ArticlePDF Available

Family Caregiving and All-Cause Mortality: Findings from a Population-based Propensity-matched Analysis

Authors:

Abstract and Figures

Previous studies have provided conflicting evidence on whether being a family caregiver is associated with increased or decreased risk for all-cause mortality. This study examined whether 3,503 family caregivers enrolled in the national Reasons for Geographic and Racial Differences in Stroke (REGARDS) Study showed differences in all-cause mortality from 2003 to 2012 compared with a propensity-matched sample of noncaregivers. Caregivers were individually matched with 3,503 noncaregivers by using a propensity score matching procedure based on 15 demographic, health history, and health behavior covariates. During an average 6-year follow-up period, 264 (7.5%) of the caregivers died, which was significantly fewer than the 315 (9.0%) matched noncaregivers who died during the same period. A proportional hazards model indicated that caregivers had an 18% reduced rate of death compared with noncaregivers (hazard ratio = 0.823, 95% confidence interval: 0.699, 0.969). Subgroup analyses by race, sex, caregiving relationship, and caregiving strain failed to identify any subgroups with increased rates of death compared with matched noncaregivers. Public policy and discourse should recognize that providing care to a family member with a chronic illness or disability is not associated with increased risk of death in most cases, but may instead be associated with modest survival benefits for the caregivers.
Content may be subject to copyright.
Original Contribution
Family Caregiving and All-Cause Mortality: Findings from a Population-based
Propensity-matched Analysis
David L. Roth*, William E. Haley, Martha Hovater, Martinique Perkins, Virginia G. Wadley,
and Suzanne Judd
*Correspondence to Dr. David L. Roth, Center on Aging and Health, 2024 East Monument Street, Suite 2-700, Baltimore, MD 21205
(e-mail: droth@jhu.edu).
Initially submitted April 9, 2013; accepted for publication July 15, 2013.
Previous studies have provided conflicting evidence on whether being a family caregiver is associated with
increased or decreased risk for all-cause mortality. This study examined whether 3,503 family caregivers enrolled in
the national Reasons for Geographic and Racial Differences in Stroke (REGARDS) Study showed differences in
all-cause mortality from 2003 to 2012 compared with a propensity-matched sample of noncaregivers. Caregivers
were individually matched with 3,503 noncaregivers by using a propensity score matching procedure based on 15
demographic, health history, and health behavior covariates. During an average 6-year follow-up period, 264
(7.5%) of the caregivers died, which was significantly fewer than the 315 (9.0%) matched noncaregivers who died
during the same period. A proportional hazards model indicated that caregivers had an 18% reduced rate of death
compared with noncaregivers (hazard ratio = 0.823, 95% confidence interval: 0.699, 0.969). Subgroup analyses
by race, sex, caregiving relationship, and caregiving strain failed to identify any subgroups with increased rates of
death compared with matched noncaregivers. Public policy and discourse should recognize that providing care to
a family member with a chronic illness or disability is not associated with increased risk of death in most cases, but
may instead be associated with modest survival benefits for the caregivers.
caregiving; cohort studies; mortality; propensity scores
Abbreviations: CHES, Caregiver Health Effects Study; REGARDS, Reasons for Geographic and Racial Differences in Stroke.
The increasing number of older adults, rising prevalence
of many chronic diseases, and greater emphasis on noninsti-
tutional care are requiring a greater number of individuals to
serve as informal caregivers of family members with chronic
illnesses or disabilities (1). These family caregivers often endure
substantial life changes and chronic stressors that several stud-
ies suggest are linked to deleterious health effects (25), includ-
ing increased riskof death (6,7). A widely cited landmark study
of spouse caregivers, the Caregiver Health Effects Study
(CHES), found that those who were providing care to a dis-
abled spouse and who reported some strain associated with
that care had a 63% elevated riskof death compared with non-
caregiving spouses (6). Increased rates of death have also
been reported for the spouses of partners who have recently
been hospitalized (7). Along with ndings from many studies
that suggest caregivers have poorer mental and physical health
status than noncaregivers (3), caregiving has been widely por-
trayed as a serious public health problem in the professional
literature (8,9) and as a threat to survival in the popular media
(e.g., the most devoted family caretakers are at risk of dying
rst themselves(10, p. 70)).
Despite these common conclusions that caregiving presents
a health risk that could extend to increased risk of death, sev-
eral other recent studies have provided opposing evidence and
suggested that caregiving mayactually be associated with pre-
served health over time and reduced risk of death. Married
participants from the Health and Retirement Study providing
14 or more hours of care per week to their spouses who had
problems with activities of daily living or instrumental activ-
ities of daily living were found to have reduced rates of death
compared with spouses who provided no such care (11). An
analysis of Northern Ireland census data found that individuals
1571 Am J Epidemiol. 2013;178(10):15711578
American Journal of Epidemiology
© The Author 2013. Published by Oxford University Press on behalf of the Johns Hopkins Bloomberg School of
Public Health. All rights reserved. For permissions, please e-mail: journals.permissions@oup.com.
Vol. 178, No. 10
DOI: 10.1093/aje/kwt225
Advance Access publication:
October 3, 2013
at University of South Florida on November 27, 2013http://aje.oxfordjournals.org/Downloaded from
who reported family caregiving responsibilities had lower 4-
year death rates than noncaregivers (12). Fredman et al. (13)
reported that older American women who engaged in infor-
mal caregiving activities had lower 8-year death rates than a
corresponding sample of noncaregiving women, and that higher
levels of physical performance (e.g., walking speed, strength
measures) were maintained over a 2-year period among the
caregivers who provided a high level of assistance with activ-
ities of daily living or instrumental activities of daily living
(14).
Several factors might partially explain the ndings of pre-
served health and lower rates of death among some caregiving
samples. These factors form the core of the healthy care-
giver hypothesis(14) and include both possible selection
processes and potential psychological and social benets of
caregiving. Selection factors concern who takes on informal
caregiving responsibilities when a family member becomes
seriously ill or disabled. One population-based study found
that healthier individuals were more likely to take on and endure
in family caregiving roles over time (15). Health and resource
factors might be especially important in the selection of non-
spouse caregivers. Other investigators have noted the potential
positive aspects of caregiving (16,17), including possible
health and longevity benets for individuals who become
more active themselves when volunteering or providing help
and support to others (1820).
One topic that is rarely addressed is whether the mortality
effects of caregiving are similar across different subgroups of
caregivers. Caregivers are an incredibly diverse group consist-
ing of many relationship subtypes (e.g., spouses, adult children,
and others) who handle different types of care recipient prob-
lems. Caregivers may or may not live with their care recipients
and may perceive different levels of caregiving strain. Exist-
ing studies of the caregiving-mortality association in the United
States have not only led to conicting ndings, but have also
been limited to spouse caregivers (6,7,11) or female caregivers
(13). Population-based studies of caregivers and matched non-
caregiving controls that include larger numbers of minority
participants and more diverse and representative relationship
subtypes are needed to further inform this important area of
investigation.
A promising analytical approach for examining potential
causal variables that cannot be subjected to random assignment
is through the use of propensity scores that can be obtained
Table 1. Baseline Characteristics of Caregivers and Noncaregivers From the REGARDS Study Before and After Propensity-based Matching,
20032007
Matching Factor Caregivers, %
(n= 3,503)
All Noncaregivers,
%(n=24,863) PValue
Propensity-matched
Noncaregivers, %
(n= 3,503)
PValue
Female sex 63.35 54.51 <0.0001 62.40 0.4145
African American race 43.11 40.22 0.0011 43.48 0.7539
Age, years 63.58 (8.93)
a
65.44 (9.44)
a
<0.0001 63.27 (9.04)
a
0.1408
Region <0.0001 0.9561
Stroke belt
b
38.00 34.36 37.91
Stroke buckle
c
21.64 21.02 21.44
Rest of United States 40.37 44.62 40.65
Education 0.0023 0.9806
Less than high school 10.85 12.27 10.85
High school graduate 24.32 26.08 24.26
Some college 28.49 26.54 28.92
College graduate 36.34 35.11 35.97
Income 0.0395 0.4703
<$20,000 16.56 17.85 15.16
$20,000$34,000 25.18 23.98 24.64
$35,000$74,000 31.40 29.80 32.40
$75,000 15.19 16.29 15.56
Refused to specify 11.60 12.08 12.25
Marital status <0.0001 0.7161
Married 66.69 58.30 67.49
Divorced 12.96 14.61 12.96
Single/never married 5.57 5.20 5.71
Widowed 12.48 19.59 11.39
Separated/refused to specify 2.31 2.30 2.46
Table continues
1572 Roth et al.
Am J Epidemiol. 2013;178(10):15711578
at University of South Florida on November 27, 2013http://aje.oxfordjournals.org/Downloaded from
from large observational data sets (2123). In this approach,
individuals who are exposed to a treatment (or a risk factor)
are compared with a selected comparison group of untreated
individuals, such that both groups are balanced on a wide range
of potential confounding factors. Large, population-based,
observational studies that assess caregiving status as 1 of a
range of contextual variables provide a unique opportunity to
implement this approach and to examine the health effects of
caregiving after controlling for many confounding variables.
However, to our knowledge, no previous study of the caregiving-
mortality association has compared caregivers with a matched
sample of noncaregivers by using an empirical, propensity
score matching procedure.
In this study, family caregivers were identied from a large
national epidemiologic study and conrmed to differ from non-
caregivers on a range of demographic, medical history, and
health behavior variables. A logistic regression, propensity-
matching algorithm was used to individually match and bal-
ance caregiving and noncaregiving subgroups, and survival
analysis methods were then used to examine subsequent all-
cause mortality rates. We also conducted supplemental anal-
yses to examine whether the mortality effects of caregiving
were comparable across races and sexes, between caregivers
of parents versus spouses, and among those with different lev-
els of self-reported caregiving strain to see if different pat-
terns would emerge among subgroups that might partly explain
the previous contradictory ndings.
MATERIALS AND METHODS
Participants
Participants in the REGARDS Study were randomly sampled
from a commerciallyavailable nationwide list. The design, enroll-
ment, and interviewing procedures for the REGARDS Study
have been previously described in detail elsewhere (2427).
Briey, exclusion criteria included age of less than 45 years,
race other than African American or white, previous diagnosis
of cancer requiring chemotherapy, residence in a nursing home,
or being on a waiting list for a nursing home. African Ameri-
cans and residents from the southern stroke beltregion of
the United States were oversampled on the basis of the stratied
random sampling design that was used in the REGARDS Study.
All procedures were reviewed and approved by the institutional
Table 1. Continued
Matching Factor Caregivers, %
(n= 3,503)
All Noncaregivers,
%(n= 24,863) PValue
Propensity-matched
Noncaregivers, %
(n= 3,503)
PValue
Have medical insurance 91.12 93.82 <0.0001 91.29 0.8002
Smoking <0.0001 0.8283
Current 15.84 13.98 15.50
Former 35.66 40.65 36.31
Never 48.50 45.37 48.19
Current alcohol use 0.0077 0.4910
Heavy 3.48 4.11 3.85
Moderate 31.83 33.77 32.66
None 64.69 62.12 63.49
Six-item cognitive screener
d
<0.0001 0.7174
04 correct 6.56 9.42 7.51
5 correct 21.30 21.64 20.92
6 correct 72.14 68.94 71.57
Self-rated health 0.2947 0.7099
Excellent 15.42 16.34 14.53
Very good 30.17 30.87 30.75
Good 36.05 34.65 35.48
Fair 15.19 14.71 16.07
Poor 3.17 3.43 3.17
Have hypertension 57.44 57.88 0.6210 56.32 0.3468
Have diabetes 21.32 22.40 0.1510 21.50 0.8613
Have cardiovascular disease 18.53 23.12 <0.0001 18.58 0.9510
a
Reported as mean (standard deviation).
b
The stroke belt included the portions of North Carolina, South Carolina, and Georgia not included in the stroke buckle, plus all of Alabama,
Mississippi, Tennessee, Arkansas, and Louisiana.
c
The stroke buckle included a coastal plain region of North Carolina, South Carolina, and Georgia.
d
The 6-item screener of global cognitive status described by Callaham et al. (29).
Family Caregiving and All-Cause Mortality 1573
Am J Epidemiol. 2013;178(10):15711578
at University of South Florida on November 27, 2013http://aje.oxfordjournals.org/Downloaded from
review boards of each participating institution. Enrollment occur-
red from 2003 through2007. Of the 30,23 9 enrolledparticipants,
1,873 (6.2%) had missing data on mortality status or at least
1 of the 15 propensity-matching covariates, leaving 28,366 par-
ticipants with complete data for the present analyses.
Procedures and measures
Trained interviewers contacted potential participants, estab-
lished eligibility, obtained verbal informed consent, and admin-
istered a computer-assisted telephone interview. Data were
obtained on the variables that are described in the following
sections.
Demographic variables. Age was calculated on the basis
of the number of days between the participants date of birth
and the baseline interview date. Sex and race (African Amer-
ican vs. white) were dichotomous variables based on self-report.
Region was analyzed on the basis of the stratied sampling
categories that were used (stroke belt, stroke buckle,”“non-
belt). The stroke buckle included a coastal plain region of
North Carolina, South Carolina, and Georgia. The stroke belt
included the remainder of these 3 states plus all of Alabama,
Mississippi, Tennessee, Arkansas, and Louisiana. The nonbelt
region included the other 40 contiguous states. Marital status,
educational level, and annual income were coded as categor-
ical variables as indicated in Table 1. Insurance coverage was
a dichotomous indicator of whether the participant reported
having any type of medical insurance.
Health behaviors. Responses to interview questions con-
cerning smoking were coded as indicated in Table 1. Alcohol
use categories of none, moderate drinking, or heavy drinking
were based on sex-specic guidelines for alcohol use within
the past week (28).
Cognitive function. The 6-item screener of global cognitive
status (29) was administered during telephone interviews that
began in December 2003. This measure was obtained from
the baseline interview for 24,448 participants and from the
rst available semiannual follow-up interview for the remaining
5,167 participants who were enrolled before this procedure was
added to the baseline interview protocol. The number of cor-
rect responses (ranging from 0 to 6) was included as a cate-
gorical variable in the propensity-matching procedure.
Health and disease history. Participants provided an
overall description of their self-rated health (excellent, very
good, good, fair, or poor). Participants were also asked a num-
ber of health history questions. Participants who reported that
they had been told by a doctor or health professional that they
had high blood pressure or hypertension or who were taking
medications for high blood pressure were coded as having a
history of hypertension. Participants were coded as having a
history of diabetes if they reported being told by a doctor
or health professional that they had diabetes or high blood
sugaror were taking medications specically for diabetes.
A history of cardiovascular disease was coded, as in a previ-
ous analysis (26), for any participants who reported a history
of myocardial infarction, stroke, transient ischemic events,
carotid endarterectomy, coronary intervention, repair of aortic
aneurism, or peripheral arterial intervention.
Caregiving status. Toward the end of the baseline inter-
view, eachparticipant was asked, Are youcurrently providing
care on an on-going basis to a family member with a chronic
illness or disability? This would include any kind of help such
as watching your family member, dressing or bathing this per-
son, arranging care, or providing transportation.Respondents
whoansweredafrmatively were categorized as caregivers and
were subsequently asked whether they lived with the care recip-
ient, their relationship with the care recipient (e.g., spouse, child),
the amount of perceived mental or emotional strain associ-
ated with that care (none, some, a lot), and the number of hours
per week they provided such care. A cutpoint of 14 hours per
week was used for the subgroup analyses, consistent with the
approach used by Brown et al. (11).
All-cause mortality. Preliminary dates of death were typi-
cally obtained from proxy reports when participants could
not be reached for routine semiannual follow-up interviews.
A death certicate was then obtained from the participants
families or state departments of health, and dates of death
were veried by using the death certicates or the National
Death Index (30,31). Analyses were based on the deaths that
occurred through April 1, 2012.
Statistical analysis
Descriptive χ
2
tests were used to compare caregivers and
noncaregivers on all covariates except for age, for which an
independent-groups Studentsttest was used. The dichoto-
mous caregiving status variable (yes or no) was then regressed
on the 15 covariates in Table 1by using a standard binary
logistic regression analysis. The propensity scores from this
analysis represented the predicted probability of being a care-
giver based each participants covariate values. Each caregiver
was then individually matched with a noncaregiving partici-
pant on this propensity score by using a modied greedy
matching algorithm without replacement (32). Matches were
accomplished by rst completing all matches that could be
made at the fth decimal place (i.e., propensity score differ-
ences < 0.00001), then the fourth decimal place, and so on,
until all caregivers were matched. In cases of tied propensity
score differences, the matching noncaregiver was selected ran-
domly from the pool of tied cases. The mean absolute value of
the propensity score differences was 0.00003, and the largest
absolute difference was 0.00768.
The descriptive comparisons between the caregivers and
the propensity-matched noncaregivers were repeated on the
covariates to conrm the balance between these 2 groups. A
Cox proportional hazards survival analysis was then con-
ducted for the propensity-matched caregivers and noncare-
givers. This analysis was based on the number of days elapsed
between the baseline interview and the date of death for the
deceased cases (median, 1,463 days) or the date of the last
semiannual follow-up interview for the living cases (median,
2,277 days). The overall median length of follow-up time was
6.1 years (2,226 days).
The caregiving subgroup analyses were conducted by repeat-
ing the logistic regression, propensity matching, group bal-
ance conrming, and proportional hazards survival analysis
sequence of steps each time for each subgroup examined sep-
arately. In many of these subgroup analyses, only 14 balanc-
ing covariates were used in the logistic regression propensity
score calculation because both the caregivers and matched
1574 Roth et al.
Am J Epidemiol. 2013;178(10):15711578
at University of South Florida on November 27, 2013http://aje.oxfordjournals.org/Downloaded from
noncaregivers were restricted to just 1 class on a remaining
demographic variable. For example, in the analysis of spouse
caregivers, only married noncaregivers were available for poten-
tial matching, and marital status was, therefore, not included as
a predictor variable in the logistic regression analysis that cal-
culated the propensity score. Likewise, female caregivers were
matched with female noncaregivers only. Similar adjustments
were made for men and for race-specic caregiving subgroup
analyses.
RESULTS
Propensity matching
Table 1summarizes the descriptive comparisons between
the 3,503 caregivers and the noncaregivers in the REGARDS
Study. Prior to matching, caregivers differed signicantly from
noncaregivers on 12 of the 15 covariates. Caregivers were
younger, on average, and more likely to be women, African
American, and married. Caregivers were less likely to have
health insurance and to report a history of cardiovascular dis-
ease. Subtle but statistically signicant differences were also
observed for education, income, smoking status, and alcohol
use. After propensity matching, the 3,503 caregivers did not
differ signicantly from their 3,503 matched noncaregivers on
any of the 15 covariates, conrming the success of the binary
logistic regression and greedy matching procedure for iden-
tifying balanced groups of caregivers and matched noncare-
givers for further analysis.
Mortality effects across all caregivers
Figure 1displays the descriptive survival curves for the
3,503 caregivers, for all of the 24,863 noncaregivers, and for
the 3,503 propensity-matched noncaregivers. Of the 3,503
caregivers, 264 (7.5%) died during the follow-up period,
whereas 2,782 of the 24,863 noncaregivers (11.2%) died
during this same period. After propensity matching, 315 of the
3,503 matched noncaregivers were deceased (9.0%), which
was a signicantly greater proportion than the 7.5% of care-
givers according to a simple χ
2
test (P= 0.0269). The Cox
proportional hazards analysis revealed that caregivers died
at approximately an 18% lower rate than their individually
matched noncaregivers over this 6-year period (hazard ratio =
0.823, 95% condence interval: 0.699, 0.969; P= 0.0196).
Mortality and caregiving subgroups
The sample of 3,503 caregivers included many different
subgroups identied by race, sex, caregiving relationship, per-
ceived caregiving strain, and amount of caregiving involve-
ment. Table 2summarizes the results of the subgroup analyses
that were conducted. In each analysis, specic caregivers were
individually matched with qualied potential noncaregiving
controls by using a new logistic regression and propensity score
matching procedure. In all cases, the propensity matching pro-
cedure was effective for balancing the caregiver and noncaregiver
groups on the relevant covariates. All Pvalues were greater
than 0.12, and 170 of the 174 possible covariate comparisons
resulted in Pvalues greater than 0.20.
The results of the Cox proportional hazards models identi-
ed 1 subgroup of caregivers with a signicantly lower death
rate. Adult child caregiverswho were providing care to a parent
were found to have a signicantly lower rate of death compared
to their propensity-matched noncaregivers (P= 0.0064). In
addition, trends that approached conventional levels of sta-
tistical signicance were observed for white caregivers (P=
0.0791), female caregivers (P= 0.0703), and caregivers who
provided 14 or more hours of care per week (P= 0.0752).
The hazard ratio for each of these subgroups was similar to
the hazard ratio for all caregivers, but the hazard ratios for the
subgroups were no longer statistically signicant at the P<
0.05 level because of reduced sample sizes and power. No
subgroup of caregivers showed a trend for increased risk of
death compared with propensity-matched noncaregivers. The
strained spouse caregiving subgroup,which included spouses
who reported either moderate or high caregiving strain, was
similar to the spouse caregivers found to have an elevated rate
of death in the CHES (6).
DISCUSSION
The present ndings contribute important new information
concerning the paradox of whether informal family caregiving
1.00
0.96
0.92
0.88
Survival
0.84
0.80
01234
Follow-up, years
5678
Figure 1. Survival curves for caregivers (black line, n= 3,503), pro-
pensity-matched noncaregivers (gray line, n= 3,503), and all noncar-
egivers (dotted line, n= 24,863) from the REGARDS Study over the 8
years of follow-up after enrollment, 20042012.
Family Caregiving and All-Cause Mortality 1575
Am J Epidemiol. 2013;178(10):15711578
at University of South Florida on November 27, 2013http://aje.oxfordjournals.org/Downloaded from
responsibilities are associated with higher or lower rates of
death, as suggested by multiple conicting previous studies
(6,7,1113). Our ndings are consistent with the studies sug-
gesting lower rates of death among caregivers, (1113)in
that the self-identied family caregivers from the REGARDS
Study, as an overall group, experienced lower all-cause mor-
tality rates than the empirically matched sample of noncare-
givers from the same epidemiologic sample. The propensity
matching procedure resulted in sufciently balanced care-
giving and noncaregiving comparison groups across key demo-
graphic, health history, and health behavior variables. The
present study used a more diverse and inclusive sample of
caregivers than have previous studies of the caregiving-
mortality association in the US population, and it is the rst
such study to demonstrate caregiving-mortality effects by
using a propensity score matching procedure. As indicated in
Figure 1, an even stronger protective effect for caregiving was
found when the caregivers were compared directly with all
noncaregivers before propensity-based matching.
The subgroup analysestypically resulted in hazard ratio point
estimates that were similar to that for all caregivers. Although
reduced power was available for the subgroup analyses, sig-
nicant effects were observed for adult child caregivers in
comparison with their respective propensity-matched noncare-
giving control group. We did not nd any subgroup of care-
givers in the REGARDS sample that appeared to be vulner-
able to increased risk of death. This includes our analyses of
spouse caregivers and spouse caregivers who experience some
caregiving strain. These subgroups did not show elevations
in their risk of death in our sample, in contrast to the previous
ndings from the CHES (6). Both the CHES and the REGARDS
Study assessed caregiving strain in the same manner. The high-
strain caregivers in the REGARDS Study have been previously
shown to have higher rates of death than the moderate- and no-
strain caregivers after adjustment for demographic and other
caregiving-related variables (33), but that analysis was limited
to caregivers only. The present ndings clarify this caregiv-
ing strain effect by showing that most caregivers report low or
moderate caregiving strain, and that those caregivers do not show
elevated rates of death when compared with propensity-matched
samples of noncaregivers.
The present results do not rule out the possibility that some
subgroups of caregivers may be vulnerable to increased risk
of death. Limitations of the current analysis include a lack of
information on the functional status of the care recipients and
the specics of the care being provided. We do not know, for
example, how many caregivers provided assistance with activ-
ities of daily living, and some may have simply visited or
watchedtheir care recipients. We were also not able to dis-
tinguish caregivers of those with dementia from other sub-
groups. Caregivers of those with dementia typically face many
unique and chronic stressors (34). These recurringstressors are
associated with alterations in circulating inammatory bio-
markers (35) that have been linked to increased all-cause mor-
tality rates (27). The nding that the hospitalization of a spouse
increased the risk of death of the nonhospitalized spouse further
showed that these risks of death were particularly elevated if the
hospitalizations were for a disabling condition such as demen-
tia (7). Future research should include indicators of care needs,
particularly for stressful types of care involvingdementia, m ental
healthrelated issues, and end-of-life situations. It may be neces-
sary to oversample some of these specic caregiving subtypes to
better dene the impact of caregiving and to investigate its diverse
effects.
Strengths of the REGARDS Study include a high level of
participation among African Americans and the geographical
Table 2. Survival Rates of All Caregivers and Caregiving Subgroups Compared With Propensity-matched
Noncaregivers in the REGARDS Study, 20042012
Caregiver Group No. Caregivers,
% died
Propensity-matched
Noncaregivers, %
died
HR 95% CI
All 3,503 7.5 9.0 0.823*0.699, 0.969
White 1,993 7.0 8.4 0.818 0.654, 1.024
African American 1,510 8.2 9.2 0.884 0.694, 1.127
Women 2,219 4.9 6.0 0.792 0.615, 1.020
Men 1,284 12.1 14.3 0.841 0.679, 1.041
Spouse 786 11.1 12.3 0.872 0.653, 1.165
Adult child 1,197 3.0 5.2 0.565** 0.375, 0.852
No-strain 1,163 8.9 10.3 0.858 0.660, 1.115
Moderate-strain 1,748 6.3 7.4 0.831 0.644, 1.071
High-strain 578 8.1 8.0 1.022 0.681, 1.535
Strained spouse 537 9.3 9.9 0.944 0.642, 1.390
14 hours of care per
week
1,588 7.3 9.1 0.801 0.628, 1.023
<14 hours of care per
week
1,915 7.7 7.7 0.992 0.789, 1.245
Abbreviations: CI, confidence interval; HR, hazard ratio.
*P< 0.05; **P< 0.01.
1576 Roth et al.
Am J Epidemiol. 2013;178(10):15711578
at University of South Florida on November 27, 2013http://aje.oxfordjournals.org/Downloaded from
diversity of its sample. However, caution is warranted in extra-
polating these ndings to other ethnic groups, such as His-
panic and Asian populations, who were not included in the
REGARDS Study. Gender roles and expectations in multi-
generational households may confer a different mix of benets
and potential risks from the family caregiving experience.
Highly stressful caregiving situations have been the focus
of considerable research and dominate media narratives about
caregiving, but the caregiving experience is incredibly diverse,
and the majority of caregivers appear to be willingly provid-
ing help to family members with relatively low levels of need.
Only 10% of caregivers report caring for a person with demen-
tia (1), and many caregivers of cognitively intact older adults
report relatively mild caregiving demands (34). Fewer than
17% of the caregivers in the present analysis reported high
levels of caregiving strain. In many cases, caregivers report
receiving benets of enhanced self-esteem, recognition, and
gratitude from their care recipients (16,17,36). Several recent
papers have reported that caregivers who report low strain or
burden have better psychological well-being than noncare-
givers (5,37). Thus, when caregiving is done willingly, at
manageable levels, and for individuals who are capable of
expressing gratitude, it is reasonable to expect that health bene-
ts might accrue in those situations. Previous research (1820)
shows that a variety of altruistic behaviors, including providing
social support and volunteering, are associated with improved
well-being and reduced morbidity and mortality. Altruism,
especially within families, is likely to have evolutionary advan-
tages, and the positive affect generated by helping others is a
mechanism through which altruism might improve physical
health (18).
This more balanced and diverse approach to caregiving
research should not eliminate the legitimate concerns about
the possible negative health effects of high-strain caregiving on
caregiversphysical, psychological, and social well-being.
More broadly, if highly stressful situations can be avoided or
managed effectively, caregiving may actually produce some
health benets for both the care recipients and the caregiv-
ers, including reduced risk of death for those providing care.
Negative public health and media portrayals of the risks of fam-
ily caregiving may do a disservice by portraying caregiving as
dangerous and could potentially deter family members from
taking on what can be a satisfying and healthy family role.
Public discussions of caregiving should more accurately bal-
ance the potential risks and gains of this universal family
role.
ACKNOWLEDGMENTS
Author afliations: Center on Aging and Health, Division
of Geriatric Medicine and Gerontology, Johns Hopkins Uni-
versity, Baltimore, Maryland (David L. Roth); School of
Aging Studies, College of Behavioral and Community Sci-
ences, University of South Florida, Tampa, Florida (William
E. Haley); Department of Biostatistics, School of Public Health,
University of Alabama at Birmingham, Birmingham, Alabama
(Martha Hovater); Department of Health Behavior, School of
Public Health, University of Alabama at Birmingham, Bir-
mingham, Alabama (Martinique Perkins); Division of Geron-
tology, Geriatrics,and Palliative Care, Schoolof Medicine, Uni-
versity of Alabama at Birmingham, Birmingham, Alabama
(Virginia G. Wadley); and Department of Biostatistics, School
of Public Health, University of Alabama at Birmingham, Bir-
mingham, Alabama (Suzanne Judd).
This work was supported by the National Institute of Neu-
rological Disorders and Stroke, National Institutes of Health,
Department of Health and Human Services (cooperative
agreement U01 NS041588) and the National Institute of
Neurological Disorders and Stroke (grants R01 NS045789
and R01 NS075047).
We thank the other investigators and the staff of the
REGARDS Study for their valuable contributions. A full list
of participating REGARDS investigators and institutions can
be found at http://www.regardsstudy.org.
The content is solely the responsibility of the authors and
does not necessarily represent the ofcial views of the National
Institute of Neurological Disorders and Stroke or the National
Institutes of Health. Representatives of the funding agency
were involved in the review of the manuscript but were not
directly involved in the collection, management, analysis, or
interpretation of the data.
Conict of interest: none declared.
REFERENCES
1. National Alliance for Caregiving and AARP. Caregiving in the
U.S. 2009. Bethesda, MD: National Alliance for Caregiving;
2009.
2. Emanuel EJ, Fairclough DL, Slutsman BA, et al.
Understanding economic and other burdens of terminal illness:
the experience of patients and their caregivers. Ann Intern
Med. 2000;132(6):451459.
3. Pinquart M, Sörensen S. Differences between caregivers and
noncaregivers in psychological health and physical health: a
meta-analysis. Psychol Aging. 2003;18(2):250267.
4. Pinquart M, Sörensen S. Correlates of physical health of
informal caregivers: a meta-analysis. J Gerontol B Psychol Sci
Soc Sci. 2007;62(2):P126P137.
5. Roth DL, Perkins M, Wadley VG, et al. Family caregiving and
emotional strain: associations with quality of life in a large
national sample of middle-aged and older adults. Qual Life
Res. 2009;18(6):679688.
6. Schulz R, Beach SR. Caregiving as a risk factor for mortality:
the Caregiver Health Effects Study. JAMA. 1999;282(23):
22152219.
7. Christakis N, Allison P. Mortality after the hospitalization of a
spouse. N Engl J Med. 2006;354(7):719730.
8. Talley RC, Crews JE. Framing the public health of caregiving.
Am J Public Health. 2007;97(2):224228.
9. Healthy Aging, Centers for Disease Control and Prevention.
Caregiving: a public health priority. Atlanta, GA: Centers for
Disease Control and Prevention; 2009. (http://www.cdc.gov/
aging/caregiving/index.htm). (Accessed February 28, 2013).
10. Sheehy G. The caregiving boomerang. Newsweek. July 5,
2010;156(1):7072.
11. Brown S, Smith D, Schulz R, et al. Caregiving behavior is
associated with decreased mortality risk. Psychol Sci. 2009;
20(4):488494.
Family Caregiving and All-Cause Mortality 1577
Am J Epidemiol. 2013;178(10):15711578
at University of South Florida on November 27, 2013http://aje.oxfordjournals.org/Downloaded from
12. OReilly D, Connolly S, Rosato M, et al. Is caring associated
with an increased risk of mortality? A longitudinal study. Soc
Sci Med. 2008;67(8):12821290.
13. Fredman L, Cauley JA, Hochberg M, et al. Mortality associated
with caregiving, general stress, and caregiving-related stress in
elderly women: results of Caregiver-Study of Osteoporotic
Fractures. J Am Geriatr Soc. 2010;58(5):937943.
14. Fredman L, Doros G, Ensrud KE, et al. Caregiving intensity
and change in physical functioning over a 2-year period:
results of the Caregiver-Study of Osteoporotic Fractures. Am J
Epidemiol. 2009;170(2):203210.
15. McCann JJ, Hebert LE, Bienias JL, et al. Predictors of
beginning and ending caregiving during a 3-year period in a
biracial community population of older adults. Am J Public
Health. 2004;94(10):18001806.
16. Beach SR, Schulz R, Yee JL, et al. Negative and positive
health effects of caring for a disabled spouse: longitudinal
ndings from the Caregiver Health Effects Study. Psychol
Aging. 2000;15(2):259271.
17. Zarit SH. Positive aspects of caregiving: more than looking on
the bright side. Aging Ment Health. 2012;16(6):673674.
18. Brown WM, Consedine NS, Magai C. Altruism relates to
health in an ethnically diverse sample of older adults. J
Gerontol B Psychol Sci Soc Sci. 2005;60(3):P143P152.
19. Post S, Neimark J. Why Good Things Happen to Good People:
The Exciting New Research that Proves the Link Between
Doing Good and Living Longer. New York, NY: Broadway
Books; 2007.
20. Poulin MJ, Brown SL, Dillard AJ, et al. Giving to others and
the association between stress and mortality. Am J Public
Health. 2013;103(9):16491655.
21. DAgostino RB. Propensity score methods for bias reduction
in the comparison of a treatment to a non-randomized control
group. Stat Med. 1998;17(19):22652281.
22. Rosenbaum PR, Rubin DB. Reducing bias in observational
studies using subclassication on the propensity score. JAm
Stat Assoc. 1984;79(387):516524.
23. Stuart EA. Matching methods for causal inference: a review
and look forward. Stat Sci. 2010;25(1):121.
24. Howard VJ, Cushman M, Pulley L, et al. The Reasons for
Geographic and Racial Differences in Stroke Study: objectives
and design. Neuroepidemiology. 2005;25(3):135143.
25. Howard G, Cushman M, Kissela BM, et al. Traditional risk
factors as the underlying cause of racial disparities in stroke:
lessons from the half-full (empty?) glass. Stroke.2011;42(12):
33693375.
26. McKnight KK, Wellons MF, Sites CK, et al. Racial and
regional differences in age at menopause in the United States:
ndings from the Reasons for Geographic and Racial
Differences in Stroke (REGARDS) Study. Am J Obstet
Gynecol. 2011;205(4):353.e1353.e8.
27. Kabagambe EK, Judd SE, Howard VJ, et al. Inammation
biomarkers and risk of all-cause-mortality in the Reasons for
Geographic and Racial Differences in Stroke (REGARDS)
Cohort. Am J Epidemiol. 2011;174(3):284292.
28. National Institute on Alcohol Abuse and Alcoholism. Helping
Patients Who Drink Too Much. A Clinicians Guide.
Rockville, MD: US Department of Health and Human
Services; 2005.
29. Callahan CM, Unverzagt FW, Hui SL, et al. Six-item screener
to identify cognitive impairment among potential subjects for
clinical research. Med Care. 2002;40(9):771781.
30. Halanych JH, Shuaib F, Parmar G, et al. Agreement on cause
of death between proxies, death certicate, and clinician
adjudicators in the Reasons for Geographic and Racial
Differences in Stroke (REGARDS) Study. Am J Epidemiol.
2011;173(11):13191326.
31. Doody MM, Hayes HM, Bilgrad R. Comparability of National
Death Index Plus and standard procedures for determining
causes of death in epidemiologic studies. Ann Epidemiol.
2001;11(1):4650.
32. Parsons LS. Reducing bias in a propensity score matched-pair
sample using greedy matching techniques. Proceedings of the
26th Annual SAS Users Group International Conference.
Cary, NC: SAS Institute, Inc; 2001.
33. Perkins M, Howard VJ, Wadley VG, et al. Caregiving strain
and all-cause mortality: evidence from the REGARDS Study.
J Gerontol B Psychol Sci Soc Sci. 2013;68(4):504512.
34. Ory MG, Hoffman RR III, Yee JL, et al. Prevalence and
impact of caregiving: a detailed comparison between dementia
and non-dementia caregivers. Gerontologist. 1999;39(2):
177185.
35. Gouin JP, Glaser R, Malarkey WB, et al. Chronic stress, daily
stressors, and circulating inammatory markers. Health
Psychol. 2012;31(2):264268.
36. Haley WE, Allen JY, Grant JS, et al. Problems and benets
reported by stroke family caregivers: results from a prospective
epidemiological study. Stroke. 2009;40(6):21292133.
37. Buyck JF, Bonnaud S, Boumendil A, et al. Informal
caregiving and self-reported mental and physical health:
results from the Gazel Cohort Study. Am J Public Health.
2011;101(10):19711979.
1578 Roth et al.
Am J Epidemiol. 2013;178(10):15711578
at University of South Florida on November 27, 2013http://aje.oxfordjournals.org/Downloaded from
... Research on caregiving emerged by the 1980s Roth et al. 2015;van Durme et al. 2012), with early and ongoing attention to how informal caregiving can negatively affect a caregiver's physical, emotional, and/or financial wellness (Beard & Fox 2008;Bruhn & Rebach 2014:161;Seeher 2013). Caregivers may face greater risks than non-caregivers for stress (Crimmins & Vasunilashorn 2016), isolation (Adler & Mehta 2014;Musich et al. 2017), depression (Roth et al. 2015), relationship challenges (Schulz & Eden 2016:104), and early mortality (Bruhn & Rebach 2014:162-163;Roth et al. 2013;Schulz & Eden 2016:100), yet caregiving is not without its positive aspects (Pinquart & Sörensen 2003;Roth et al. 2015). Negative consequences for caregivers appear to be moderated by the caregiver's age, race, and gender (Bruhn & Rebach 2014:111;Schulz & Eden 2016:63). ...
... Both the intensity and the duration of care increase thelikelihood of negative outcomes for the 16 million dementia caregivers in the U.S. (Gaugler et al 2019; Bruhn & Rebach 2014:131; Seltzer & Li 1996); further, 66 percent of caregivers cohabitate with their care recipients, more than any other form of care (Gaugler et al. 2019:31; Pinquart & Sörensen 2003), leaving them with little reprieve. Dementia care demonstrates negative effects on measures of caregiver wellness in both prevalence and severity(Roth et al. 2013; Pinquart & Sörensen 2004;Seeher 2013), and the "functional limitations and behavioral symptoms" associated with dementia(Schulz & Eden 2016:85) are noted to affect subjective well-being more than high counts of care needs(Schulz & Eden 2016:103; Pinquart & Sörensen 2004). A sense of guilt over what more can or should be done is common, if usually unwarranted(Bruhn & Rebach 2014:163;Seaman 2018). ...
Thesis
Full-text available
In the coming decades, there may not be enough professional caregivers to meet the needs of elderly and disabled persons; family and friends will take on more informal care roles than ever before, with great social impact, yet the social literature lacks consistency and breadth. This exploratory, quantitative study organizes the disparate social literature on caregiving and dementia care and suggests a critical theoretical framework to interpret social factors more fully. Utilizing the Caregiving in the U.S. dataset from 2015, three hypotheses test this framework for a dementia care population to identify variable combinations that exert outsized influence on care outcomes. Findings confirm two hypotheses: people of color tend to develop dementia far earlier than whites and persons who are marginalized in multiple ways tend to perform more activities of daily living; these findings demonstrate a role for intersectionality and queer subjectivity in dementia care analyses and interpretation.
... Caregiver burden poses challenges for both the caregiver and care recipient due to increased stress, which can lead to lower quality of care for care recipients (Schulz & Eden, 2016). Emotional distress can result from caregiver burden as caregivers feel overwhelmed (Di Mattei et al., 2008), which can negatively affect their physical (O'Reilly et al., 2008;Roth et al., 2013Roth et al., , 2015 and mental health (Cooper et al., 2008). Multiple factors, such as high number of hours spent caregiving, can increase risk for caregiver burden (Adelman et al., 2014). ...
Article
The Supporting Older Adults & Caregivers: Integrative Service Learning (SOCIAL) partnership trains undergraduate social work students to provide practical home-based support for older adults with chronic illness and their family caregivers, serving as a pipeline for future leaders in older adult care. More than 2 million older adults are homebound, and 5 million need help leaving their homes. Family members often assume daily caregiving tasks to assist their loved ones, navigate health care systems, and provide much needed emotional support. In this teaching note, we provide an example of an integrative service learning model which can serve older adults and caregivers while offering valuable pedagogical experiences to baccalaureate students, along with strategies for curriculum building, community engagement, research and evaluation, and program sustainability.
... Furthermore, prosocial behavior is associated with personal benefits for the individual provider of help. This is supported by mounting evidence of different types of prosocial behavior being related to greater physical and mental health with studies showing that volunteering, family caregiving, and the provision of tangible help to close others are associated with greater longevity (Harris and Thoresen, 2005;Poulin et al., 2013;Roth et al., 2013) and that spending money on others augments emotional well-being (Aknin et al., 2013). Past research has shown that prosocial behavior buffers against the negative effects of daily stress on emotional well-being (Raposa et al., 2016). ...
Article
Full-text available
The coronavirus disease (COVID-19) pandemic fundamentally disrupted humans’ social life and behavior. Public health measures may have inadvertently impacted how people care for each other. This study investigated prosocial behavior, its association well-being, and predictors of prosocial behavior during the first COVID-19 pandemic lockdown and sought to understand whether region-specific differences exist. Participants ( N = 9,496) from eight regions clustering multiple countries around the world responded to a cross-sectional online-survey investigating the psychological consequences of the first upsurge of lockdowns in spring 2020. Prosocial behavior was reported to occur frequently. Multiple regression analyses showed that prosocial behavior was associated with better well-being consistently across regions. With regard to predictors of prosocial behavior, high levels of perceived social support were most strongly associated with prosocial behavior, followed by high levels of perceived stress, positive affect and psychological flexibility. Sociodemographic and psychosocial predictors of prosocial behavior were similar across regions.
... A populationbased study has shown that healthier individuals are more likely to take on a caregiving role [31]. Furthermore, other population-based studies have found that non-Hispanic white, female, adult child caregivers who provide care to their parents, in particular, have a significantly lower rate of death compared to non-caregivers [24,32]. Good physical health may be an expectation for being a caregiver so that they can care for others [31]. ...
Article
Full-text available
One in four Baby Boomers fills the informal caregiver role in the United States. The objectives of this study were to estimate the prevalence of Baby Boomers who are informal caregivers for people living with dementia and compare their physical and mental health status to caregivers for persons with conditions other than dementia using 2015–2018 Behavioral Risk Factor Surveillance System data (n = 10,602). We identified caregiving status (assisting a family member/friend with a long-term illness or disability in the past month, managing personal care, and not caring for a child/grandchild) and whether the care recipient’s major health condition was dementia. We calculated weighted estimates and used chi-square tests and log-binomial regression for comparisons of selected characteristics. Among Baby Boomer caregivers, 15.4% were caring for someone with dementia. Dementia caregivers were more likely to be female, caring for a parent/parent-in-law, and providing care longer than caregivers for persons without dementia. After adjusting for sociodemographic and caregiving characteristics, the prevalence of fair/poor health, frequent mental distress, and chronic conditions were similar across types of caregivers. Although no differences in caregiver’s physical and mental health by care recipient’s dementia status were found, we should underscore the importance of maintaining Baby Boomer caregivers’ health and well-being.
... Bis zu 50% der Betreuenden leiden im Verlauf dieser Aufgabe an psychiatrischen Symptomen (Schulz et al. 1995). Trotzdem empfinden viele Menschen auch eine innere Befriedigung in der Betreuung ihrer Angehörigen, besonders, wenn diese dadurch zu Hause weiter versorgt werden können und nicht in einer stationären Pflegeeinrichtung untergebracht werden müssen (Roth et al. 2013). In der vergangenen Legislaturperiode (2013-2017) wurden eine Reihe von Gesetzesinitiativen auf den Weg gebracht, die nicht nur auf die Pflegebedürftigen und die Pflegeeinrichtungen, sondern auf die pflegenden Angehörigen abzielen. ...
Thesis
Hintergrund: Die Veränderung der Bevölkerungsstruktur im Zuge des demografischen Wandels geht mit der Zunahme von Menschen mit Demenz einher. Die Angehörigen übernehmen den größten Teil der häuslichen Pflege. Mit dieser Entwicklung werden zunehmend assistive Technologien als Unterstützung diskutiert. Ansätze zu Informations-und Beratungsmöglichkeiten sowie zur didaktisch-methodischen Umsetzung von transformativen Konzepten stehen dazu bisher aus. Zielsetzung: Die Arbeit verfolgt das Ziel, die Informations-und Beratungsbedarfe von pflegenden Angehörigen von Menschen mit Demenz zur Integration assistiver Technologien in den Pflegealltag offenzulegen sowie die theoretische Konzeptualisierung eines transformativen Didaktisierungsansatzes zu beschreiben. Methode: Es wurde ein sequenziell-exploratives Studiendesign einer Mixed Methods-Erhebung gewählt. Dazu wurden im qualitativen Erhebungsteil Fokusgruppen mit pflegenden Angehörigen (n=46) sowie Expert*inneninterviews mit Berater*innen (n=5) durchgeführt und um eine quantitative Erhebung mittels standardisierten Fragebogen bei Berater*innen nach SGB XI (n=47) ergänzt. Ergebnisse: Assistive Technologien scheinen in pflegerelevanten Informations-und Beratungsstrukturen bisher keine Rolle zu spielen. Der frühzeitige Kompetenzaufbau wird von pflegenden Angehörigen und von Berater*innen im Rahmen eines angeleiteten Unterstützungsprozesses explizit erbeten. Sie favorisieren das persönliche Erleben assistiver Technologien, z.B. durch moderierte Testmöglichkeiten. Auch wünschen sie sich gezielte Schulungsmöglichkeiten, um neuartige, digital-unterstützte assistive Technologien ausprobieren und Rückmeldungen zur Praxisintegration geben zu können. Trotz des geringen Wissens-und Verbreitungsgrades von assistiven Technologien sehen sie die Notwendigkeit, technische Lösungen in die Praxis zu überführen und die Potenziale für die Unterstützung pflegender Angehöriger von Menschen mit Demenz zu nutzen.
Article
Objectives Higher inflammation has been linked to poor physical and mental health outcomes, and mortality, but few studies have rigorously examined whether changes in perceived stress and depressive symptoms are associated with increased inflammation within family caregivers and non-caregivers in a longitudinal design. Design Longitudinal Study. Setting REasons for Geographic And Racial Differences in Stroke cohort study. Participants Participants included 239 individuals who were not caregivers at baseline but transitioned to providing substantial and sustained caregiving over time. They were initially matched to 241 non-caregiver comparisons on age, sex, race, education, marital status, self-rated health, and history of cardiovascular disease. Blood was drawn at baseline and approximately 9.3 years at follow-up for both groups. Measurements Perceived Stress Scale, Center for Epidemiological Studies-Depression, inflammatory biomarkers, including high-sensitivity C-reactive protein, D dimer, tumor necrosis factor alpha receptor 1, interleukin (IL)-2, IL-6, and IL-10 taken at baseline and follow-up. Results Although at follow-up, caregivers showed significantly greater worsening in perceived stress and depressive symptoms compared to non-caregivers, there were few significant associations between depressive symptoms or perceived stress on inflammation for either group. Inflammation, however, was associated with multiple demographic and health variables, including age, race, obesity, and use of medications for hypertension and diabetes for caregivers and non-caregivers. Conclusions These findings illustrate the complexity of studying the associations between stress, depressive symptoms, and inflammation in older adults, where these associations may depend on demographic, disease, and medication effects. Future studies should examine whether resilience factors may prevent increased inflammation in older caregivers.
Chapter
Caregiving for a family member or loved one is one of the most taxing roles that human beings endure. Caregiver assessments should include exploration into the following areas: mental health, physical health, financial well‐being, social support, and spiritual well‐being. Several screening measures are available to assist providers in supporting caregivers. Some of these assessments can help providers identify areas in which caregivers may need additional support or to assess their overall well‐being. Interventions to support people with chronic health issues often focus on one group or the other rather than the dyad of the caregiver and their loved one. Interventions for caregivers are multidimensional in structure, providing psychological, social/emotional, and educational support. These programs often include a variety of components, such as education, counselling, support groups, problem‐solving, behaviour management training, and respite or adult daycare. Caregiving varies across different cultural and ethnic groups as values surrounding family and elders look different across the world.
Article
Objectives Despite adverse physical and mental health outcomes related to caregiving, family caregivers also experience lower mortality rates compared to noncaregivers. However, research has not yet examined the role of caregiving intensity and religiosity with health and mortality among spousal caregivers. Methods Data include spousal caregivers ( n=5,214 person-wave observations) and noncaregivers ( n=50,311 person-wave observations) from the Health and Retirement Study (2004–2014 waves). Multinomial logistic regression was used to explore how caregiving intensity and religiosity were associated with health and mortality among spousal caregivers, compared health and mortality between caregivers and noncaregiving peers, and examined gender differences in these mechanisms. Results Greater religious salience and attending religious services, although dependent on gender and caregiving intensity, are protective for caregivers’ health and mortality. Discussion Religiosity may buffer adverse effects of caregiving on health and mortality for spousal caregivers. Continuation of prior religiosity may enhance positive aspects of caregiving and decrease caregiver burden.
Article
ntroduction Many informal caregivers experience a significant caregiving burden, which may interfere with their health behaviors. Caregiver health behaviors may vary by disease context, but this has rarely been studied. This study compares the health behaviors of prevalent groups of chronic illness caregivers (i.e., dementia, cancer, chronic obstructive pulmonary disease/emphysema, diabetes) with those of noncaregivers and examines whether caregiving intensity is associated with these behaviors. Methods In 2021, using pooled cross-sectional 2015–2019 Behavioral Risk Factor Surveillance System data, health behaviors (i.e., physical activity, diet, alcohol use, smoking, sleep, and influenza immunization) of caregivers of patients with dementia (n=5,525), cancer (n=4,246), chronic obstructive pulmonary disease/emphysema (n=1,959), and diabetes (n=2,853) and noncaregivers (n=203,848) were compared. Relationships between caregiving intensity (e.g., hours, type of tasks) and caregiver health behaviors were examined. Regression analyses were used to compare groups. Results Compared with noncaregivers, caregiver groups were more likely to report engaging in both risky (i.e., smoking, shorter sleep duration) and health-promoting (i.e., physical activity, vegetable consumption, abstaining from heavy drinking) behaviors, whereas nonsignificant differences were observed for influenza immunization. Longer caregiving hours and providing help with personal care were associated with poorer health behaviors (e.g., shorter sleep duration). Few differences in health behaviors were observed between caregivers of patients with dementia and other caregiver groups. Conclusions Results suggest that caregivers are more likely to engage in both risky and health-promoting behaviors than noncaregivers. Furthermore, findings suggest that greater caregiving responsibilities are associated with certain risky health behaviors. Findings support the development and implementation of strategies to improve caregivers’ health behaviors across disease contexts.
Article
Full-text available
Importance More than 17 million people in the US provide uncompensated care for adults with physical or cognitive limitations. Such caregiving is associated with worse mental and physical health, yet little research has investigated how publicly funded home care might ameliorate these harms. Objective To investigate the association between Medicaid home care services and family caregivers’ health. Design, Setting, and Participants This longitudinal cohort study used data from the 1996 to 2017 Medical Expenditures Panel Survey. Data on all household members were collected in 5 interviews over 2 years. Person-level difference-in-difference models were used to isolate within-person changes associated with new onset of Medicaid home care. The Medical Expenditures Panel Survey longitudinal data sets included 331 202 individuals (approximately 10% excluded owing to loss to follow-up). Adult (age ≥21 years) members of households that contained at least 1 person with limited activities of daily living were included in our study. The analysis itself was performed from March to August of 2020. Exposures New onset of regular (≥1 time per month) Medicaid home care in the household. Main Outcomes and Measures Self-rated mental and physical health (planned prior to beginning the study). Results The study population was 14 013 adults; 7232 were “likely caregivers,” or nondisabled adult coresidents of someone with activities of daily living limitations. Overall, 962 likely caregivers were ever exposed to Medicaid home care in the household; for 563, we observed the onset. Of likely caregivers exposed to Medicaid home care, 479 (50%) were women; 296 (31%) were White non-Hispanic, 309 (31%) were Hispanic or Latinx, and 279 (29%) were Black non-Hispanic individuals, respectively; 326 (34%) had less than a high school education; and 300 (31%) were in or near poverty. Median age of participants was 51 (interquartile range, 39-62) years. New-onset Medicaid home care was associated with a 0.08 standard deviation improvement in likely caregivers’ self-rated mental health (95% CI, 0.01-0.14; P = .02) measured 1 to 6 months after onset, equivalent to a 3.39% improvement (95% CI, 0.05%-6.33%) over their average preonset mental health. No association with self-rated physical health was found (<0.001 standard deviations; 95% CI, −0.06 to 0.06; P = .99). Conclusions and Relevance In this cohort study, Medicaid home care was associated with improvement in caregiver self-rated mental health, but not with any short-term change in self-rated physical health. When evaluating the social value of home care programs, policy makers should consider spillover benefits to caregivers.
Article
In observational studies, investigators have no control over the treatment assignment. The treated and non-treated (that is, control) groups may have large differences on their observed covariates, and these differences can lead to biased estimates of treatment effects. Even traditional covariance analysis adjustments may be inadequate to eliminate this bias. The propensity score, defined as the conditional probability of being treated given the covariates, can be used to balance the covariates in the two groups, and therefore reduce this bias. In order to estimate the propensity score, one must model the distribution of the treatment indicator variable given the observed covariates. Once estimated the propensity score can be used to reduce bias through matching, stratification (subclassification), regression adjustment, or some combination of all three. In this tutorial we discuss the uses of propensity score methods for bias reduction, give references to the literature and illustrate the uses through applied examples. © 1998 John Wiley & Sons, Ltd.
Article
The propensity score is the conditional probability of assignment to a particular treatment given a vector of observed covariates. Previous theoretical arguments have shown that subclassification on the propensity score will balance all observed covariates. Subclassification on an estimated propensity score is illustrated, using observational data on treatments for coronary artery disease. Five subclasses defined by the estimated propensity score are constructed that balance 74 covariates, and thereby provide estimates of treatment effects using direct adjustment. These subclasses are applied within sub-populations, and model-based adjustments are then used to provide estimates of treatment effects within these sub-populations. Two appendixes address theoretical issues related to the application: the effectiveness of subclassification on the propensity score in removing bias, and balancing properties of propensity scores with incomplete data.
Article
In observational studies, investigators have no control over the treatment assignment. The treated and non-treated (that is, control) groups may have large differences on their observed covariates, and these differences can lead to biased estimates of treatment effects. Even traditional covariance analysis adjustments may be inadequate to eliminate this bias. The propensity score, defined as the conditional probability of being treated given the covariates, can be used to balance the covariates in the two groups, and therefore reduce this bias. In order to estimate the propensity score, one must model the distribution of the treatment indicator variable given the observed covariates. Once estimated the propensity score can be used to reduce bias through matching, stratification (subclassification), regression adjustment, or some combination of all three. In this tutorial we discuss the uses of propensity score methods for bias reduction, give references to the literature and illustrate the uses through applied examples.
Article
Background: Terminal illness imposes substantial burdens-economic and otherwise-on patients and caregivers. The cause of these burdens is not understood. Objective: To determine the mechanism for economic and noneconomic burdens of terminal illness and to identify potential ameliorating interventions. Design: In-person interviews of terminally ill patients and their caregivers. Setting: Six randomly selected U.S. sites: Worcester, Massachusetts; St. Louis, Missouri; Tucson, Arizona; Birmingham, Alabama; Brooklyn, New York; and Mesa County, Colorado. Participants: 988 terminally ill patients and 893 caregivers. Measurements: Needs for transportation, nursing care, homemaking, and personal care; subjective perception of economic burden; expenditure of more than 10% of household income on health care costs; caregiver depression and sense of interference with his or her life; and patient consideration of euthanasia or physician-assisted suicide. Results: Of all patients, 34.7% had substantial care needs. Patients who had substantial care needs were more likely to report that they had a subjective sense of economic burden (44.9% compared with 35.3%; difference, 9.6 percentage points [95% Cl, 3.1 to 16.1]; P = 0.005); that 10% of their household income was spent on health care (28.0% compared with 17.0%; difference, 11.0 percentage points [Cl, 4.8 to 17.1]; P ≤ 0.001); and that they or their families had to take out a loan or mortgage, spend their savings, or obtain an additional job (16.3% compared with 10.2%; difference, 6.1 percentage points [Cl, 1.4 to 10.6]; P = 0.004). Patients with substantial care needs were more likely to consider euthanasia or physician-assisted suicide (P = 0.001). Caregivers of these patients were more likely to have depressive symptoms (31.4% compared with 24.8%; difference, 6.6 percentage points [Cl, 0.4 to 12.8]; P = 0.01) and to report that caring for the patients interfered with their lives (35.6% compared with 24.3%; difference, 11.3 percentage points [Cl, 5.0 to 17.7]; P = 0.001). Caregivers of patients whose physicians listened to patients' and caregivers' needs had fewer burdens. Conclusions: Substantial care needs are an important cause of the economic and other burdens imposed by terminal illness. Through empathy, physicians may be able to ameliorate some of these burdens.
Article
Matching members of a treatment group (cases) to members of a no treatment group (controls) is often used in observational studies to reduce bias and approximate a randomized trial. There is often a trade-off when matching cases to controls and two types of bias can be introduced. While trying to maximize exact matches, cases may be excluded due to incomplete matching. While trying to maximize cases, inexact matching may result. Bias is introduced by both incomplete matching and inexact matching. Propensity scores are being used in observational studies to reduce bias. It has been shown that matching on a propensity score can result in similar matched populations. This paper will describe how to reduce matched-pair bias caused by incomplete matching and inexact matching. Cases will be matched to controls on the propensity score using the presented matching algorithm. SAS/STAT LOGISTIC procedure code will be given to create the propensity score. A user-written SAS macro will be given to create a propensity score matched- pair sample using greedy matching techniques. The results of using the presented code, run on a large observational database of myocardial infarction patients, will be given as an example.
Article
There is strong consensus that caring for an elderly individual with disability is burdensome and stressful to many family members and contributes to psychiatric morbidity. Researchers have also suggested that the combination of loss, prolonged distress, the physical demands of caregiving, and biological vulnerabilities of older caregivers may compromise their physiological functioning and increase their risk for physical health problems, leading to increased mortality. To examine the relationship between caregiving demands among older spousal caregivers and 4-year all-cause mortality, controlling for sociodemographic factors, prevalent clinical disease, and subclinical disease at baseline. Prospective population-based cohort study, from 1993 through 1998 with an average of 4.5 years of follow-up. Four US communities. A total of 392 caregivers and 427 noncaregivers aged 66 to 96 years who were living with their spouses. Four-year mortality, based on level of caregiving: (1) spouse not disabled; (2) spouse disabled and not helping; (3) spouse disabled and helping with no strain reported; or(4) spouse disabled and helping with mental or emotional strain reported. After 4 years of follow-up, 103 participants (12.6%) died. After adjusting for sociodemographic factors, prevalent disease, and subclinical cardiovascular disease, participants who were providing care and experiencing caregiver strain had mortality risks that were 63% higher than noncaregiving controls (relative risk [RR], 1.63; 95% confidence interval [CI], 1.00-2.65). Participants who were providing care but not experiencing strain (RR, 1.08; 95 % CI, 0.61-1.90) and those with a disabled spouse who were not providing care (RR, 1.37; 95% CI, 0.73-2.58) did not have elevated adjusted mortality rates relative to the noncaregiving controls. Our study suggests that being a caregiver who is experiencing mental or emotional strain is an independent risk factor for mortality among elderly spousal caregivers. Caregivers who report strain associated with caregiving are more likely to die than noncaregiving controls.
Article
Objectives: We sought to test the hypothesis that providing help to others predicts a reduced association between stress and mortality. Methods: We examined data from participants (n = 846) in a study in the Detroit, Michigan, area. Participants completed baseline interviews that assessed past-year stressful events and whether the participant had provided tangible assistance to friends or family members. Participant mortality and time to death was monitored for 5 years by way of newspaper obituaries and monthly state death-record tapes. Results: When we adjusted for age, baseline health and functioning, and key psychosocial variables, Cox proportional hazard models for mortality revealed a significant interaction between helping behavior and stressful events (hazard ratio [HR] = 0.58; P < .05; 95% confidence interval [CI] = 0.35, 0.98). Specifically, stress did not predict mortality risk among individuals who provided help to others in the past year (HR = 0.96; 95% CI = 0.79, 1.18), but stress did predict mortality among those who did not provide help to others (HR = 1.30; P < .05; 95% CI = 1.05, 1.62). Conclusions: Helping others predicted reduced mortality specifically by buffering the association between stress and mortality.