Associations Among Perceptions of Social Support, Negative Affect, and
Quality of Sleep in Caregivers and Noncaregivers
Beverly H. Brummett, Michael A. Babyak, and
Ilene C. Siegler
Duke University Medical Center
Peter P. Vitaliano
University of Washington
Edna L. Ballard, Lisa P. Gwyther, and Redford B. Williams
Duke University Medical Center
The authors used structural equation modeling to examine associations among perceptions of negative
affect, social support, and quality of sleep in a sample of caregivers (n ? 175) and noncaregiver control
participants (n ? 169). The authors hypothesized that caregiver status would be related to sleep quality
directly and also indirectly by way of negative affect and social support. This hypothesis was partially
supported in that caregiving was found to be indirectly related to sleep quality. However, after accounting
for the indirect effects of negative affect and social support, the direct effect of caregiving on sleep
quality was no longer statistically significant. The structural model accounted for approximately 43% of
the variance in sleep quality. The present findings may be useful in the development of successful sleep
interventions for caregivers.
Keywords: sleep, caregiving, depression, social support
Approximately four million individuals in the United States
suffer from dementia. The provision of care to these patients is a
significant societal burden that continues to grow as the population
ages. Family members who provide such care often sacrifice their
own health and well-being. A large body of research documents
the psychological impact of caregiving. For example, caregivers
have been shown to have higher ratings of negative affect, includ-
ing measures of depression, anxiety, and psychological distress
(for a review of this research, see Pinquart & Sorensen, 2003;
Russo, Vitaliano, Brewer, Katon, & Becker, 1995; Schulz,
O’Brien, Bookwala, & Fleissner, 1995).
The enormous amount of time that caregivers devote to caring
for a dementia patient is also likely to decrease the amount of time
available to caregivers for social interaction. Less time to spend
with others, coupled with the progressive loss of a loved one, may
result in worsened perceptions of social support (Bergman-Evans,
1994). This lack of social support is one factor that may contribute
to the negative psychological symptoms experienced by caregiv-
ers; several studies in caregiver samples have shown that lower
ratings of social support are associated with greater symptoms of
depression (e.g., Cannuscio et al., 2004; Schulz & Williamson,
Poorer social support is also associated with poorer sleep quality
among caregivers (Vitaliano et al., 2002); caregivers have been
shown to have poorer sleep quality than noncaregivers (Carter,
2002; Kiecolt-Glaser, Dura, Speicher, Trask, & Glaser, 1991; Sato,
Kanda, Anan, & Watanuki, 2002; Teel & Press, 1999). Under-
standing the sleep quality of caregivers is critical for a number of
reasons. Sleep is an extremely important aspect of well-being that
is related to quality of life (Zammit, Weiner, Damato, Sillup, &
McMillan, 1999) and health outcomes such as glucose regulation
(Spiegel, Leproult, & Van Cauter, 1999) and cortisol levels (Lep-
roult, Copinschi, Buxton, & Van Cauter, 1997; Spiegel et al.,
1999). Apart from the deleterious health consequences attributable
to lack of sleep, the ability to provide proper care to dementia
patients may be compromised when caregivers are not rested. In
addition, caregivers’ sleep problems contribute to the decisions to
place patients in nursing homes (Chenier, 1997; Vitiello & Borson,
2001). Considering the known disruption of sleep in caregivers and
the importance of sleep for this group, examinations of this health
measure in those who provide care have been surprisingly sparse.
In a study of 90 older women who were caregivers of a family
member with dementia, relationship to care recipient was unrelated
to sleep problems, but other factors related to emotion and psy-
chological distress were related to poor sleep quality (Wilcox &
King, 1999). A pilot study of changes in sleep and depression
ratings in caregivers found that both measures tended to fluctuate
somewhat from week to week and that both were affected by
factors such as caregiver anxiety (Carter, 2002). Another study
followed 80 spouse caregivers of persons with Alzheimer’s disease
and 85 demographically similar spouse noncaregivers, and re-
searchers observed that caregivers reported poorer sleep quality
than did noncaregivers at both study entry and 18 months later
Beverly A. Brummett, Michael A. Babyak, Ilene C. Siegler, Edna L.
Ballard, Lisa P. Gwyther, and Redford B. Williams, Department of Psy-
chiatry and Behavioral Sciences, Duke University Medical Center; Peter P.
Vitaliano, Department of Psychiatry and Behavioral Sciences, University
This research was supported by the National Institute on Aging Grant
R01AG19605, with cofunding by the National Institute of Environmental
Health Sciences; by the Clinical Research Unit Grant M01RR30; and by
the National Institute of Mental Health Grant R01MH57663.
Correspondence concerning this article should be addressed to Beverly
H. Brummett, Box 2969, Duke University Medical Center, Durham, NC
27710. E-mail: firstname.lastname@example.org
2006, Vol. 25, No. 2, 220–225
Copyright 2006 by the American Psychological Association
0278-6133/06/$12.00 DOI: 10.1037/0278-622.214.171.124
(Vitaliano et al., 1999). Moreover, sleep problems were related to
self-reported and clinical ratings of depression.
In the present study we used a structural equation model (SEM)
to examine the associations among negative affect, perceptions of
social support, and quality of sleep in a sample of caregivers and
noncaregiver participants. Apart from its inherent flexibility, SEM
offers a number of methodological advantages over more conven-
tional regression-type analysis. First, the variables under study can
be modeled using multiple indicators of each variable (when
available), resulting in more reliable renderings of the variables. In
the present study, for example, we modeled several of the variables
of interest as latent variables, which carry the shared, or reliable,
variance among the items that are used to create the construct.
Second, mediation hypotheses can be tested directly, using the
indirect effect test available in standard SEM software packages.
We based our hypothesis on (a) extensive research linking
caregiving with poor psychosocial outcomes (e.g., Brummett et al.,
2005; Pinquart & Sorensen, 2003; Schulz et al., 1995); (b) research
that shows negative affect is positively related to self-report mea-
sures of poor sleep in caregivers (Vitaliano et al., 1999) and (c)
studies that show lower ratings of social support are associated
with higher ratings of negative affect in caregivers (e.g., Cannuscio
et al., 2004; Schulz & Williamson, 1991). Looking at these find-
ings, we hypothesized that caregiving would be related to sleep
quality directly and indirectly by way of negative affect and social
support. Specifically, we expected that caregiver status would be
associated with greater negative affect and lower social support,
which would be associated in turn with decrements in quality of
sleep. In addition, because sleep disturbances in caregivers may be
caused by functional events that may act independently of negative
affect, such as being awakened by a patient in the middle of the
night, we also expected that caregiving would be related to poor
sleep quality over and above the effects of negative affect and
social support. Results of these analyses may help identify care-
givers who are especially vulnerable to sleep disturbances.
We recruited participants to be part of a study designed to examine the
underlying biological and behavioral mechanisms whereby stressful social
and physical environments (e.g., caregiving responsibilities or poor neigh-
borhood characteristics, respectively), lead to health disparities among
different socioeconomic groups. The full study sample consisted of 344
participants, with 175 adults who reported significant caregiving respon-
sibility for a relative (94.5% were caring for a parent) or a spouse diag-
nosed with dementia; and 169 control participants who did not have
caregiving responsibilities. Those in the caregiver group were older than
those in the control group by an average of five years. Within the caregiver
group, 84 (48%) were spouses and 91 (52%) were relatives.1There were
varying amounts of missing data, with the most missing cases (n ? 14) on
one of the sleep parameters.
The study was conducted at Duke University Medical Center. We
recruited caregivers with flyers, ads in the local media, and outreach efforts
conducted under the auspices of a Community Outreach and Education
Program. We recruited control participants by asking each caregiver to
nominate two to five friends who lived in their neighborhood and were like
them with respect to the key demographic factors (i.e., age, race, gender).
All individuals gave informed consent prior to their participation in the
study. Participants enrolled in the study received $250 for their participa-
tion. We excluded individuals who were experiencing any acute major
medical or psychiatric disorders. We collected data in two venues—during
a home visit by a nurse and during a visit to the General Clinical Research
Center (GCRC) at Duke University Medical Center. A questionnaire bat-
tery containing the information used in the present study was given to
participants during the home visit and returned upon their visit to the
GCRC (scheduled within the week), when study personnel went over the
questionnaires to check with the participant regarding any queries or
each coded as dichotomous variables. Age was measured in years.
The Pittsburgh Sleep Quality Index (Buysse, Reynolds, Monk,
Berman, & Kupfer, 1989) was used to measure sleep quality. The scale
consists of 19 items that assess various aspects of sleep during the past
month (e.g., time needed to fall asleep, aspects related to sleep disruption).
The items on the scale are used to form the following subscales: overall
sleep quality, sleep latency (time needed to fall asleep upon going to bed),
sleep duration, sleep efficiency (100 times the result of the following: time
asleep divided by time in bed), sleep disturbances, problems with daytime
functioning, and medications taken for sleep. Typically, the seven sub-
scales may be summed to provide a global score, with higher scores
reflecting poor sleep quality. A score of 5 or higher on the global rating
suggests moderate sleep problems in three or more areas, or more severe
problems in at least two areas. Acceptable psychometric properties have
been demonstrated with respect to internal homogeneity, test–retest reli-
ability, and validity (Buysse et al., 1989). We also used six of the seven
subscales to create a latent variable in the SEM.
The Interpersonal Support Evaluation List (ISEL; Co-
hen, Mermelstein, Kamarck, & Hoberman, 1985) was used to assess
perceptions of social support. The ISEL consists of 40 items that assess the
following dimensions of support: appraisal, self-esteem, belonging, and
tangible. We used a shortened 16-item version of the ISEL that retains the
four subscales (Brummett et al., 1998, 2004). Items were rated on a 4-point
scale, with a potential range of 0–48 for total ISEL scores; higher scores
reflect greater perceived support. We used these four subscales as indicator
variables of a social support latent variable in the SEM.
Negative affect was assessed with four scales that
measure symptoms of depression, hopelessness, perceptions of stress, and
state-trait anxiety. The Center for Epidemiologic Studies Depression Scale
(CES-D) (Radloff, 1977) is a 20-item self-report scale designed to measure
depressive symptomatology that has been frequently used in studies of
caregivers (Schulz et al., 1995). The questions refer to symptoms experi-
enced during the week prior to the interview, capturing affective, somatic,
well-being, and interpersonal symptomatology. Measures of internal con-
sistency for the CES-D are acceptable, with alpha coefficients of .85 in a
general population and .90 in a patient sample. Test-retest correlations
range between .45 and .70 (Radloff, 1977). We supplemented the CES-D
with the following four items that measure feelings of hopelessness:
“During the past month” (a) “Have you felt so sad, discouraged, hopeless
or had so many problems that you wondered if anything was worthwhile?”
(b) “Have you felt downhearted and blue?”(c) “How have you been feeling
in general?” and (d) “How depressed or cheerful have you been?”. Feelings
of hopelessness have been shown to be especially potent predictors of
Group status (caregiver or control) and gender were
1Spouse caregivers did not differ significantly from nonspouse-relative
caregivers on any of the measures in the present study (e.g., symptoms of
depression, trait anxiety, hopelessness, perceptions of stress, social support,
nor sleep quality).
QUALITY OF SLEEP IN CAREGIVERS
health outcomes that are independent of other aspects of depression (Anda
et al., 1993). The Perceived-Stress Scale (Cohen & Williamson, 1988)
consists of 10 items that are evaluated on a 5-point Likert scale. Items tap
the degree to which individuals feel that events in their lives are unpre-
dictable and uncontrollable. Comparisons of the 10-item version with the
original 14-item version of the scale have revealed that the shorter version
is psychometrically superior. Cronbach’s alpha reliability was .78 in a
national survey. Finally, the Spielberger Trait Anxiety Scale (STAI-Trait)
(Spielberger, 1983) was used to assess chronic (trait) levels of anxiety. This
scale is a standard measure with acceptable demonstrated reliability, for
example, trait anxiety retest reliability has ranged from r ? .73 to .86 and
state internal consistency has ranged from ? ? .83 to .92. Higher scores on
each of the 4 measures reflect greater negative affect. As in the case of the
previous measures, we used the four scales described above as indicators
of a broader latent construct labeled Negative Affect.
We used structural equation modeling to evaluate relationships among
caregiver status, social support, negative affect, and sleep quality. We
represented caregiver status as a binary (yes–no) variable, whereas social
support, negative affect, and sleep status were operationalized as latent
variables. Latent variables represent the shared variance among a set of
indicator variables that are thought to capture information about a common
construct. Because the latent variable contains only the shared variance of
its indicators, much of the method-related error variance from an individual
instrument is removed from the construct, thereby improving its reliability.
We used the CES-D, Spielberger Trait Anxiety Scale Perceived Stress
Scale, and hopelessness scores as indicator variables of the negative affect
latent variable, and we used the four ISEL subscales as the indicators for
social support. For the sleep quality latent variable, we used the following
Pittsburgh Sleep Quality Index subscales as indicators: sleep quality, sleep
latency, sleep duration, sleep disturbances, daytime dysfunction, and sleep
medications. We did not use the sleep efficiency (100 times the result of the
following: time asleep divided by time in bed) subscale, owing to a very
low correlation (factor loading) with the latent construct. In addition to the
above variables, we also included age and gender as adjustment variables
with respect to negative affect, social support, and the sleep outcome. The
structure of the model is shown in Figure 1. Each arrow or series of arrows
represents a hypothesis about the relation among the variables in the model.
In this case, we hypothesized that caregiver status would be related to Sleep
Quality directly, and also indirectly by way of its relation to Negative
Affect and Social Support.
We estimated the model with Muthe ´n’s M-Plus software (Muthe ´n &
Muthe ´n, 2004), using the maximum likelihood minimization and the raw
data input option. We managed missing data with the full-information
maximum likelihood method available in M-Plus. We evaluated fit using
the Comparative Fit Index (CFI), the Tucker-Lewis Index (TLI), the
chi-square values, and the root mean square error of approximation
(RMSEA). For the primary analysis, we estimated only the full a priori
model, making no ad hoc changes to improve fit (Green & Babyak, 1997).
We also conducted ancillary analyses that examined the relation between
caregiving and Sleep Quality, Negative Affect and Sleep Quality, and
Social Support and Sleep Quality, each in a separate model. These latter
analyses yield an estimate of the relation between each of the three
variables and sleep quality, adjusted only for age and gender.
Table 1 presents the characteristics of the sample by group
status. Compared with controls, caregivers exhibited significantly
higher scores on all measures of negative affect, poorer perceived
social support, and poorer sleep quality.
Table 2 shows the factor loadings from the latent variables that
are included in the structural model. These loadings can be inter-
preted in a manner similar to conventional common factor analy-
sis. The loadings for the indicators of the Negative Affect and
Social Support variable are all relatively high, indicating a fair
amount of unidimensionality and shared variance among the indi-
cators on their respective latent variables. The indicators for the
Sleep Quality are more variable in their magnitude but still within
an acceptable range. It should also be borne in mind, however, that
because the latent construct carries only the shared variance among
its indicators, low loadings do not impinge on the reliability of the
latent variable—the contribution to the latent variable is only what
it shares with the rest of the indicators.
The overall fit of the model was modest (chi-square [107 df] ?
284, p ? .001; CFI ? .931, TLI? 915, RMSEA ? .069), indi-
cating that there were parameters that might be freed that might
improve the fit. Ancillary analysis in which parameters (cross
loadings and/or correlated errors) were freed improved the fit
considerably but made very little difference in the path coefficient
estimates, which are of primary concern in the present study.
We modeled age and gender as predictors of all latent con-
structs, meaning that all structural (path) coefficients pointing
toward the latent variables were adjusted for gender and age. The
path model (see Figure 1) shows that caregiver status is signifi-
cantly related to both Social Support (p ? .001) and Negative
Affect (p ? .001), and that Negative Affect is strongly and
significantly related to Sleep Quality (p ? .001). Social Support,
age, and gender were not directly related to Sleep Quality. Care-
giver status also was not significantly related to Sleep Quality in
this full SEM.
Structural equation modeling also affords a convenient means of
calculating indirect effects, that is, the relation between a predictor
and outcome by way of one or more intervening, or mediating,
variables. A statistically significant indirect effect with a nontrivial
effect size can be interpreted as support for the hypothesis that the
intervening variable mediates the relation between a predictor and
outcome. The present model shows three indirect effects. The first,
Caregiver Status 3 Negative Affect 3 Sleep Quality, was statis-
tically significant (standardized structural coefficient ? .21, p ?
.001). The indirect effect of Caregiver Status 3 Social Support 3
Negative Affect 3 Sleep Quality also was significant (standard-
ized structural coefficient ? .12, p ? .001). Finally, the Caregiver
Status 3 Social Support 3 Sleep Quality indirect effect was not
statistically significant (standardized structural coefficient ?
?.02, p ? .342). The relation of caregiving status to Sleep Quality,
significant (p ? .05) and values in italics are not statistically significant.
Results of structural model. Values in bold are statistically
BRUMMETT ET AL.
when adjusted for only age and gender (and not Negative Affect
and Social Support), was considerably larger and statistically sig-
nificant (standardized structural coefficient ? .34, p ? .001).
Taken together, the model suggests that the data are consistent
with the possibility that the relation between caregiving and Sleep
Quality may be mediated by Negative Affect and Social Support.
Total variance in sleep accounted for by model is ? 42.6%.
The present findings replicate an extensive body of research that
indicates that caregiving is related to higher ratings of negative
affect and worsened perceptions of social support. In addition,
current univariate and multivariate analyses show that sleep qual-
ity in the present sample of caregivers is compromised, as com-
pared with that of noncaregivers. Most important, our findings
indicate that a large percentage of the variance in sleep quality may
be attributable to associations among caregiving, negative affect,
and social support. Specifically, the current findings suggest that
negative affect mediates the association between caregiving and
poor sleep quality. However, the inverse association between
negative affect and social support may also explain part of the
mediating role of negative affect with respect to caregiving and
Multiple causal pathways come to mind as we consider the sleep
problems reported by caregivers. Some of these pathways are
directly attributable to the emotional distress suffered by caregiv-
ers, while others may seem primarily functional in nature. For
example, caregivers may be frequently awakened by dementia
patients who themselves suffer from nocturnally disrupted sleep
(Bliwise, 2004). Family caregivers may also sleep lightly because
of the feeling that they are on duty both day and night, listening for
calls for help. Our findings suggest, however, that such functional
disruptions may ultimately lead to poor sleep quality due to their
ability to produce negative affect. Specifically, when the effects of
negative affect and perceptions of support were accounted for, the
direct association between caregiving and sleep quality was not
Given the evidence supporting poor sleep quality in caregivers,
recent research has examined interventions aimed at addressing
this problem. Related research examining group cognitive–
behavioral therapy as a means of reducing anxiety in caregivers
found that individuals assigned to the therapy condition demon-
strated less anxiety as compared to wait-listed participants and that
a subsample of caregivers also demonstrated improvements in
sleep (Akkerman & Ostwald, 2004). McCurry, Logsdon, Vitiello,
and Teri (1998) have shown that a behavioral intervention con-
sisting of standard sleep strategies, community resource education,
stress management, and methods to limit disruptive behavior in
patients is associated with improvement in quality of sleep. The
above studies point toward viable behavioral techniques for im-
proving sleep quality in caregivers; the present results suggest that
interventions focused on reductions in negative affect are likely to
be beneficial with respect to improvements in sleep quality.
We note that a number of plausible alternative models could be
proposed that would fit the data equally well, especially given the
cross-sectional study design. Indeed, by mathematical definition,
the direction of path coefficients could be varied in a number of
ways and still achieve the same overall model fit (Lee & Hersh-
Sample Characteristics by Group Status
(n ? 175)
(n ? 169)
pM SDM SD
Symptoms of depression (total CES-D)
Anxiety (total STAI-Trait)
Perceived stress (total PSS)
Social support (total ISEL)
PSQI global score
Noncaregivers were male. CES-D ? Center for Epidemiologic Studies Depression Scale; STAI ? State–Trait
Anxiety Inventory; PSS ? Perceived Stress Scale; ISEL ? Interpersonal Support Evaluation List; PSQI ?
Pittsburg Sleep Quality Index.
Sample size may vary slightly by measure. 45 (25.7%) of the caregivers were male; 41 (24.3%) of the
Latent Construct Factor Loadings
Poor sleep quality
Overall sleep quality rating
Use of sleep medications
Symptoms of depression
STAI ? State–Trait Anxiety Inventory.
QUALITY OF SLEEP IN CAREGIVERS
berger, 1990). For example, the well-known relationship between
sleep and depressive symptoms suggests an obvious alternative
model that would simply reverse the order of sleep quality and
negative affect. This model would have fit the data exactly as well
as the model we present, but would be subjected to the same
criticism. Ideally, we would have preferred to model a bidirec-
tional path between Negative Affect and Sleep Quality, but we
were not able to do so for technical reasons (the model did not
converge). Inasmuch as we were constrained to choose one model
over the other for estimation, we maintained our hypothesized
model on the assumption that the causal influence of negative
affect on sleep might be somewhat stronger than the opposite
configuration (see Davis, 1985). Related to the above, the fit of the
present model was only modest. As we mention in the Results
section, we were able to improve the fit of the model considerably
by freeing parameters among the latent variables and indicator
variables. However, in the present study, we are far less concerned
with the fit of the model than we are with the path estimates.
Moreover, post hoc modifications are subject to the same problem
of multiple testing (Green & Babyak, 1997), and these changes did
not materially alter the path coefficients.
According to a 1998 report from the National Family Caregivers
Association (as cited by the National Family Caregivers Associa-
tion, 2005), approximately half of those who provide high levels of
caregiving report that they suffer from sleeplessness. Diminished
sleep quality is clinically meaningful, as it not only detracts from
general well-being but may also lead to additional health problems
over time. Indeed, our previous work has shown that poor sleep
quality is associated with lower levels of natural killer cell activity
(the ability of the body to fight tumors) in caregivers but not in
noncaregivers (Irwin et al., 1994; Vitaliano et al., 1999). Caregiv-
ing also predicted a distress composite (including poor sleep
quality and negative affect), which predicted poor health habits
(i.e., diet, exercise) and greater physiological dysregulation (ele-
vated insulin and glucose) (Vitaliano et al., 2002). Hence, sleep
and negative affect together mediated the relationship between
caregiving and physiological risk. This latter work is consistent
with studies in general populations that show relationships be-
tween poor sleep and risk for glucose dysregulation (Spiegel et al.,
1999). It has also been shown that caregivers have poorer memo-
ries than do noncaregivers and that this is mediated by poor sleep
quality (Caswell et al., 2003). Caregiver cognitive status is critical
because it may influence poorer self-care in caregivers and poorer
care of patients. Finally, a meta-analysis of caregivers of persons
with dementia and demographically similar noncaregivers (Vi-
taliano, Zhang, & Scanlan, 2003) found that caregivers had a 23%
higher level of stress hormones (ACTH, catecholamines, cortisol,
etc.) and a 15% lower level of antibodies (e.g., poorer IgG re-
sponses to vaccinations and high levels of herpes simplex antibod-
ies in absence of a challenge, etc.) than did noncaregivers. The
current work is relevant to this meta-analysis for two reasons.
First, poor sleep quality is associated with elevated stress hor-
mones, and chronically elevated epinephrine and cortisol may lead
to persistent elevations in blood pressure and glucose and in-
creased risk for hypertension and diabetes (Chrousos & Gold,
1992; Lovallo, 1997; Lovallo & Thomas, 2000). Second, poor
sleep quality is associated with compromised immunity (Irwin et
al., 1994), and a lower antibody production may increase a care-
giver’s risk for influenza.
Several possible limitations should be noted with respect to the
present findings. First, the present study used self-report measures.
However, the affect and social support measures have been thor-
oughly examined over the years for validity and reliability; they
have been widely accepted as reasonable measures of these con-
structs. The issue of self-reported sleep quality is a more complex
matter. Our findings are consistent with Wells, Day, Carney,
Freedland, and Duntley (2004), who showed that depression was
related to a self-report measure of sleep quality. The same study
found that depression was not as strongly related to polysomno-
graphic assessment. Taken together with other evidence that self-
reported sleep rating correlates only modestly with objective sleep
measurement in both healthy (Baker, Maloney, & Driver, 1999)
and clinical populations (Weiss, McPartland, & Kupfer, 1973),
inferences from the present study therefore may pertain only with
respect to self-report ratings of sleep quality. However, Wells et al.
(2004) also showed that self-reported sleep quality was strongly
correlated with oxygen desaturation. In addition, Hall et al. (2000)
posit that self-reported sleep quality may be related to hyper-
arousal during sleep, resulting in one’s perception of wakefulness
during the night despite being asleep. Therefore, beyond the im-
portance of subjective well-being per se, self-reported sleep quality
in our sample may have some importance with respect to physio-
logical correlates. It is also important to note again that the present
data are cross-sectional, limiting the ability to draw causal infer-
ence. A final limitation concerns the fact that we did not model
certain medical constructs that might have influenced the findings
(e.g., obesity, sleep apnea).
Given the poor sleep quality reported in informal caregivers and
the increasingly important societal role that they play, it is critical
to better understand the nature of their sleep problems. It is hoped
that practitioners can use the present findings to continue to
develop successful sleep interventions. To that end, our results
indicate that sleep interventions in those who care for dementia
patients should focus on the reduction of negative affect and that
interventions in caregivers aimed at increasing social support may
be effective in improving sleep quality only to the extent that they
decrease negative affect.
Akkerman, R. L., & Ostwald, S. K. (2004). Reducing anxiety in Alzhei-
mer’s disease family caregivers: The effectiveness of a nine-week
cognitive-behavioral intervention. American Journal of Alzheimer’s Dis-
ease & Other Dementias, 19, 117–123.
Anda, R., Williamson, D., Jones, D., Macera, C., Eaker, E., Glassman, A.,
& Marks, J. (1993). Depressed affect, hopelessness, and the risk of
ischemic heart disease in a cohort of U.S. adults. Epidemiology, 4,
Baker, F. C., Maloney, S., & Driver, H. S. (1999). A comparison of
subjective estimates of sleep with objective polysomnographic data in
healthy men and women. Journal of Psychosomatic Research, 47, 335–
Bergman-Evans, B. F. (1994). Alzheimer’s and related disorders: Loneli-
ness, depression, and social support of spousal caregivers. Journal of
Gerontology and Nursing, 20, 6–16.
Bliwise, D. L. (2004). Sleep disorders in Alzheimer’s disease and other
dementias. Clinical Cornerstone, 6(Suppl. 1A), S16–28.
Brummett, B. H., Babyak, M. A., Barefoot, J. C., Bosworth, H. B.,
Clapp-Channing, N. E., Siegler, I. C., et al. (1998). Social support and
hostility as predictors of depressive symptoms in cardiac patients one
BRUMMETT ET AL.
month following hospitalization: A prospective study. Psychosomatic
Medicine, 60(6), 707–713.
Brummett, B. H., Babyak, M. A., Mark, D. B., Clapp-Channing, N. E.,
Siegler, I. C., & Barefoot, J. C. (2004). Prospective study of perceived
stress in cardiac patients. Annals of Behavioral Medicine, 27, 22–30.
Brummett, B. H., Siegler, I. C., Rohe, W. M., Barefoot, J. C., Vitaliano,
P. P., Surwit, R. S., et al. (2005). Neighborhood characteristics moderate
effects of caregiving on glucose metabolism. Psychosomatic Medicine,
Buysse, D. J., Reynolds, C. F., Monk, T. H., Berman, S. R., & Kupfer, D. J.
(1989). The Pittsburgh Sleep Quality Index: A new instrument for
psychiatric practice and research. Psychiatric Research, 28, 193–213.
Cannuscio, C. C., Colditz, G. A., Rimm, E. B., Berkman, L. F., Jones,
C. P., & Kawachi, I. (2004). Employment status, social ties, and care-
givers’ mental health. Social Science and Medicine, 58, 1247–1256.
Carter, P. A. (2002). Caregivers’ descriptions of sleep changes and depres-
sive symptoms. Oncology Nursing Forum, 29, 1277–1283.
Caswell, L., Vitaliano, P. P., Croyle, K., Scanlan, J. M., Zhang, J., &
Daruwala, A. (2003). Negative associations of chronic stress and cog-
nitive functioning in older adult spouse caregivers. Experimental Aging
Research, 29, 303–318.
Chenier, M. C. (1997). Review and analysis of caregiver burden and
nursing home placement. Geriatric Nursing, 18, 121–126.
Chrousos, G. P., & Gold, P. W. (1992). The concepts of stress and stress
system disorders. Journal of American Medical Association, 267, 1244–
Cohen, S., Mermelstein, R., Kamarck, T., & Hoberman, H. (1985). Mea-
suring the functional components of social support. In I. G. Sarason &
B. R. Sarason (Eds.), Social support: Theory, research and application
(pp. 73–94). The Hague, Holland: Martinus Nijhoff.
Cohen, S., & Williamson, G. M. (1988). Perceived stress in a probability
sample of the United States. In S. Spacapan & S. Oskamp (Eds.), The
social psychology of health: The Claremont Symposium on Applied
Social Psychology (pp. 31–67). Newbury Park, CA: Sage.
Davis, J. A. (1985). The logic of causal order. Thousand Oaks, CA: Sage.
Green, S. B., & Babyak, M. A. (1997). Control of type I errors with
multiple tests of constraints in structural equation modeling. Multivari-
ate Behavioral Research, 32, 39–51.
Hall, M., Buysse, D. J., Nowell, P. D., Nofzinger, E. A., Houck, P.,
Reynolds, C. F., et al. (2000). Symptoms of stress and depression as
correlates of sleep in primary insomnia. Psychosomatic Medicine, 62,
Irwin, M., Mascovich, A., Gillin, J. C., Willoughby, R., Pike, J., & T. L.,
S. (1994). Partial sleep deprivation reduces natural killer cell activity in
humans. Psychosomatic Medicine, 56, 493–498.
Kiecolt-Glaser, J. K., Dura, J. R., Speicher, C. E., Trask, J., & Glaser, R.
(1991). Spousal caregivers of dementia victims: Longitudinal changes in
immunity and health. Psychosomatic Medicine, 53, 345–362.
Lee, S., & Hershberger, S. (1990). A simple rule for generating equivalent
models in structural equation modeling. Multivariate Behavioral Re-
search, 25, 313–334.
Leproult, R., Copinschi, G., Buxton, O., & Van Cauter, E. (1997). Sleep
loss results in an elevation of cortisol levels the next evening. Sleep, 20,
Lovallo, W. R. (1997). Stress & Health. Thousand Oaks, CA: Sage.
Lovallo, W. R., & Thomas, T. L. (2000). Stress hormones in psychophys-
iological research: Emotional, behavioral, and cognitive implications. In
J. T. Cacioppo, L. G. Tassinary, & G. Berntson (Eds.), Handbook of
psychophysiology (pp. 342–367). New York: Cambridge University
McCurry, S. M., Logsdon, R. G., Vitiello, M. V., & Teri, L. (1998).
Successful behavioral treatment for reported sleep problems in elderly
caregivers of dementia patients: A controlled study. Journals of Geron-
tology Series B: Psychological Sciences and Social Sciences, 53, 122–
Muthe ´n, L. K., & Muthe ´n, B. O. (2004). Mplus user’s guide. (3rd ed.). Los
National Family Caregivers Association. (2005). Caregiving statistics.
Retrieved November 4, 2004, from http://www.thefamilycaregiver.org/
Pinquart, M., & Sorensen, S. (2003). Differences between caregivers and
noncaregivers in psychological health and physical health: A meta-
analysis. Psychology and Aging, 18, 250–267.
Radloff, L. S. (1977). The CES-D scale: A self report depression scale for
research in the general population. Applied Psychological Measurement,
Russo, J., Vitaliano, P. P., Brewer, D., Katon, W., & Becker, J. (1995).
Psychiatric disorders in spouse caregivers of care-recipients with Alz-
heimer’s disease and matched controls: A Diathesis-Stress model of
psychopathology. Journal of Abnormal Psychology, 104, 197–204.
Sato, R., Kanda, K., Anan, M., & Watanuki, S. (2002). Sleep EEG patterns
and fatigue of middle-aged and older female family caregivers providing
routine nighttime care for elderly persons at home. Perceptual & Motor
Skills, 95, 815–829.
Schulz, R., O’Brien, A. T., Bookwala, J., & Fleissner, K. (1995). Psychi-
atric and physical morbidity effects of dementia caregiving: Prevalence,
correlates, and causes. The Gerontologist, 35, 771–791.
Schulz, R., & Williamson, G. M. (1991). A 2-year longitudinal study of
depression among Alzheimer’s caregivers. Psychology and Aging, 6,
Spiegel, K., Leproult, R., & Van Cauter, E. (1999). Impact of sleep debt on
metabolic and endocrine function. Lancet, 354, 1435–1439.
Spielberger, C. D. (1983). Manual for the State-Trait Anxiety Inventory.
Palo Alto, CA: Consulting and Psychologists Press.
Teel, C. S., & Press, A. N. (1999). Fatigue among elders in caregiving and
noncaregiving roles. Western Journal of Nursing Research, 21, 498–
Vitaliano, P. P., Scanlan, J., Moe, K., Siegler, I. C., Prinz, P. N., & Ochs,
H. D. (1999). Stress, sleep problems, and immune function in persons
with cancer histories. Cancer Research Therapy and Control, 10, 167–
Vitaliano, P. P., Scanlan, J. M., Zhang, J., Savage, M. V., Hirsch, I. B., &
Siegler, I. C. (2002). A path model of chronic stress, the metabolic
syndrome, and coronary heart disease. Psychosomatic Medicine, 64,
Vitaliano, P. P., Zhang, J., & Scanlan, J. M. (2003). Is caregiving hazard-
ous to one’s physical health? A meta-analysis. Psychological Bulletin,
Vitiello, M. V., & Borson, S. (2001). Sleep disturbances in patients with
Alzheimer’s disease: Epidemiology, pathophysiology and treatment.
CNS Drugs, 15, 777–796.
Weiss, B. L., McPartland, R. J., & Kupfer, D. J. (1973). Once more: The
inaccuracy of non-EEG estimation of sleep. American Journal of Psy-
chiatry, 11, 1282–1285.
Wells, R. D., Day, R. C., Carney, R. M., Freedland, K. E., & Duntley, S. P.
(2004). Depression predicts self-reported sleep quality in patients with
obstructive sleep apnea. Psychosomatic Medicine, 66, 692–697.
Wilcox, S., & King, A. C. (1999). Sleep complaints in older women who
are family caregivers. Journals of Gerontology Series B: Psychological
Sciences and Social Sciences, 54, 189–198.
Zammit, G. K., Weiner, J., Damato, N., Sillup, G., & McMillan, C. (1999).
Quality of life in people with insomnia. Sleep, 22, 379–385.
QUALITY OF SLEEP IN CAREGIVERS