Informal Care and Sleep Disturbance Among Caregivers in Paid Work: Longitudinal Analyses From a Large Community-Based Swedish Cohort Study
Informal Care and Sleep Disturbance Among Caregivers in Paid Work: Longitudinal Analyses From a Large Community-Based Swedish Cohort Study
Lawrence B. Sacco Emalila:
Constanze Leineweber 0
Loretta G. Platts
0 S,UtrKe,ss Research Institute, Stockholm University, Stockholm, Sweden Work performed: Stress Research Institute, Stockholm University , Stockholm , Sweden
1 Department of Global Health and Social Medicine, Institute of Gerontology, King's College London , London
Study Objectives: To examine cross-sectionally and prospectively whether informal caregiving is related to sleep disturbance among caregivers in paid work. Methods: ParticipantsN( = 21604) in paid work from the Swedish Longitudinal Occupational Survey of Health. Sleeping problems were measured with a validated scale of sleep disturbance (Karolinska Sleep Questionnaire). Random-effects modeling was used to examine the cross-sectional association between informal caregiving (self-reports: none, up to 5 hours per week, over 5 hours per week) and sleep disturbance. Potential sociodemographic and health confounders were controlled for and interactions between caregiving and gender included. Longitudinal random-effects modeling of the effects of changes in reported informal caregiving upon sleep disturbance and change in sleep disturbance was performed. Results: In multivariate analyses controlling for sociodemographics, health factors, and work hours, informal caregiving was associated cross-sectionally with sleep disturbance in a dose-response relationship (compared with no caregiving, up to 5 hours of caregivingβ: = 0.03; 95% CI: 0.01, 0.06, and over 5 hours:β = 0.08; 95% CI: 0.02, 0.13), results which varied by gender. Cessation of caregiving was associated with reductions in sleep disturbanβce=(−0.08; 95% CI: −0.13, −0.04). Conclusions: This study provides evidence for a causal association of provision of informal care upon self-reported sleep disturbance. Even low-intensity care provision was related to sleep disturbance among this sample of carers in paid work. The results highlight the importance of addressing sleep disturbance in caregivers.
carer; cohort study; epidemiology; informal care; Karolinska Sleep Questionnaire; SLOSH; sleep disturbance
Sleep disturbances are a common health problem. One likely cause of sleep disturbance is provision of informal care
to an elderly, ill, or disabled person. In this prospective analysis of 21 604 participants in paid work, caregiving was -associ
ated with self-reported sleep disturbance, particularly at higher numbers of hours of informal care. The effects of -provid
ing informal care at different intensities varied between women and men. In analyses of change in caregiving, providing
caregiving at the first but not the second wave was associated with reduction in sleep disturbance. These findings support
the hypothesis that provision of informal care affects the carer’s sleep quality. Further large-scale studies to examine the
mechanisms of sleep disturbance among in-work carers are required.
Introduction and longitudinal associations of caregiving with sleep -dis
turbance. Important health and demographic confounders of
Sleeping problems are common: in Swedish adults aged 18–84, the caregiving-sleep relationship are controlled for. Firstly, we
insomnia disorder, that is, insomnia symptoms and daytime examine whether greater intensity of caring, operationalized
consequences, has an estimated prevalence of around 10.5% and using weekly hours of care provided, is associated with sleep
symptoms of insomnia are reported by one-quarter of the pop-u disturbance in cross-sectional analyses. We hypothesize that
]. Evidence is growing that people with insomnia sym-p providing informal care will be associated with sleep distu-rb
toms have a raised risk of developing physical illnes[2s], specific- ance, although the benefits of “role enhancement” may mean
ally cardiometabolic disease, occupational injuries, and all-cause that carers providing low-intensity caregiving experience lower
mortality[3, 4]. In terms of mental illness, disturbed sleep and levels of sleep disturbance than noncaregivers. Since women
tiredness are common symptoms of mood disorders such as tend to provide more informal care than me[n7], and reports
depression . Consequently, addressing the causes of insomnia of insomnia symptoms are more frequent for women than for
may contribute to improvements in population heal[t6h]. men [
], gender interactions are included in these analyses in
One likely important cause of sleep disturbance is provision case the relationship between caregiving and sleep disturbance
of informal care to an elderly, ill, or disabled person. Informal differs by gender. Secondly, in longitudinal analyses, we exa-m
care is assistance provided by people from the intimate envir-on ine the effects of changes in caregiving upon sleep disturbance
ment of the dependent person, who do not receive any training and upon change in sleep disturbance. We hypothesize that
or economic compensation. It is common: in the United States, commencing informal caregiving will be associated with greater
16.6% of adults[
] and in Sweden, 10.9% of adults are cur- sleep disturbance and cessation of caregiving with reduced
rently informal carers, rates that are expected to increase in linesleep disturbance. To the best of our knowledge, only one study
with population aging and cut-backs in publicly funded ca[r9e]. has previously examined the relationship between starting and
Informal caregiving has been associated, albeit inconsistently, stopping provision of informal care on sleeping problems, in a
with a range of poor health outcomes among carers in large small sample of spousal Alzheimer caregiver[2s6]. In short, the
cross-sectional and prospective studie[s10–13], depending on longitudinal analyses performed in this study may offer a-dd
the characteristics of the carer and person cared for, type of itional evidence for a causal relationship between informal c-are
health outcome, type of care provided, and duration of follow- giving and sleep disturbance in a large sample of in-work carers.
up [14, 15]. However, some studies find that caregiving may be
neutral or benefit carers’ health, e.g., by providing an additional
rewarding role r(ole enhancement) and improving the quality of
]. Among other factors, it is pro-b METHODS
able that the intensity and duration of care provision determine
whether and to what degree caring has a positive or negative Study Population
impact on the caregiver. Participants in the Swedish Longitudinal Occupational Survey
Although it has been argued that various aspects of the care-giv of Health (SLOSH) formed the study population. SLOSH is a
ing role impact on sleep specificall[y21], recent reviews of the ev-i biennial postal survey, which follows a subsample of gainfully
dence note that few adequately powered, community-based, lon-gi employed people aged 16–64 from the Swedish Labour Force
tudinal studies have examined the impacts of caregiving on sleep Survey who were recruited into the Swedish Work Environment
]. Existing studies point to poorer sleep among caregivers, Surveys (SWES) 2003–2011 [
]. Both SLOSH and the present
but apart from certain exceptions, tend to be cross-sectional, use study have been approved by the Regional Research Ethics Board
small samples which make effects hard to detect, or only examine in Stockholm. All participants provided informed consent.
specific groups of carers without comparing them to the general SLOSH participants respond to one of the two versions of
population. Important confounders such as the caregiver’s edu-ca the postal questionnaire: an in-work questionnaire for those
tion level or health may not be controlled for, factors which lead tocurrently in paid work at least 30% of full-time and another for
both selection into the caregiving role and sleep disturban[1c7e]. those working less or who are outside the labor force whether
Stronger evidence would also be provided by prospective designs permanently or temporarily. The present study is based on
linking changing provision of care to changes in sleep, but almost information provided in the in-work questionnaires only from
no research has so far done so[25, 26]. 2010, 2012, 2014, and 2016. Response rates to the SLOSH que-s
Although recent reviews have argued for greater focus on tionnaires were 57% in 2010 and 2012, 53% in 2014, and 51% in
subpopulations of carers in terms of factors such as ag[2e7], 2016. In total, 21604 participants responding to at least one
inor care recipient characteristic[s28], to the best of our know- work questionnaire were included in the study, providing 42928
ledge, prior research has not examined the impact of ca-re observations for the analysis. The change analyses required p-ar
giving on sleep in the important and large population of -car ticipants to respond to the in-work questionnaire in at least two
egivers who are simultaneously engaged in paid work. This is consecutive waves, a requirement which reduced the sample to
despite evidence that difficulties combining work tasks with 12253 people.
family obligations can generate stress and health proble[m29s,
30], Specifically, the demands of paid work may conflict with
requirements to provide informal care, which may lead troole Sleep Disturbance
overload, in which an individual with a finite amount of time and Self-reported sleep disturbance was measured at each wave
energy experiences strain from trying to fulfil multiple ro[1l6e]s. using a validated measure of disturbed sleep composed of four
Consequently, this study focuses on caregivers and no-n questions from the Karolinska Sleep Questionnair[e32–34]. The
caregivers who are in paid work, and uses a large community- measure contains the core symptoms of insomnia; specifically
based prospective dataset to examine both the cross-sectional participants were asked how often they had been disturbed in
the previous 3 months by difficulties falling asleep, repeated the previous 3 months (specifically headache, pain in the neck
awakenings with difficulties going back to sleep, premature or shoulders, pain in the lower back, or other pain). For both
(final) awakening, or disturbed or restless sleep, with response variables, participants could respond: no; yes, but this does not
options ranging from 0 (“never”) to 5 (“always/five times a affect their life at all; yes, this affects their life a little; or yes, this
week”). Item nonresponse rates were low: 1.1% for one missing affects their life a lot. The two variables were dichotomized such
item, 0.2% for each of two and three missing items, and 1.1% that reports that chronic disease or pain/discomfort affected the
for four missing items. An average sleep disturbance score was participant’s life either a little or a lot were coded as 1, otherwise
calculated for participants who provided responses to at least they were coded as 0.
three of the four items. Change in sleep disturbance from one Self-rated health and reports of depressive symptoms were
wave to the next was calculated by subtracting the score in the included in the analyses as both potential confounders and
earlier wave from the score in the next wave. mediators of the relationship between informal caregiving and
sleep disturbance. To measure self-rated health, respondents
were asked: “How would you rate your general state of health”
Provision of Informal Care with the responses: very good, good, neither good nor bad, quite
At each study wave, questions about participants’ time use in poor, or very poor, a single-item measure of general physical
a typical work week were used to identify how much care in and mental health which predicts mortali[t3y9, 40]. A brief
sixhours participants provided for a relative other than a child or item version of the Symptom Checklist Core Depression Scale
grandchild. Original response categories were 0 hour (85.8% of (SCL-CD6) was used to measure depressive symptomatology
sample members), 1–5 hours (12.4%), 6–10 hours (1.2%), 11–15 [41–43]. Questions in this scale evaluate how much (from 0 = not
hours (0.3%), and >15 hours (0.4%). The last three categories were at all to 4 = very much) participants were troubled by lethargy
collapsed into >5 hours per week since relatively few respo-nd or low energy, feeling blue, blaming oneself, excessive wor-ry
ents provided high numbers of hours of informal care. A dich-ot ing, feelings of no interest in things, and feeling that everything
omized variable of caregiving was used in analyses examining is an effort. An overall score (0–24) was obtained by summing
change between consecutive waves: no care-giving and prov-i responses to all items.
sion of any care (at least 1 hour per week). This generated the Self-rated health and depressive symptoms were considered
combinations: no caregiving at either wave; no caregiving at the to be potential confounders of the informal caring–sleep r-ela
first wave, caregiving at the second wave; caregiving at the first tionship because poor self-rated health and depressive sym-p
wave but not at the second; and caregiving at both waves. toms might prevent carers from providing help to relatives and
also be likely to increase sleep disturbance. However, self-rated
health and depressive symptoms might also mediate the re-la
Covariates tionship from informal caregiving to sleep disturbance; i.e., they
may be on the causal path. Previous research has highlighted
Selection of covariates was performed by drawing directed ac-yc the importance of physical fatigue and tiredness to how p-eo
lic graphs based on existing knowledge of factors which might ple evaluate their self-rated healt[4h0, 44]. Furthermore, fatigue
affect both sleep disturbance and possibilities to provide in-for and sleep disturbance are common symptoms of depressive
]. The sociodemographic variables of gender, age, episodes in the ICD-10 classification of mental and behavioral
marital status, and education were included in the analyses as disorders.
possible confounders of the relationship between caring and Paid work intensity was measured in four categories indic-at
sleep disturbance. Information regarding gender, age, education, ing time spent weekly in paid work created from participants’
and marital status (0 = married/cohabiting and 1 = nonmarried/ reports (<10 hours, 10–19 hours, 20–29 hours, and ≥30 hours). This
divorced/widowed/single) was obtained through linkage using covariate could be both a confounder or mediator of the in-for
individual person numbers to Statistics Sweden’s Longitudinal mal caring–sleep relationship: a confounder because time in paid
Individual Data Base (LISA) administrative register. Age was c-en work could affect both the number of hours dedicated to informal
tered at its mean of 50 years. In order to model nonlinear r-ela caregiving and sleep disturbance, and a mediator because hours
tionships between age and sleep disturbance, squared and cubic spent providing informal care could lead to a decrease in hours
functions of age were generated. Highest education level was of paid work, which in turn may affect sleep disturbance if, for
grouped into the following categories: compulsory schooling; example, this generated financial difficulti[e4s5].
2 years upper secondary/vocational training or 4 years upper
secondary; university or equivalent shorter than 3 years; and at
least 3 years of university or equivalent.
Two variables recorded participants’ perceptions of whether Statistical Methods
their lives were affected by (1) any of a range of chronic co-n Statistical analysis was carried out in Stata 14.2 (Stata
ditions and (2) physical pain or discomfort. These variables Corporation, College Station, TX, USA). We performed r-an
were included as confounders because such health problems dom intercept (random effect) modeling, which allows the
might prevent carers from providing help to relatives and are use of full information provided by longitudinal data, while
also likely to increase sleep disturbanc[e36–38]. Specifically, accounting for the dependency of repeated measures within
participants indicated whether they had any of the foll-ow individuals . These models use both the between- and
ing chronic conditions (hypertension, cardiovascular disease, within-subject components of the variability that are present
diabetes, rheumatic disorder, musculoskeletal disorder, m-en in longitudinal data.
tal illness, asthma, obstructive pulmonary disease, migraine, Two main analyses were performed, which both present
physical disability, cancer, and other illness) and in a separate results from the random intercept models as regression c-oef
question whether they had experienced pain or discomfort over ficients with 95% confidence intervals. First, cross-sectional
associations between intensity of informal caregiving (no ca-re disturbance even after controlling for the average older age of
giving, up to 5 hours, and over 5 hours) and sleep disturbance caregivers.
were examined. The reference group contained participants In the random effects model, presented inTable 2, informal
who provided no informal care. These bivariate associations are care provision was associated with sleep disturbance before
presented as Model 1. In Model 2, variables were included that (Model 1) and after adjustment for possible socioeconomic and
were considered to be potential confounders (sociodemograp-h health confounders (Model 2). Specifically, after adjustment,
ics, physical pain, and chronic disease) because they might affect compared with participants who were not providing informal
both sleep disturbance and possibilities to provide informal care, those who were providing care up to 5 hours per week
care. In Model 3, additional variables (self-rated health, dep-res reported higher levels of sleep disturbancβe =( 0.07; 95% CI: 0.04,
sive symptoms, and hours in paid work) were included which 0.09), whereas those who were providing over 5 hours of care
might be both confounders and mediators of the informal c-ar reported highest levels of sleep disturbanceβ (= 0.17; 95% CI:
ing–sleep relationship. In other words, in addition to being likely 0.11, 0.23). In Model 3, we adjusted for depressive symptoms,
confounders, these factors may also lie on the causal pathway self-rated health and hours in paid work, and variables which
from informal caregiving to sleep disturbance. Consequently, could be both confounders and mediators of the association
adjusting on them may remove part or all of any effect of in-for between caregiving and sleep. Following this adjustment, ca-re
mal caregiving on sleep disturbance (overadjustment). Potential giving was still associated with higher levels of reported sleep
interactions between gender and care-giving intensity were also disturbance (up to 5 hours of careβ: = 0.03; 95% CI: 0.01, 0.06 and
examined (Model 4). over 5 hours of care:β = 0.08; 95% CI: 0.02, 0.13).
Second, we examined changes in informal caregiving Gender interacted significantly with carinTgab(le 2, Model 4
(whether providing any care or not) between two consecutive and Figure 1). Men who did not provide care and who provided
waves in relation to sleep disturbance at the second of the up to 5 hours of care had similar levels of sleep disturbance.
two waves, as well as in relation to change in sleep distu-rb Both groups had significantly lower sleep disturbance than men
ance between the waves. All of the changes between the four who provided over 5 hours of care. Women who provided up to
available waves were simultaneously modeled in random int-er 5 hours of informal care weekly reported slightly more sleep
cept models. Covariates were taken from the second of the two disturbance than women who provided no care; differences
waves. The reference group contained participants providing no between those groups and women providing more than 5 hours
informal care in either of the two waves. of care were not discernibleF(igure 1).
In Table 3, random effects models of change in caregiving
on both sleep disturbance and change in sleep disturbance are
RESULTS presented. The left-hand set of analyses present change in ca-re
Descriptive and bivariate statistics of the total study sample giving from one wave to the next on sleep disturbance at the
(N = 21604) are presented inTable 1 using information from second of the two waves. Compared with participants who did
the baseline survey year for each individual. Although most of not provide informal care in either wave, those who provided
the sample (18531 participants) did not report providing in-for care during at least one time point had greater sleep dist-urb
mal care at baseline, 2675 participants (12.4%) reported pro-vid ance both before (Model 1) and after adjustment (Model 2) for a
ing informal care up to 5 hours per week, and 398 participants range of possible confounders at the current wave. Specifically,
(1.8%) reported higher number of hours of care. Caregivers were compared with participants who did not provide informal care
more likely to be female, be older, have a lower education level, in either wave, those who provided care at both waves had s-ig
report that their life was affected by physical pain or discomfort nificantly greater sleep disturbancβe =( 0.10; 95% CI: 0.04, 0.15),
or by chronic illness, report poorer self-rated health, depressive as did those who provided care only at the second of the two
symptoms, and sleep disturbance, and be in paid work under 20 waves (β = 0.07; 95% CI: 0.03, 0.12), and those who provided care
hours per week. In sensitivity analyses (not shown), caregivers at the first, but not at the second wavβe =( 0.05; 95% CI: 0.00,
had poorer physical and mental health as well as greater sleep 0.10). Results from Wald tests showed that differences in sleep
Caregiving > 5 h per
disturbance among these three categories in which participants at both waves:χ2: 4.45, p = .035). Performing caregiving in both
provided care in at least one wave were not significant at the 5% waves and providing care only at the current wave were not
level. Only the coefficient for high sleep disturbance for those associated with change in sleep disturbance compared with the
who were caring in both waves remained significant at the 95% reference group composed of participants who did not provide
level after including variables in Model 3 (self-rated health, care at either wave.
depressive symptoms, and time spent in paid work), which may
be both confounders and mediators of the caregiving–sleep
The right-hand set of analyses present change in caregiving In this large sample of Swedish people in paid work, prov-id
from one wave to the next upon change in sleep disturbance ing informal care predicted higher levels of self-reported sleep
over the same period. Compared with participants who did not disturbance, particularly when more than 5 hours of care per
provide informal care in either wave, providing caregiving at the week were provided, associations which were robust to incl-u
previous but not the current wave was associated with a red-uc sion of demographic and health covariates as well as time in
tion in sleep disturbance over the same time period, results paid work. These results correspond to previous research su-g
which changed little after adjustment for possible confounders gesting that informal caregiving impacts on self-reported and
and mediators (Model 2:β = −0.07; 95% CI: −0.12, −0.02 and Model objective sleep[
]. and that providing informal care for a
3: β = −0.08; 95% CI: −0.13, −0.04). Results from Wald tests showed higher number of hours is more strongly associated with sleep
that this group who provided care only at the first wave also disturbance[48, 49], Studies have shown that provision of nigh-t
reported a reduction in sleep disturbance in comparison to the time care is associated with poorer self-reported slee[4p9], as
other two caregiving groups ( compared with caregiving only at is (albeit inconsistently) when carers and care recipients are
the second wave: χ2: 7.85, p = .005; compared with caregiving coresident[50–52], It might be expected that providing relatively
Table 2. Sleep Disturbance in Relation to Level of Informal Caregiving (Swedish Longitudinal Occupational Survey of HeNa l=t h21, 604, Random
low intensity of informal care (up to 5 hours a week) would be with little or no choice about engaging in caregivin[5g5]. The
associated with better sleep through the mechanism of role impacts of caregiving upon sleep disturbance extend beyond
enhancement. However, in this Swedish working sample, even the time taken for care tasks: caregivers may be on 24-hour call
low intensity caregiving was associated with self-reported sleep [21, 56] and the emotional labor involved in caregiving can g-en
disturbance, suggesting that in-work carers may be experie-nc erate substantial worry and distress. Accordingly, introduction
ing stress generated by role overload in managing the comp-et of depressive symptoms and self-rated health into the models
ing commitments of paid work and informal caregiving. attenuated the relationship between provision of informal care
In terms of the stress process model, in which primary stre-ss and sleep disturbance. This result accords with prior research
ors generated by informal caregiving lead to problems in other which has highlighted the comorbidity of insomnia symptoms
domains, such as job-caregiving conflicts[
], such role over- with depression[
] and self-rated healt h[
], and found
crossload may be generated due to the difficulty of apportioning time sectional and longitudinal associations between sleep and affect
and energy to two relatively inflexible activiti[e5s4]. In-work in caregivers[
25, 58, 59
carers have partial control at best over the timing of car-egiv In contrast to a previous study which failed to find gender
ing and paid work activities, leading to scheduling difficulties differences in the impact of caregiving upon sleeping problems
(e.g., providing care at unpredictable times, at night, and during , we found significant interactions between gender and inf-or
the working week), and the difficulty of catching up on sleep mal caring in relation to sleep disturbance, specifically similar
during the day due to the fixed timing of most jobs. Informal levels of sleep disturbance in men providing no caregiving and
caregiving, unlike other unremunerated activities, is often p-er up to 5 hours of caregiving and much higher sleep disturbance
formed out of a sense of moral obligation, particularly when the among men providing more than 5 hours of care. Women had
care recipient is a close family member, leaving some caregivers higher levels of sleep disturbance overall; differences among
CI = confidence interval.
*p < .05; **p < .01; ***p < .001.
Model 1 was unadjusted.
Model 2 included gender, age, age-squared, age-cubed, education level, marital status, pain, and chronic disease at the second of the two waves.
Model 3 additionally contained self-rated health, depressive symptoms, and hours in paid work at the second of the two waves.
aThe outcome is sleep disturbance at1.t
bThe outcome is change in sleep disturbance between−1tand t1.
women by caregiving status were smaller than among men. It of spousal death on nighttime wakening and sleep time (mea-s
is possible that gender differences in sleep disturbance at m-od ured with actigraph data), but no effect on self-reported sleep
erate levels of caregiving are due to different tasks being p-er complaints.
formed by male and female carers, e.g., more personal care and
routine tasks being performed by female carer[s60,61]. Higher
intensity caregiving is more often spousal care, where gender Strengths and Limitations
differences in care tasks are smalle[r60]. It would be valuable for The major strengths of this study are its use of a validated
future research to examine gender differences in large samples, measure of sleep disturbance with good measurement pro-p
where information about the carer burden, who is cared for and erties in a longitudinal analysis of a large sample with a wide
the tasks performed, is available, in order to examine possible range of demographic and health covariates. In addition, the
mechanisms generating gendered differences in the impact of random effects model is a robust approach to dealing with
caregiving upon sleep disturbance. unbalanced panels and missing data due to attrition, which is
Relatively little prior research has examined the relationship generally found in longitudinal analyses. Rather than having to
between starting and stopping provision of informal care and assume that data are missing completely at random, random
sleep disturbance, an approach which provides additional e-vi effects models make the less restrictive missing at random
dence for the direction of effects. We found that, compared with assumption, i.e., the propensity for data to be missing is not
not providing informal care in either wave, patterns of care-giv related to the missing data after accounting for relationships
ing between two study waves that would correspond to beg-in with observed data[63–65]. However, the study has certain lim- i
ning, terminating, or continuing to provide informal care were tations: first, although the original sample of the SLOSH study
associated with higher sleep disturbance at the second of those was drawn from a representative sample of the Swedish pop-u
study waves. The finding that reporting providing care only at lation within the age-range 16–64 years (i.e., the Swedish Labour
the first wave was associated with sleep disturbance at the -sec Force Survey), the findings are not generalizable to those wo-rk
ond wave is in line with qualitative research suggesting that ing less than 30% of full-time or outside this age range. In a-dd
sleep disturbance extends beyond the end of informal care p-ro ition, selective attrition from the SLOSH study has taken place,
vision . with SLOSH respondents more likely than nonrespondents to
In analyses relating changes in sleep disturbance from be older, female, married, Swedish-born, with university qua-li
one wave to the next to patterns of caregiving over the same fications, and to work in the government sector. Second, it is
period, individuals who provided informal care only at the first possible that individuals with disturbed sleep are less likely to
wave (which would correspond to ceasing caregiving) expe-ri provide informal care, as a result of the impact of health c-on
enced improved self-reported sleep compared with the ref-er ditions associated with disturbed sleep (e.g., depression and
ence group of participants not providing informal care at either chronic pain). Although this would have a conservative effect on
wave. These results point to cessation of caregiving generating the results, weakening the caregiving–sleep disturbance ass-oci
reductions in sleep disturbance, even if not to the level of those ations, such reverse causation has been reduced by taking such
who had not provided informal care at either wave. They stand health variables into account. Third, although we were able to
in contrast to the sole earlier study examining transitions out control for a wide range of demographic and health variables,
of caregiving, which used a small sample of spousal Alzheimer there may be residual confounding related to unobserved ch-ar
caregivers and differentiated caregiving ceasing as a result of acteristics which could generate a spurious positive association
spousal institutionalization or deat[2h6]. That study, by von between participation in informal care and sleep disturbance.
Känel et al., found no impact on self-reported or objective sleep Consequently, we performed a sensitivity analysis using fixed
measures of spousal institutionalization and a negative impact effect modeling, which partials out unobserved time-invariant
individual differences as well as baseline differences in sleep supported by FAS (2005-0734) and the Swedish Research Council
disturbance by utilizing only information on changes in depe-nd (VR, 2009-6192 and 2013-1645). The work was carried out within
ent and independent variables within individuals. The results the framework of the Stockholm Stress Center, a FORTE Centre
are presented inSupplementary Table S1and confirm the con- of Excellence (FORTE, 2009-1758).
clusions presented in the main analyses. Fourth, measures of
care-giving were only provided at each wave; therefore, it was
not possible to ascertain what was happening between the Disclosure statement
waves, e.g., whether a carer providing care at two consecutive None declared.
waves had been providing care continuously during the 2-year
period between those waves. Last, it was only possible to exa-m
ine associations with weekly hours of care provision because the References
SLOSH study lacks information about the nature of tasks -per
formed and characteristics of the care recipient (e.g., whether
co-resident, nature of disability, or illness), an important avenue
for future research in large, community-based samples.
1. Mallon L , Broman JE , Akerstedt T , Hetta J . Insomnia in Sw-e den: a population-based survey . Sleep Disord . 2014 ; 2014: Article ID 843126, 1 - 7 . doi: 10 .1155/ 2014 /843126.
2. Hall MA , Fernandez-Mendoza J , Kline CE , Vgontzas AN .
Concluding Remarks Principles and Practice of Sleep Medicine. 6th ed . Philadelphia, PA: Elsevier; 2017 : 794 - 803 .
This study has found that among workers, providing informal 3 . Kecklund G , Axelsson J . Health consequences of shift work care is an independent predictor of sleep disturbance after c-on and insufficient sleep . BMJ . 2016 ; 355 : i5210 .
trolling for demographic and health variables. Even low-in-ten 4 . Luyster FS , Strollo PJJr , Zee PC , Walsh JK ; Boards of Directors sity caregiving, measured in hours per week, was associated of the American Academy of Sleep Medicine and the Sleep with self-reported sleep disturbance . Sleep disturbance among Research Society. Sleep: a health imperativeS.leep . 2012 ; carers is an under-recognized and under-treated problem [ 24 ], 35 ( 6 ): 727 - 734 .
which is related to a variety of serious health outcomes, includ- 5. World Health OrganizationT.he ICD-10 Classification of Mening major depressive disorder and physical health complaints. tal and Behavioural Disorders: Clinical Descriptions and DiagThe data are from Sweden, a country with a welfare model a-im nostic Guidelines . Vol 1 . Geneva , Switzerland: World Health ing to minimize conflict between paid work and caring comm- it Organization; 1992 .
ments among informal carers, although recent years have seen 6 . Rod NH , Vahtera J , Westerlund HSleep disturbances and cuts to formal care service[s66]. A promising avenue for further cause-specific mortality: results from the GAZEL cohort research may be to investigate the caregiving-sleep distu-rb study . Am J Epidemiol . 2011 ; 173 ( 3 ): 300 - 309 .
7. National Alliance for Caregiving. AARP Public Policy In-sti ance association in diverse welfare and social care contex[t6s7]. tute. Caregiving in the U .S. 2015 . Washington, DC: National That a caregiving-sleep association was observed in a country Alliance for Caregiving and AARP; 2015 .
with relatively comprehensive formal care provision suggests 8 . Berglund E , Lytsy P , Westerling R . Health and wellbeing in the need for policy actors everywhere to consider measures to informal caregivers and non-caregivers: a comparative support employed caregivers, concerning both government and cross-sectional study of the Swedish general population .
employers, as well as to protect formal care services . Since it Health Qual Life Outcomes . 2015 ; 13 : 109 . doi: 10 .1186/s12955 - appears that informal carers are at a high risk of sleep dist-urb 015-0309-2 ance, improved recognition and management by healthcare 9 . Agree E , Glaser K. Demography of informal caregiving. In: staff of sleeping problems in carers may also improve carers ' Uhlenberg P, ed. International Handbook of Population Aging.
health and quality of life . Vol1 . Netherlands: Springer; 2009 : 647 - 668 .
10. Buyck JF , Ankri J , Dugravot AInformal caregiving and the risk for coronary heart disease: the Whitehall II studJGy . erSupplementary material ontol A Biol Sci Med Sci . 2013 ; 68 ( 10 ): 1316 - 1323 .
11. Choi KS , Stewart R , Dewey M. Participation in productive Supplementary materials are available at SLEEP online activities and depression among older Europeans: survey of health, ageing and retirement in Europe (SHAREI)n .t J Geriatr Psychiatry . 2013 ; 28 ( 11 ): 1157 - 1165 .
Acknowledgments 12. Legg L , Weir CJ , Langhorne P , Smith LN , Stott DJ . Is inf-or mal caregiving independently associated with poor health? We gratefully acknowledge advice from Göran Kecklund during A population-based study . J Epidemiol Community Health.
the preparation of this manuscript . 2013 ; 67 ( 1 ): 95 - 97 .
13. Vlachantoni A , Robards J , Falkingham J , Evandrou M. T-ra jectories of informal care and healthS .SM - Popul Health .
Funding 2016 ; 2 : 495 - 501 . doi: 10 .1016/j.ssmph. 2016 . 05 .009 14. Vlachantoni A , Evandrou M , Falkingham J , Robards J . Inf-or LBS was supported by the British cross-research council mal care, health and mortalitMy .aturitas. 2013 ; 74 ( 2 ): 114 - Lifelong Health and Wellbeing (LLHW) program under Extending 118 .
Working Lives as part of an interdisciplinary consortium on 15 . Pinquart M , Sörensen S . Differences between caregivers Wellbeing, Health, Retirement, and the Lifecourse (WHERL) (ES/ and noncaregivers in psychological health and physical L002825/1). LGP received funding from the Swedish Council for health: a meta-analysisP . sychol Aging . 2003 ; 18 ( 2 ): 250 - Working Life and Social Research (FAS, 2012 - 1743 ). SLOSH was 267.
16. Rozario PA , Morrow-Howell N , Hinterlong JE . Role enhan-ce place demands, control, support, and sleep problemSsle .ep.
ment or role strain assessing the impact of multiple -pro 2011 ; 34 ( 10 ): 1403 - 1410 .
ductive roles on older caregiver well-beingR . es Aging . 35 . Greenland S , Pearl J , Robins JM . Causal diagrams for epid-e 2004 ; 26 ( 4 ): 413 - 428 . miologic research.Epidemiology. 1999 ; 10 ( 1 ): 37 - 48 .
17. Schulz R , Sherwood PR . Physical and mental health effects 36 . Foley D , Ancoli-Israel S , Britz P , Walsh J . Sleep disturbances of family caregiving . Am J Nurs . 2008 ; 108 ( 9 Suppl) : 23 - 7 . and chronic disease in older adults: results of the 2003 18 . Broese van Groenou MI, de Boer A , Iedema J . Positive and National Sleep Foundation Sleep in America SurveJy . Psynegative evaluation of caregiving among three different chosom Res . 2004 ; 56 ( 5 ): 497 - 502 .
types of informal care relationshipsE . ur J Ageing . 2013 ; 37. McCann JJ , Hebert LE , Bienias JL , Morris MC , Evans DA . Pr-e 10 ( 4 ): 301 - 311 . dictors of beginning and ending caregiving during a 3-year 19 . O'Reilly D , Connolly S , Rosato M , Patterson C . Is caring a-sso period in a biracial community population of older adults .
ciated with an increased risk of mortality? A longitudinal Am J Public Health . 2004 ; 94 ( 10 ): 1800 - 1806 .
study.Soc Sci Med . 2008 ; 67 ( 8 ): 1282 - 1290 . 38 . Lee C , Gramotnev H . Transitions into and out of careg-iv 20 . O'Reilly D , Rosato M , Maguire A . Caregiving reduces mort-al ing: health and social characteristics of mid-age Australian ity risk for most caregivers: a census-based record linkage women . Psychol Health . 2007 ; 22 ( 2 ): 193 - 209 .
study. Int J Epidemiol . 2015 ; 44 ( 6 ): 1959 - 1969 . 39. Quesnel-Vallée A . Self-rated health: caught in the cr-oss 21 . Arber S , Venn S. Caregiving at night: understanding the fire of the quest for 'true ' healthIn?t J Epidemiol . 2007 ; 36 ( 6): impact on carers . J Aging Stud . 2011 ; 25 ( 2 ): 155 - 165 . 1161 - 1164 .
22. Kotronoulas G , Wengstrom Y , Kearney N. Sleep patterns and 40 . Singh-Manoux A , Martikainen P , Ferrie J , Zins M , Marmot M , sleep-impairing factors of persons providing informal care Goldberg M. What does self rated health measure? Results for people with cancer: a critical review of the literature. from the British Whitehall II and French Gazel cohort s-tud Cancer Nurs . 2013 ; 36 ( 1 ): E1 - 15 . ies. J Epidemiol Community Health . 2006 ; 60 ( 4 ): 364 - 372 .
23. Fonareva I , Oken BS . Physiological and functional con-se 41 . Derogatis LR , Lipman RS , Rickels K , Uhlenhuth EH , Covi quences of caregiving for relatives with dementiaIn .t Psy- L. The Hopkins Symptom Checklist (HSCL): a self-report chogeriatr . 2014 ; 26 ( 5 ): 725 - 747 . symptom inventory. Behav Sci . 1974 ; 19 ( 1 ): 1 - 15 .
24. Peng HL , Chang YP . Sleep disturbance in family caregivers 42 . Lipman RS . Depression scales derived from the Hopkins of individuals with dementia: a review of the literaturPe .r- Symptom Checklist. In: Sartorius N , Ban TA , edsA. ssessment spect Psychiatr Care . 2013 ; 49 ( 2 ): 135 - 146 . of Depression. Berlin, Heidelberg: Springer Berlin Heidelberg; 25. Kochar J , Fredman L , Stone KL , Cauley JA ; Study of Oste-o 1986 : 232 - 248 . doi: 10 .1007/978-3- 642 -70486-4_ 22 .
porotic Fractures . Sleep problems in elderly women care-giv 43 . Magnusson Hanson LL , Westerlund H , Leineweber CThe ers depend on the level of depressive symptoms: results of Symptom Checklist-core depression (SCL-CD6) scale: the caregiver-study of osteoporotic fractureJsA.m Geriatr psychometric properties of a brief six item scale for the Soc . 2007 ; 55 ( 12 ): 2003 - 2009 . assessment of depression . Scand J Public Health . 2014 ; 42 ( 1 ): 26 . von Känel R , Mausbach BT , Ancoli-Israel SSleep in spousal 82-88.
Alzheimer caregivers: a longitudinal study with a focus on 44 . Au N , Johnston DW . Self-assessed health: what does it the effects of major patient transitions on sleeSple . ep . 2012 ; mean and what does it hide? Soc Sci Med . 2014 ; 121 : 21 - 28 .
35 ( 2 ): 247 - 255 . doi: 10 .1016/j.socscimed. 2014 . 10 .007 27. Etcher L. Sleep disruption in older informal caregivers . 45. Basner M , Fomberstein KM , Razavi FMAmerican time use Home Healthc Nurse . 2014 ; 32 ( 7 ): 415 - 419 . survey: sleep time and its relationship to waking activities.
28. McCurry SM , Song Y , Martin JL . Sleep in caregivers: what we Sleep . 2007 ; 30 ( 9 ): 1085 - 1095 .
know and what we need to learnC . urr Opin Psychiatry . 2015 ; 46 . Laird NM , Ware JH . Random-effects models for longitudinal 28(6 ): 497 - 503 . data. Biometrics . 1982 ; 38 ( 4 ): 963 - 974 .
29. Melchior M , Berkman LF , Niedhammer I , Zins M , Goldberg 47. Maltby KF , Sanderson CR , Lobb EA , Phillips JL . Sleep dist-ur M. The mental health effects of multiple work and fa-m bances in caregivers of patients with advanced cancer: a ily demands. A prospective study of psychiatric sickness systematic review . Palliat Support Care . 2017 ; 15 ( 1 ): 125 - 140 .
absence in the French GAZEL study . Soc Psychiatry Psychiatr 48 . Maher J , Green H.Carers 2000 . London: National Statistics, Epidemiol. 2007 ; 42 ( 7 ): 573 - 582 . The Stationery Office; 2002 .
30. Razavi T , Clark C , Stansfeld SA . Work-family conflict as a 49 . Wilcox S , King AC . Sleep complaints in older women who predictor of common mental disorders in the 1958 British are family caregivers . J Gerontol B Psychol Sci Soc Sci . 1999 ; birth cohortL. ongitud Life Course Stud . 2015 ; 6 ( 3 ): 264 - 278 . 54( 3 ): P189 - P198 .
31. Magnusson Hanson LL , Theorell T , Oxenstierna G , Hyde 50 . Gibson R , Gander P , Alpass F , Stephens C . Effect of caregi-v M, Westerlund H. Demand, control and social climate as ing status on the sleep of older New ZealanderAs.ustralas J predictors of emotional exhaustion symptoms in working Ageing . 2015 ; 34 ( 3 ): 155 - 159 .
Swedish men and women . Scand J Public Health . 2008 ; 36 ( 7 ): 51 . Kumari M , Green R , Nazroo J . Sleep duration and sleep d-is 737-743. turbance . In:Financial Circumstances, Health and Well-Being of 32 . Nordin M , Åkerstedt T , Nordin S. Psychometric evaluation the Older Population in England: The 2008 English Longitudinal and normative data for the Karolinska Sleep Questionnaire . Study of Ageing (Wave 4) . London: The Institute for Fiscal Sleep Biol Rhythms . 2013 ; 11 ( 4 ): 216 - 226 . Studies; 2010 : 178 - 226 .
33. Akerstedt T , Knutsson A , Westerholm P , Theorell T , Alfre-ds 52. Simpson C , Carter P. The impact of living arrangements on son L, Kecklund G. Sleep disturbances, work stress and dementia caregivers' sleep qualityA .m J Alzheimers Dis Other work hours: a cross-sectional studyJ . Psychosom Res . 2002 ; Demen. 2015 ; 30 ( 4 ): 352 - 359 .
53 ( 3 ): 741 - 748 . 53 . Pearlin LI , Mullan JT , Semple SJ , Skaff MM . Caregiving and 34 . Hanson LL , Åkerstedt T , Näswall K , Leineweber C , Theorell the stress process: an overview of concepts and their mea-s T, Westerlund H . Cross-lagged relationships between wo-rk ures . Gerontologist . 1990 ; 30 ( 5 ): 583 - 594 .
54. DePasquale N , Zarit SH , Mogle J , Moen P , Hammer LB , 61 . Szebehely M , Ulmanen P , Sand A -BA. tt Ge Omsorg Mitt Almeida DM. Double- and triple-duty caregiving men an I Livet: Hur Påverkar Det Arbete Och Försörjning? [Providing Inexamination of subjective stress and perceived schedule formal Care in the Middle of Life: How Does It Affect Work and controlJ . Appl Gerontol . 2016 . doi: 10 .1177/0733464816641391. Livelihoods?]. Stockholm: Institutionen för socialt arbete; Epub ahead of print. 2014 .
55. Burr JA , Choi NG , Mutchler JE , Caro FG . Caregiving and v-ol 62 . Corey KL , McCurry MK . When caregiving ends: the exper-i unteering: are private and public helping behaviors linkeJd? ences of former family caregivers of people with dementia .
Gerontol B Psychol Sci Soc Sci . 2005 ; 60 ( 5 ): S247 - S256 . The Gerontologist. 2016 ; 00 ( 00 ): 1 - 10 . doi: 10 .1093/geront/ 56. Creese J , Bédard M , Brazil K , Chambers L. Sleep disturbances gnw20 in spousal caregivers of individuals with Alzheimer's d-is 63 . Laird NM . Missing data in longitudinal studieSs . tat Med .
ease. Int Psychogeriatr . 2008 ; 20 ( 1 ): 149 - 161 . 1988 ; 7 ( 1 -2): 305 - 315 .
57. Washington KT , Parker Oliver D , Smith JB , McCrae CS , Balch-an 64 . Carrière I , Bouyer J . Choosing marginal or random-effects dani SM , Demiris G . Sleep problems, anxiety, and global self- models for longitudinal binary responses: application to rated health among hospice family caregiverAsm. J Hosp Palliat self-reported disability among older personBsM . C Med Res Med . 2017 . doi: 10 .1177/1049909117703643. Epub ahead of print. Methodol . 2002 ; 2 ( 1 ): 15 .
58. von Känel R , Mausbach BT , Ancoli-Israel SPositive affect 65 . Hu FB , Goldberg J , Hedeker D , Flay BR , Pentz MA . Comparison and sleep in spousal Alzheimer caregivers: a longitudinal of population-averaged and subject-specific approaches for study . Behav Sleep Med . 2014 ; 12 ( 5 ): 358 - 372 . analyzing repeated binary outcomesA. m J Epidemiol . 1998 ; 59. McCrae CS , Dzierzewski JM , McNamara JP , Vatthauer KE , 147 ( 7 ): 694 - 703 .
Roth AJ , Rowe MA . Changes in sleep predict changes in 66 . Rostgaard T , Szebehely M. Changing policies, changing p-at affect in older caregivers of individuals with alzheimer's terns of care: Danish and Swedish home care at the cro-ss dementia: a multilevel model approachJ.Gerontol B Psychol roads . Eur J Ageing . 2012 ; 9 ( 2 ): 101 - 109 .
Sci Soc Sci . 2016 ; 71 ( 3 ): 458 - 462 . 67 . Plaisier I , Broese van Groenou MI , Keuzenkamp S. Com60. Arber S , Ginn J . Gender differences in informal caring. bining work and informal care: the importance of caring Health Soc Care Community . 1995 ; 3 ( 1 ): 19 - 31 . organisations. Hum Resour Manag J . 2015 ; 25 ( 2 ): 267 - 280 .