/10.1093/geronb/gbaa001
Pudrovska, T., Schieman, S., Pearlin, L. I., & Nguyen, K. (2005, Oct). The sense of mastery as a
mediator and moderator in the association between economic hardship and health in late
life. J Aging Health, 17(5), 634-660. https://doi.org/10.1177/0898264305279874
Quiñones, A. R., Botoseneanu, A., Markwardt, S., Nagel, C. L., Newsom, J. T., Dorr, D. A., &
Allore, H. G. (2019). Racial/ethnic differences in multimorbidity development and
chronic disease accumulation for middle-aged adults. PLoS One, 14(6), e0218462.
https://doi.org/10.1371/journal.pone.0218462
Quiñones, A. R., Markwardt, S., & Botoseneanu, A. (2016). Multimorbidity combinations and
disability in older adults. Journals of Gerontology Series A: Biomedical Sciences and
Medical Sciences, 71(6), 823-830.
Rafnsson, S. B., Shankar, A., & Steptoe, A. (2017, 2017/01/02). Informal caregiving transitions,
subjective well-being and depressed mood: Findings from the English Longitudinal Study
of Ageing. Aging & Mental Health, 21(1), 104-112.
https://doi.org/10.1080/13607863.2015.1088510
Rawlings, G. H., & Bains, M. (2020). Experiences of depression in older adults from non-white
and ethnic minority groups: A thematic synthesis of qualitative studies. Journal of
Affective Disorders, 266, 341-348.
Reyes, A. M., Thunell, J., & Zissimopoulos, J. (2021). Addressing the Diverse Needs of Unpaid
Caregivers Through New Health-Care Policy Opportunities. Public Policy & Aging
Report, 31(1), 19-23. https://doi.org/10.1093/ppar/praa039
115
Rhee, Y. S., Yun, Y. H., Park, S., Shin, D. O., Lee, K. M., Yoo, H. J., Kim, J. H., Kim, S. O.,
Lee, R., & Lee, Y. O. (2008). Depression in family caregivers of cancer patients: the
feeling of burden as a predictor of depression. Journal of Clinical Oncology, 26(36),
5890-5895.
Ribé, J. M., Salamero, M., Pérez-Testor, C., Mercadal, J., Aguilera, C., & Cleris, M. (2018,
Mar). Quality of life in family caregivers of schizophrenia patients in Spain: caregiver
characteristics, caregiving burden, family functioning, and social and professional
support. Int J Psychiatry Clin Pract, 22(1), 25-33.
https://doi.org/10.1080/13651501.2017.1360500
Roberto, K. A., & Jarrott, S. E. (2008). Family caregivers of older adults: A life span
perspective. Family Relations, 57(1), 100-111.
Robinson, C. A., Bottorff, J. L., Pesut, B., Oliffe, J. L., & Tomlinson, J. (2014, 2014/09/01). The
Male Face of Caregiving: A Scoping Review of Men Caring for a Person With Dementia.
American Journal of Men's Health, 8(5), 409-426.
https://doi.org/10.1177/1557988313519671
Rosenfield, S. (2012, Jun). Triple jeopardy? Mental health at the intersection of gender, race, and
class. Soc Sci Med, 74(11), 1791-1801. https://doi.org/10.1016/j.socscimed.2011.11.010
Roth, D. L., Brown, S. L., Rhodes, J. D., & Haley, W. E. (2018, Jun). Reduced mortality rates
among caregivers: Does family caregiving provide a stress-buffering effect? Psychol
Aging, 33(4), 619-629. https://doi.org/10.1037/pag0000224
Roth, D. L., Fredman, L., & Haley, W. E. (2015, Apr). Informal caregiving and its impact on
health: a reappraisal from population-based studies. Gerontologist, 55(2), 309-319.
https://doi.org/10.1093/geront/gnu177
116
Roth, D. L., Haley, W. E., David Rhodes, J., Sheehan, O. C., Huang, J., Blinka, M. D., Yuan, Y.,
Irvin, M. R., Jenny, N., Durda, P., Cushman, M., Walston, J. D., & Howard, V. J. (2020,
Sep). Transitions to family caregiving: enrolling incident caregivers and matched non-
caregiving controls from a population-based study. Aging Clin Exp Res, 32(9), 1829-
1838. https://doi.org/10.1007/s40520-019-01370-9
Ruscio, J., & Ruscio, A. M. (2008). Categories and Dimensions Advancing Psychological
Science Through the Study of Latent Structure. Current directions in psychological
science, 17(3), 203-207. https://doi.org/10.1111/j.1467-8721.2008.00575.x
Samuel, L. J., Dwivedi, P., Hladek, M., Cudjoe, T. K. M., Drazich, B. F., Li, Q., & Szanton, S.
L. (2022, 2022/04/07). The effect of COVID-19 pandemic-related financial challenges on
mental health and well-being among US older adults [https://doi.org/10.1111/jgs.17808].
Journal of the American Geriatrics Society, n/a(n/a).
https://doi.org/https://doi.org/10.1111/jgs.17808
Schmitz, H., & Westphal, M. (2015). Short-and medium-term effects of informal care provision
on female caregivers’ health. Journal of health economics, 42, 174-185.
Schulz, R., & Beach, S. R. (1999, Dec 15). Caregiving as a risk factor for mortality: the
Caregiver Health Effects Study. Jama, 282(23), 2215-2219.
https://doi.org/10.1001/jama.282.23.2215
Schulz, R., Beach, S. R., Czaja, S. J., Martire, L. M., & Monin, J. K. (2020). Family Caregiving
for Older Adults. Annual review of psychology, 71, 635-659.
https://doi.org/10.1146/annurev-psych-010419-050754
117
Schulz, R., Beach, S. R., & Friedman, E. M. (2021, 2021/03/01/). Caregiving Factors as
Predictors of Care Recipient Mortality. The American Journal of Geriatric Psychiatry,
29(3), 295-303. https://doi.org/https://doi.org/10.1016/j.jagp.2020.06.025
Schulz, R., & Eden, J. (2016). In R. Schulz & J. Eden (Eds.), Families Caring for an Aging
America. National Academies Press (US)
Copyright 2016 by the National Academy of Sciences. All rights reserved.
https://doi.org/10.17226/23606
Schulz, R., & Sherwood, P. R. (2008). Physical and mental health effects of family caregiving.
The American journal of nursing, 108(9 Suppl), 23-27.
https://doi.org/10.1097/01.NAJ.0000336406.45248.4c
Skolarus, L. E., Freedman, V. A., Feng, C., Wing, J. J., & Burke, J. F. (2016). Care Received by
Elderly US Stroke Survivors May Be Underestimated. Stroke, 47(8), 2090-2095.
https://doi.org/10.1161/STROKEAHA.116.012704
Slaboda, J. C., Nelson, S. H., Agha, Z., & Norman, G. J. (2021, 2021/01/29). A national survey
of caregiver’s own experiences and perceptions of U.S. health care system when
addressing their health and caring for an older adult. BMC Health Services Research,
21(1), 101. https://doi.org/10.1186/s12913-021-06086-z
Smith, V. A., Lindquist, J., Miller, K. E. M., Shepherd-Banigan, M., Olsen, M., Campbell-
Kotler, M., Henius, J., Kabat, M., & Van Houtven, C. H. (2019). Comprehensive Family
Caregiver Support and Caregiver Well-Being: Preliminary Evidence From a Pre-post-
survey Study With a Non-equivalent Control Group. Front Public Health, 7, 122.
https://doi.org/10.3389/fpubh.2019.00122
118
Spetz, J., Trupin, L., Bates, T., & Coffman, J. M. (2015, Jun). Future Demand For Long-Term
Care Workers Will Be Influenced By Demographic And Utilization Changes. Health Aff
(Millwood), 34(6), 936-945. https://doi.org/10.1377/hlthaff.2015.0005
Stall, N. M., Kim, S. J., Hardacre, K. A., Shah, P. S., Straus, S. E., Bronskill, S. E., Lix, L. M.,
Bell, C. M., & Rochon, P. A. (2019, Mar). Association of Informal Caregiver Distress
with Health Outcomes of Community-Dwelling Dementia Care Recipients: A Systematic
Review. J Am Geriatr Soc, 67(3), 609-617. https://doi.org/10.1111/jgs.15690
Steffick, D. E., Wallace, R. B., & Herzog, A. R. (2000). Documentation of affective functioning
measures in the Health and Retirement Study. Ann Arbor, MI: University of Michigan.
Stockdale, S. E., Lagomasino, I. T., Siddique, J., McGuire, T., & Miranda, J. (2008, Jul). Racial
and ethnic disparities in detection and treatment of depression and anxiety among
psychiatric and primary health care visits, 1995-2005. Med Care, 46(7), 668-677.
https://doi.org/10.1097/MLR.0b013e3181789496
Stuart, E. A. (2010). Matching methods for causal inference: A review and a look forward.
Statistical science: a review journal of the Institute of Mathematical Statistics, 25(1), 1.
Sung, P., Goh, V. S., Azman, N. D., Visaria, A., & Malhotra, R. (2021). Types of Caregiving
Experience and Their Association With Caregiver Depressive Symptoms and Quality of
Life. Journal of aging and health, 08982643211051568.
https://doi.org/10.1177/08982643211051568
Swinkels, J., Tilburg, T. v., Verbakel, E., & Broese van Groenou, M. (2019). Explaining the
Gender Gap in the Caregiving Burden of Partner Caregivers. The Journals of
Gerontology: Series B, 74(2), 309-317. https://doi.org/10.1093/geronb/gbx036
119
Torres, O. V., & O'Dell, L. E. (2016, Feb 4). Stress is a principal factor that promotes tobacco
use in females. Prog Neuropsychopharmacol Biol Psychiatry, 65, 260-268.
https://doi.org/10.1016/j.pnpbp.2015.04.005
Turvey, C. L., Schultz, S. K., Beglinger, L., & Klein, D. M. (2009). A longitudinal community-
based study of chronic illness, cognitive and physical function, and depression. The
American Journal of Geriatric Psychiatry, 17(8), 632-641.
Turvey, C. L., Wallace, R. B., & Herzog, R. (1999). A revised CES-D measure of depressive
symptoms and a DSM-based measure of major depressive episodes in the elderly.
International psychogeriatrics, 11(2), 139-148.
Vaingankar, J. A., Chong, S. A., Abdin, E., Picco, L., Jeyagurunathan, A., Zhang, Y.,
Sambasivam, R., Chua, B. Y., Ng, L. L., Prince, M., & Subramaniam, M. (2016, Feb).
Care participation and burden among informal caregivers of older adults with care needs
and associations with dementia. Int Psychogeriatr, 28(2), 221-231.
https://doi.org/10.1017/s104161021500160x
Vitaliano, P. P. (2010). An ironic tragedy: are spouses of persons with dementia at higher risk for
dementia than spouses of persons without dementia? Journal of the American Geriatrics
Society, 58(5), 976-978. https://doi.org/10.1111/j.1532-5415.2010.02843.x
Vitaliano, P. P., Zhang, J., & Scanlan, J. M. (2003, Nov). Is caregiving hazardous to one's
physical health? A meta-analysis. Psychol Bull, 129(6), 946-972.
https://doi.org/10.1037/0033-2909.129.6.946
Walls, M. L., & Whitbeck, L. B. (2012). Advantages of stress process approaches for measuring
historical trauma [doi:10.3109/00952990.2012.694524]. Informa Healthcare.
120
Wang, M. (2014, 2014/12/01). Generalized Estimating Equations in Longitudinal Data Analysis:
A Review and Recent Developments. Advances in Statistics, 2014, 303728.
https://doi.org/10.1155/2014/303728
Weller, B. E., Bowen, N. K., & Faubert, S. J. (2020, 2020/05/01). Latent Class Analysis: A
Guide to Best Practice. Journal of Black Psychology, 46(4), 287-311.
https://doi.org/10.1177/0095798420930932
Weuve, J., Barnes, L. L., Mendes de Leon, C. F., Rajan, K. B., Beck, T., Aggarwal, N. T.,
Hebert, L. E., Bennett, D. A., Wilson, R. S., & Evans, D. A. (2018, Jan). Cognitive Aging
in Black and White Americans: Cognition, Cognitive Decline, and Incidence of
Alzheimer Disease Dementia. Epidemiology, 29(1), 151-159.
https://doi.org/10.1097/ede.0000000000000747
Wharton, T., & Zivin, K. (2017). Relationships Among Caregiving, Income, Gender, and Health:
A Cross-Sectional Examination of a Representative Sample of Older Americans. J Soc
Serv Res, 43(1), 141-148. https://doi.org/10.1080/01488376.2016.1239597
Wolff, J. L., & Kasper, J. D. (2006). Caregivers of Frail Elders: Updating a National Profile. The
Gerontologist, 46(3), 344-356. https://doi.org/10.1093/geront/46.3.344
Xiong, C., Biscardi, M., Astell, A., Nalder, E., Cameron, J. I., Mihailidis, A., & Colantonio, A.
(2020). Sex and gender differences in caregiving burden experienced by family
caregivers of persons with dementia: A systematic review. PLoS One, 15(4), e0231848-
e0231848. https://doi.org/10.1371/journal.pone.0231848
Yee, J. L., & Schulz, R. (2000). Gender differences in psychiatric morbidity among family
caregivers: A review and analysis. The Gerontologist, 40(2), 147-164.
121
Young, H. M., Bell, J. F., Whitney, R. L., Ridberg, R. A., Reed, S. C., & Vitaliano, P. P. (2020).
Social Determinants of Health: Underreported Heterogeneity in Systematic Reviews of
Caregiver Interventions. The Gerontologist, 60(Supplement_1), S14-S28.
https://doi.org/10.1093/geront/gnz148
Yu, Y., Tang, B.-w., Liu, Z.-w., Chen, Y.-m., Zhang, X.-y., & Xiao, S. (2018). Who cares for the
schizophrenia individuals in rural China—A profile of primary family caregivers.
Comprehensive psychiatry, 84, 47-53.
Zarit, S. H. (2018, 2018/06/03). Past is prologue: how to advance caregiver interventions. Aging
& Mental Health, 22(6), 717-722. https://doi.org/10.1080/13607863.2017.1328482
Zeger, S. L., & Liang, K.-Y. (1986). Longitudinal Data Analysis for Discrete and Continuous
Outcomes. Biometrics, 42(1), 121-130. https://doi.org/10.2307/2531248
Zodikoff, B. D. (2007). Gender differences in the community service use attitudes of older
spousal caregiver-care recipient couples. Home Health Care Serv Q, 26(2), 1-20.
https://doi.org/10.1300/J027v26n02_01
122
Appendix
Table 1. Comparison between matched and unmatched spousal caregivers for Chapter 2
Cohort
Variables
Matched
(N=535)
Unmatched
(n=142)
Sig.
Level
1 Self-rated health at T1 3.19 (1.07) 2.57 (1.09)
***
Self-rated health at T2 3.01 (1.06) 2.76 (1.17)
*
Change in self-rated health -0.18 (0.74) 0.19 (1.07)
***
Depressive symptoms at T1 1.20 (1.66) 2.55 (2.41)
***
Depressive symptoms at T2 1.52 (1.83) 2.24 (2.37)
**
Chang in depressive symptoms 0.31 (1.58) -0.31 (2.56)
**
Cognitive functioning at T1 14.94 (4.05) 11.44 (3.76)
***
Cognitive functioning at T2 14.49 (4.07) 11.68 (3.54)
***
Change in cognitive functioning -0.44 (3.28) 0.15 (3.79)
Spouse’s gender (female) 287 (53.64%) 71 (50.00%)
Spouse’s race
• Non-Hispanic White
• Non-Hispanic Black
• Hispanics
422 (78.88%)
59 (11.03%)
54 (10.09)
89 (62.68%)
23 (16.20%)
30 (21.13%)
***
Spouse’s age at T1 71.00 (7.59) 73.46 (9.97)
*
Partner’s age at T1 72.45 (7.33) 73.30 (8.50)
Spouse’s education (college and
above)
171 (31.96%) 37 (26.06%)
Spouse’s comorbidities 2.27 (1.42) 2.49 (1.56)
Partner’s comorbidities 2.64 (1.45) 3.21 (1.53)
**
Spouse’s ADL at T1 0.30 (0.81) 0.54 (1.28)
**
Spouse’s IADL at T1 0.05 (0.23) 0.13 (0.34)
**
Partner’s ADL at T1 0.24 (0.66) 0.92 (1.31)
***
Partner’s IADL at T1 0.09 (0.37) 0.35 (0.65)
***
Household income (quintile) 3.47 (1.10) 2.92 (1.25)
***
Length of current marriage at T1 42.82 (15.43) 44.45 (16.42)
Number of living children at T1 3.68 (2.03) 3.62 (2.01)
2 Self-rated health at T1 3.22 (0.99) 2.62 (1.06)
***
Self-rated health at T2 3.11 (0.98) 2.74 (1.10)
**
Change in self-rated health -0.12 (0.84) 0.13 (0.83)
**
Depressive symptoms at T1 1.09 (1.66) 2.07 (2.24)
***
Depressive symptoms at T2 1.53 (2.04) 1.50 (2.04)
Chang in depressive symptoms 0.48 (1.70) -0.43 (2.53)
***
Cognitive functioning at T1 15.12 (3.60) 12.52 (4.72)
***
Cognitive functioning at T2 14.36 (4.08) 13.16 (4.71)
**
Change in cognitive functioning -0.83 (3.39) 0.47 (3.11)
***
Spouse’s gender (female) 465 (58.71%) 90 (60.40%)
Spouse’s race
• Non-Hispanic White
• Non-Hispanic Black
• Hispanics
626 (79.04%)
67 (8.46%)
99 (12.50%)
76 (51.01%)
20 (13.42%)
53 (35.57%)
***
123
Spouse’s age at T1 71.83 (8.29) 71.45 (9.79)
Partner’s age at T1 73.79 (7.44) 72.75 (9.26)
Spouse’s education (college and
above)
289 (36.49%) 32 (21.48%)
***
Spouse’s comorbidities 2.35 (1.29) 2.63 (1.62)
*
Partner’s comorbidities 2.80 (1.42) 3.01 (1.30)
Spouse’s ADL at T1 0.21 (0.66) 0.51 (1.02)
***
Spouse’s IADL at T1 0.06 (0.29) 0.14 (0.45)
**
Partner’s ADL at T1 0.28 (0.79) 0.99 (1.06)
***
Partner’s IADL at T1 0.13 (0.45) 0.26 (0.55)
**
Household income (quintile) 3.41 (1.16) 2.46 (1.24)
***
Length of current marriage at T1 41.39 (15.67) 42.17 (15.67)
Number of living children at T1 3.92 (2.16) 4.01 (2.61)
3 Self-rated health at T1 3.22 (0.95) 2.74 (1.00)
***
Self-rated health at T2 3.08 (1.03) 2.95 (0.99)
Change in self-rated health -0.14 (0.91) 0.17 (0.82)
**
Depressive symptoms at T1 1.01 (1.57) 2.71 (2.42)
***
Depressive symptoms at T2 1.45 (1.87) 2.03 (2.01)
**
Chang in depressive symptoms 0.43 (1.67) -0.57 (2.35)
***
Cognitive functioning at T1 15.33 (3.75) 13.00 (4.82)
***
Cognitive functioning at T2 14.64 (4.09) 12.38 (4.47)
***
Change in cognitive functioning -0.71 (3.78) -0.69 (3.83)
Spouse’s gender (female) 334 (64.23%) 55 (43.65%)
***
Spouse’s race
• Non-Hispanic White
• Non-Hispanic Black
• Hispanics
418 (80.38%)
66 (12.69%)
36 (6.92%)
62 (49.21%)
37 (29.37%)
27 (21.43%)
***
Spouse’s age at T1 71.52 (9.23) 68.56 (11.09)
**
Partner’s age at T1 73.88 (8.66) 70.67 (8.84)
**
Spouse’s education (college and
above)
209 (40.19%) 43 (34.13%)
Spouse’s comorbidities 2.46 (1.25) 2.57 (1.23)
Partner’s comorbidities 3.06 (1.38) 3.17 (1.67)
Spouse’s ADL at T1 0.21 (0.62) 0.35 (0.73)
*
Spouse’s IADL at T1 0.08 (0.35) 0.26 (0.48)
***
Partner’s ADL at T1 0.28 (0.79) 0.96 (1.39)
***
Partner’s IADL at T1 0.09 (0.32) 0.76 (0.89)
***
Household income (quintile) 3.54 (1.10) 3.13 (1.25)
**
Length of current marriage at T1 41.94 (16.89) 40.60 (16.31)
Number of living children at T1 3.68 (2.14) 4.17 (2.34)
*
4 Self-rated health at T1 3.24 (1.00) 2.95 (1.06)
**
Self-rated health at T2 3.16 (1.04) 2.69 (0.84)
***
Change in self-rated health -0.07 (0.97) -0.26 (0.84)
Depressive symptoms at T1 1.02 (1.47) 2.05 (2.45)
***
Depressive symptoms at T2 1.46 (1.89) 1.95 (2.21)
*
Chang in depressive symptoms 0.44 (1.60) -0.14 (2.45)
**
124
Cognitive functioning at T1 15.19 (4.13) 11.67 (4.17)
***
Cognitive functioning at T2 14.53 (4.27) 11.37 (4.24)
***
Change in cognitive functioning -0.71 (3.68) -0.23 (3.42)
Spouse’s gender (female) 257 (52.88%) 69 (57.98%)
Spouse’s race
• Non-Hispanic White
• Non-Hispanic Black
• Hispanics
330 (67.90%)
81 (16.67%)
75 (15.43%)
50 (42.02%)
28 (23.53%)
41 (34.45%)
***
Spouse’s age at T1 70.58 (9.10) 69.22 (13.60)
Partner’s age at T1 72.18 (8.48) 71.62 (12.01)
Spouse’s education (college and
above)
201 (41.36%) 44 (36.97%)
Spouse’s comorbidities 2.24 (1.33) 2.34 (1.74)
Partner’s comorbidities 2.65 (1.49) 3.03 (1.42)
*
Spouse’s ADL at T1 0.21 (0.63) 0.34 (1.04)
Spouse’s IADL at T1 0.07 (0.31) 0.34 (0.65)
***
Partner’s ADL at T1 0.20 (0.57) 0.98 (1.13)
***
Partner’s IADL at T1 0.08 (0.30) 0.34 (0.56)
***
Household income (quintile) 3.53 (1.21) 2.78 (1.40)
***
Length of current marriage at T1 41.87 (16.00) 40.21 (21.40)
Number of living children at T1 3.66 (2.02) 4.16 (2.21)
*
5 Self-rated health at T1 3.25 (0.86) 2.71 (1.15)
***
Self-rated health at T2 3.14 (0.92) 2.80 (1.13)
**
Change in self-rated health -0.11 (0.77) 0.09 (1.11)
*
Depressive symptoms at T1 0.90 (1.38) 2.33 (2.47)
***
Depressive symptoms at T2 1.12 (1.83) 1.76 (2.23)
**
Chang in depressive symptoms 0.22 (1.83) -0.41 (1.61)
**
Cognitive functioning at T1 15.18 (4.50) 12.73 (4.62)
***
Cognitive functioning at T2 14.90 (4.29) 12.87 (5.29)
**
Change in cognitive functioning -0.32 (3.72) 0.04 (3.91)
Spouse’s gender (female) 218 (57.22%) 59 (59.00%)
Spouse’s race
• Non-Hispanic White
• Non-Hispanic Black
• Hispanics
275 (72.75%)
45 (11.90%)
58 (15.34%)
45 (45.00%)
19 (19.00%)
36 (36.00%)
***
Spouse’s age at T1 71.34 (9.46) 68.04 (11.45)
**
Partner’s age at T1 73.27 (8.71) 70.13 (9.81)
**
Spouse’s education (college and
above)
183 (48.03%) 45 (45.00%)
Spouse’s comorbidities 2.28 (1.30) 2.52 (1.38)
Partner’s comorbidities 2.73 (1.48) 2.79 (1.43)
Spouse’s ADL at T1 0.22 (0.66) 0.65 (1.27)
***
Spouse’s IADL at T1 0.08 (0.43) 0.32 (0.83)
**
Partner’s ADL at T1 0.25 (0.76) 0.83 (1.04)
***
Partner’s IADL at T1 0.08 (0.29) 0.36 (0.66)
***
Household income (quintile) 3.62 (1.18) 2.76 (1.27)
***
125
Length of current marriage at T1 42.59 (17.00) 38.51 (16.58)
*
Number of living children at T1 3.60 (2.09) 3.81 (1.99)
Note: Sig. level: level of significance
*
p<.05,
**
p<.01,
***
p<.001
126
Table 2. Comparison in baseline characteristics between complete cases and missing cases for Chapter 4
Variables Complete cases
(n=509; 79.66%)
Missing cases
(n=130; 20.34%)
Sig.
level
Outcomes
Self-rated health at T1 2.81 (1.00) 2.42 (1.02)
***
Depressive symptoms at T1 2.53 (2.65) 3.10 (3.31)
*
Spousal caregiver latent classes
• High-support SCG
• Low-stress SCG
• Intensive SCG
210 (41.26%)
185 (36.35%)
114 (22.40%)
50 (38.36%)
41 (31.54%)
39 (30.00%)
Control variables
Care-recipient’s age at T1 79.03 (6.87) 80.04 (7.47)
Care-recipient’s gender (female) 218 (42.83%) 48 (36.92%)
Care-recipient’s race/ethnicity
• non-Hispanic White
• non-Hispanic Black
• others
400 (79.52%)
86 (17.10%)
17 (3.38%)
101 (78.91%)
17 (13.28%)
10 (7.81%)
Care-recipient’s education (above high school) 235 (47.00%) 68 (53.12%)
Care-recipient’s number of comorbidities
(other than dementia) at T1
2.83 (1.49) 2.94 (1.60)
Household income at T1 (in 2016$) 76080.62 (240120.90) 66746.66 (73272.71)
Note: Sig. level=level of significance;
*
p<.05,
**
p<.01,
***
p<.001
127
Table 3. Generalized estimating equation models of self-rated health over time across spousal
caregiver classes on complete cases for Chapter 4 (n=509)
Self-rated health
Model 1 Model 2 Model 3
Baseline
health
status
Overall 2.81 (2.73, 2.90) 3.03 (2.90, 3.16) 2.50 (1.42, 3.60)
Class 1 Ref Ref
Class 2 -0.29 (-0.48, -0.10) -0.21 (-0.40, -0.03)
Class 3 -0.50 (-0.72, -0.28) -0.37 (-0.58, -0.16)
Rate of
change
Time -0.07 (-0.15, 0.01) -0.05 (-0.17, 0.08) -0.04 (-0.16, 0.09)
Class 1 * Time Ref Ref
Class 2 * Time 0.03 (-0.16, 0.21) 0.04 (-0.14, 0.22)
Class 3 * Time -0.15 (-0.36, 0.06) -0.14 (-0.35, 0.07)
Covariates Age 0.01 (-0.00, 0.02)
Female 0.11 (-0.03, 0.25)
Race (ref: White)
• Black
• others
-0.26 (-0.44, -0.07)
-0.56 (-0.94, -0.18)
Education 0.23 (0.08, 0.37)
Comorbidities -0.18 (-0.22, -0.13)
Income 0.01 (-0.05, 0.08)
Note: Class 1 = Low-stress low-support spousal caregivers; Class 2 = Medium-stress high-
support spousal caregivers; Class 3 = High-stress medium-support spousal caregivers
95% confidence interval in parenthesis; bolded values indicate statistical significance at p<.05
level based on Wald test
128
Table 4. Generalized estimating equation models of depressive symptoms over time across
spousal caregiver classes on complete cases for Chapter 4(n=509)
Depressive symptoms
Model 1 Model 2 Model 3
Baseline
health
status
Overall 2.53 (2.30, 2.77) 2.10 (1.74, 2.47) 2.92 (-0.27, 6.10)
Class 1 Ref Ref
Class 2 0.45 (-0.08, 0.98) 0.39 (-0.13, 0.92)
Class 3 1.18 (0.57, 1.79) 1.04 (0.42, 1.65)
Rate of
change
Time 0.20 (-0.03, 0.43) 0.28 (-0.07, 0.64) 0.27 (-0.08, 0.62)
Class 1 * Time Ref Ref
Class 2 * Time -0.38 (-0.90, 0.13) -0.40 (-0.91, 0.12)
Class 3 * Time 0.24 (-0.35, 0.83) 0.21 (-0.38, 0.81)
Covariates Age -0.02 (-0.05, 0.01)
Female 0.18 (-0.24, 0.59)
Race (ref: White)
• Black
• others
-0.18 (-0.73, 0.37)
1.92 (0.81, 3.04)
Education -0.52 (-0.94, -0.09)
Comorbidities 0.28 (0.15, 0.42)
Income -0.00 (-0.19, 0.18)
Note: Class 1 = Low-stress low-support spousal caregivers; Class 2 = Medium-stress high-
support spousal caregivers; Class 3 = High-stress medium-support spousal caregivers
95% confidence interval in parenthesis; bolded values indicate statistical significance at p<.05
level based on Wald test
129
Abstract (if available)
Abstract
Spousal caregivers to older adults are at the intersection of partners and caregivers; they need to simultaneously fulfill their supportive roles and cope with the stress caused by caring activities. Compared to other family caregivers such as adult child, spousal caregivers are more likely to encounter more stressors and poorer physical and mental health, because of the declined physical condition and increased daily functional needs for both the care-recipients and themselves. Meanwhile, studies have suggested increasing evidence showing beneficial results from caregiving such as increased interpersonal relationships, increased quality of life, and lower mortality. Explanations for the different findings include selection bias: (1) when comparing caregivers to non-caregivers, studies recruiting participants from convenience samples could exaggerate the negative effects of caregiving due to socially active individuals’ self-selection into participation as non-caregivers, and (2) when healthier individuals uptake caregiving role or remain in caring activities, their self-selection into caregiving because of better initial health status could confound the findings on health outcomes especially mortalities. The other explanation concerns the small effect size or insignificant findings when comparing caregivers to non-caregivers, indicating that caregivers are heterogeneous, and some caregivers, such as spousal caregivers, could have higher levels of stress than other types of caregivers and subsequently worse health outcomes.
The dissertation, therefore, chose to focus on a relatively homogenous caregiver subgroup – spousal caregivers. With the stress process model as the theoretical foundation, the first study (Chapter 2) examined the whole spousal caregiver group in comparison to their spousal non-caregiver counterparts, aiming to have an overall understanding of the health impacts of caregiving that are independent of selection bias. The second (Chapter 3) and third(Chapter 4) studies emphasized the heterogeneity among spousal caregivers and their care recipients. The two studies explored spousal caregivers’ latent classes based on the relative intensities of co-occurring stressors and resources they encounter and possess, and the association between identified latent classes and spousal caregivers’ profiles as well as care-recipients’ health outcomes.
Using data from the Health and Retirement Study, the first study compared 2,741 spousal caregivers with 18,043 spousal non-caregivers using coarsened exact matching. Results indicated that caregiving was associated with a 0.27-unit increase in depressive symptoms in the subsequent wave. This highlighted the elevated needs to address mental health among spousal caregivers.
The second and third studies utilized data from National Health and Aging Trends Study and National Study of Caregiving. Using latent class analysis, the second study identified three spousal caregiver latent classes based on the relative intensities of co-occurring stressors and resources from 793 observations across 639 spousal caregivers. The three latent classes were: low-stress low-support spousal caregivers, medium-stress high-support caregivers, and high-stress medium-support caregivers. Compared to low-stress low-support spousal caregivers, high-stress medium-support caregivers were found to have worse self-rated health and higher levels of depressive symptoms after controlling for covariates. Based on the identified spousal caregiver latent classes, the third study further explored how caregivers’ class membership was associated with their care-recipients’ health at baseline and rates of changes in health. Results indicated that compared to the counterparts taken care of by low-stress low-support caregivers, recipients taken care of by medium-stress high-support spousal caregivers had lower levels of self-rated health at baseline, and those with high-stress medium-support spousal caregivers had both worse self-rated heath and higher levels of depressive symptoms at baseline. No statistically significant differences in health changes were detected across the three groups within the one-year time interval.
The findings indicated that spousal caregiving had a negative impact on caregivers’ mental health. Moreover, results highlighted that the combination of stressors and resources featured by their co-occurring relative intensities is related to spousal caregivers’ and recipients’ health outcomes: spousal caregivers and the corresponding recipients have the worst health outcomes when the stressors caregivers encounter are at a higher level relative to the resources they possess. Social workers and policy makers should, therefore, focus on mental health of spousal caregivers in general, and pay special attention to the spousal dyads whose level of stressors exceeds that of the resources they can mobilize.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
The roles of social isolation and secondary caregiver support in the stress process among primary caregivers of older adults
PDF
My Life, My Wishes, a death education program for community-dwelling Chinese older adults: development and pilot testing
PDF
A targeted culturally-informed approach for caregiver stress among Vietnamese caregivers of family members [capstone paper]
PDF
A targeted culturally-informed approach for caregiver stress among Vietnamese caregivers of family members [executive summary]
PDF
A targeted culturally-informed approach for caregiver stress among Vietnamese caregivers of family members [oral defense]
PDF
An exploratory study of the role of religion and religious involvement among Latinos with schizophrenia and family caregivers
PDF
Examining the longitudinal influence of the physical and social environments on social isolation and cognitive health: contextualizing the role of technology
PDF
The effects of wisdom-related personality traits on caregivers’ health: an application of the resilience model
PDF
Adoption of virtual healthcare, self-sufficient wages and paid neighbors concept will ensure optimal living for vulnerable people and their paid caregivers
PDF
Exploring the social determinants of health in a population with similar access to healthcare: experiences from United States active-duty army wives
PDF
Complicated relationships: the prevalence and correlates of sexual assault, dating violence victimization, and minority stress among sexual minority adolescents throughout the United States
PDF
Mechanisms of the effect of involuntary retirement on older adults' health and mental health
PDF
Life transitions, leisure activity engagement, and cognition among older adults
PDF
The dollars and sense of caregiving: exploring the financial financial realities of providing informal care to an older adult
PDF
The effects of time and space on health status of older adults in China
PDF
Family relationships and their influence on health outcomes over time among older adults in rural China
PDF
Acculturation team-based clinical program: pilot program to address acculturative stress and mental health in the Latino community
PDF
The role of public policy in the decisions of parents and caregivers: an examination of work, fertility, and informal caregiving
PDF
Is stress exposure enough? Race/ethnic differences in the exposure and appraisal of chronic stressors among older adults
PDF
An examination of the association between spousal support and type 2 diabetes self-management
Asset Metadata
Creator
Liu, Ruotong
(author)
Core Title
Impacts of caregiving on wellbeing among older adults and their spousal caregivers in the United States
School
Suzanne Dworak-Peck School of Social Work
Degree
Doctor of Philosophy
Degree Program
Social Work
Degree Conferral Date
2022-08
Publication Date
07/27/2022
Defense Date
04/15/2022
Publisher
University of Southern California
(original),
University of Southern California. Libraries
(digital)
Tag
caregiving,coarsened exact matching,latent class analysis,Mental Health,OAI-PMH Harvest,stress process model
Format
application/pdf
(imt)
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Chi, Iris (
committee chair
), Wu, Shinyi (
committee chair
), Jang, Yuri (
committee member
), Zissimopoulos, Julie (
committee member
)
Creator Email
ruotongl@usc.edu,uscmona@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-oUC111375368
Unique identifier
UC111375368
Legacy Identifier
etd-LiuRuotong-11014
Document Type
Dissertation
Format
application/pdf (imt)
Rights
Liu, Ruotong
Type
texts
Source
20220728-usctheses-batch-962
(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
Tags
caregiving
coarsened exact matching
latent class analysis
stress process model