Physical Disability Contributes to Caregiver Stress
in Dementia Caregivers
David G. Bruce,1Glenys A. Paley,1Pamela Nichols,1David Roberts,2
Peter J. Underwood,3and Frank Schaper4
1School of Medicine and Pharmacology, University of Western Australia, Perth.
2School of Nursing, Edith Cowan University, Perth.
3Department of Environmental Science, Murdoch University, Perth.
4Alzheimer’s Association of Western Australia, Perth.
Backgound. Previous findings of studies on the impact of physical illness on caregiver health have been inconsistent.
The authors wanted to determine whether physical disability, as determined by the SF-12 survey that provides information
on both physical and mental health problems, contributes to caregiver stress.
Methods. The authors interviewed 91 primary caregivers (aged 38–85 years) of persons with dementia who had been
referred by their family physicians for the first time for formal support services or memory evaluation. Caregivers
completed the SF-12 version of the Medical Outcomes Study Short Form Health Survey that generates Mental Component
Summary (MCS) and Physical Component Summary (PCS) scores and reported on caregiver stress and concurrent
medical conditions and medications.
Results. Most caregivers reported stress (76.9%), having medical conditions (72.4%), or taking medications (67%). The
MCS but not the PCS scores were significantly lower than community norms, indicating an excess of disability due to
mental health problems. Nevertheless, 40.7% had PCS scores indicating some degree of physical disability. Using
multiple logistic regression analysis, PCS scores but not the presence of medical problems were independently associated
with caregiver stress.
Conclusions. Chronic disability as assessed by SF-12 PCS scores is independently associated with caregiver stress.
These data suggest that caregivers of persons with dementia should be assessed for disabling physical conditions and
mental health problems. In addition, reducing the impact of physical disability could ameliorate caregiver stress.
particularly in relation to the stress and strain that results
from caring for a person with dementia (1). Although the
impact of caregiving varies, caregivers of persons with
dementia are at increased risk for suffering from mental
health disorders including anxiety and depression (2,3).
Several studies indicate that the stress of caregiving may
also be associated with physical health effects including
poorer self-rated overall health (3) and decreased immunity
(4), but most studies have found little or no difference in
physical illnesses and chronic conditions, and a review of
the literature concluded that the evidence linking caregiving
with physical morbidity was weak (3). Recently, a large
community-based study found that elderly caregivers who
reported emotional strain had a high mortality rate compared
with similarly aged persons who were not caregivers (5).
Because physical health and mental health are closely
linked, research has suggested that caregiver studies should
include measures of both types of outcomes (3). To this end,
we included the 12-item version (SF-12) of the Medical
Outcomes Study Short Form Health Survey as part of our
assessment of caregivers of persons with dementia in an
ongoing health promotion study aimed at improving demen-
tia care. The SF-12 generates a mental component summary
(MCS) score and a physical component summary (PCS)
score that provide a broad estimate of disability resulting
HE potential for mental health problems in family
caregivers of patients with dementia is well recognized,
from mental or physical health problems (6). The SF-12 has
been widely used in studies of physical and psychiatric
disorders, and local normative data are available. In this
investigation, we wanted to determine whether physical
health problems contribute to caregiver stress.
Caregivers of persons with dementia were identified and
recruited for a research interview using methods developed
in previous studies aimed at identifying successive patients
with dementia and caregivers referred for the first time to
local geriatric services (7,8). In our health care system, this is
virtually always the first time these families have interacted
with the formal community support system (7,8). Between
April and December 2001, we invited 145 caregivers to be
interviewed at their homes. Of these, 92 agreed and were
interviewed. Reasons for refusing an interview included
caregiver illness and the caregiver being too stressed or too
busy. One person refused to complete the SF-12, so this
report includes 91 caregivers representing 62.8% of the
eligible caregivers identified. All had been caring for
relatives with verified dementia for at least 6 months and
all identified themselves as the primary caregiver. The Fre-
mantle Hospital Ethics Committee approved the protocol,
and all participants signed a consent form.
Journal of Gerontology: MEDICAL SCIENCES
2005, Vol. 60A, No. 1, 000–000
Copyright 2005 by The Gerontological Society of America
Proof only. Not for distribution.
The SF-12 provides a generic measure of health status
and disability because it addresses limitations resulting from
physical or mental health status. The SF-12 has excellent
psychometric characteristics and is recommended for use as
an outcome measure for mental health problems (9). The
SF-12 has not been used previously in caregiver studies,
although the longer version (SF-36) has been (10). The PCS
and MCS scores are derived from eight measured con-
cepts (physical functioning, role limitations due to physical
health problems, bodily pain and general health for the PCS
and vitality, social functioning, role limitations resulting
from emotional problems, and psychological distress for
The research interview included questions aboutthecaring
role such as the duration and extent of caregiving, the
caregiver’s knowledge and attitudes about dementia, and
how to access services. Demographic data included the care-
giver’s relationship to the persons with dementia, whether
household income. Specific questions about caregiver health
included: (a) whether they had a health problem (yes/no) and
the type of health problem; (b) whether they were taking
medications for a health problems (yes/no) and for how long;
and (c) whether they were taking medications for an emo-
tional problem or nervous condition (yes/no) and for how
long. They were also asked whether they were feeling
stressed, to which they could answer yes/no/not sure. The
research assistant could clarify that stress could mean feeling
anxious,overwhelmed,distressed, agitated, exhausted,under
pressure, downhearted, or feeling low.
Data Analysis and Statistics
The MCS and PCS scores were calculated using weight-
ings from the United States that are appropriate in
the Australian context (11) despite previous debate (12).
The question on stress was dichotomized by combining the
‘‘yes’’ and ‘‘not sure’’ responses (7 participants responded
‘‘not sure’’). Educational level was assessed by years of
schooling, and responses were dichotomized to less than 10
years schooling or not. Income was assessed by annual total
family income, and responses were dichotomized to less than
$A20,000 or not. The MCS and PCS scores were compared
with normative data supplied by the Western Australian
Health and Well-being Surveillance System, Department of
Health, Western Australia. This population-based survey of
more than 10,000 community-living adults aged 18 years
and older was administered by telephone and included the
SF-12 survey (13). Because the age ranges of the caregivers
were bimodal (Figure 1), with a median age of close to 65
years, comparative data were obtained for comparable age
ranges (30–65 years and 65–85 years).
Statistical analyses were performed using SPSS for
Windows (SPSS, Chicago, IL). The PCS and MCS scores
are presented as the mean 6 SD or by 95% confidence
intervals. Two-sample comparisons for normally distributed
variables and for MCS and PCS scores were by the Student t
test and for non-normally distributed variables by Mann–
Whitney U tests. Comparisons of proportions were made
using the Pearson chi-squared test, and correlations were
made using the Spearman rho test. A significance level of
p , .05 was used. Stepwise multiple logistic regression was
performed to investigate independent factors associated with
the MCS and PCS scores, with variable selection entry of
p , .05 and removal at p . .10. Multiple logistic regression
analyses using the forward conditional method explored
factors associated with caregiver stress, and entry and
removal criteria were p , .05 and p . .10, respectively.
Description of the Sample
Caregiver ages ranged from 38 to 85 years, and the
distribution of ages was bimodal (Figure 1), with the mid-
point of the 2 age distributions close to the median age of 62
years. Most were female (71.4%), and 41 were spouses.
Nonspouse caregivers included 36 daughters, 10 sons, 1
daughter-in-law, and 3 friends. As expected, child care-
givers were significantly younger than spouse (and friend)
caregivers (median age, 52.4 vs 76 years; p , .001), al-
though there was overlap (age range for spouses and friends,
49–85 years; age range for children, 38–68 years). Fifty-
seven caregivers (62.6%) lived with the dementia sufferer,
and this group included all spouses and 34% of child care-
givers. Consequently, live-in caregivers were significantly
older than non–live-in caregivers (median age, 72 vs 51
years; p , .001). Table 1 includes data on the duration of
care and the type and degree of assistance required.
Although the duration of symptoms and need for care was
lengthy (more than 1 year for the majority), in 79% the
referrals from family physicians included a request for an
assessment of possible dementia. The majority of the per-
sons with dementia were independent in performing basic
activities of daily living but required assistance with cook-
ing, shopping, managing finances, and medications.
Figure 1. Bimodal distribution of ages in 91 dementia caregivers.
SF-12 IN DEMENTIA CAREGIVERS
Physical and Mental Health of the Caregivers
SF-12 MCS scores.—The mean MCS scores for the
caregivers were significantly lower than the state population
norms as demonstrated by comparison of confidence inter-
vals (Table 2). In terms of level of mental health–related
disability assessed from MCS scores using the method of
Sanderson and Andrews (14), 33% had mild disability
(MCS scores between 40 and 49), 12.1% had moderate
disability (scores between 30 and 39), and 9.9% had severe
mental health–related disability (scores ,30). There was no
significant difference in MCS scores by age, sex, income,
education, whether the caregiver lived with the patient,
whether the caregiver was a spouse, or by hours of care
given by caregivers or activities-of-daily-living dependence
of the patients.
SF-12 PCS scores.—The mean PCS scores were similar
to state norms (Table 2), and 40.7% of the caregivers had
some degree of disability (PCS score ,50), with 29.7%
having scores less than 40, representing more severe levels
of physical disability. We found no association with sex,
caregiver type, income, or by the burden of care (hours of
care, or activities of daily living dependence). We noted
a significant correlation with age, indicating that older
caregivers had worse PCS scores (Spearman rho¼?.24, p¼
.025), and those who had completed less than 10 years of
schooling also had significantly lower PCS scores (43.1 6
11.7 vs 49.3 6 11.7; t ¼ 2.32, p ¼ .023).
Self-reported stress.—Most caregivers (76.9%) reported
stress associated with the caring role. Self-reported stress
was not related to age, sex, caregiver type (live-in or
spouse), burden of care (hours of care, activities of daily
living dependence), education, or income.
Medical conditions.—Most caregivers reported having
physical health problems (72.4%) and taking medications
(67%) (Table 2). Only 4 reported mental health problems,
and 15 (16.5%) reported taking medications for nervous
conditions. The 4 caregivers who reported mental health
problems were female, aged 43–63 years, and reported
having anxiety or depression or both. Physical health pro-
blems were not related to sex, education, or income but were
more commonly reported by older caregivers (by 90% of
those older than 65 years vs 58% of younger caregivers; chi-
squared ¼ 11.3, p ¼ .001) and by spouse caregivers (85.4%
of spouses vs 59.6% of others; chi-squared¼8.1, p¼.018).
Health problems included arthritis and related conditions
(35.2%), cardiovascular risk factors (predominantly hyper-
tension, 34%), cardiac disease (11%), chronic lung disease
(8.8%), cancer (4.3%), and a variety of other conditions.
Associations Among Stress, Health Problems, and
Table 3 presentsthe MCS and PCS scores in caregivers
according to whether they reported feeling stressed, having
medical problems, or taking medications. The MCS scores
were significantly lower in those who reported stress, mental
health problems, or taking medications for nervous con-
ditions. The PCS scores were also significantly lower in
those who reported stress and in those with physical health
problems and who were taking any medications.
Correlates of SF-12.—Stepwise multiple linear regression
analyses were used to determine whether these variables
(age, sex, self-reported stress, mental health problem,
physical health problem, taking medications for nervous
conditions, taking any medications) were independently
associated with either the MCS or PCS scores. Stress
(Beta ¼ 0.31, p ¼ .002), mental health problems (Beta ¼
5.31, p , .001), and taking medications for nervous
conditions (Beta ¼ 0.26, p ¼ .009) were the only variables
independently associated with (poorer) MCS scores (model
R2¼.25). Physical health problems (Beta¼0.44, p , .001)
and stress (Beta ¼ 0.21, p ¼ .035), but none of the other
variables, were independently associated with low PCS
scores (model R2¼ .25). This latter model was repeated by
substituting the medical conditions (arthritic condition,
cardiovascular risk, heart disease, lung disease, cancer) for
physical health problems. In this model, the PCS scores
were independently associated only with stress (Beta¼0.20,
p ¼ .028), arthritic conditions (Beta ¼ 0.44, p , .001), and
taking medications (Beta¼0.22, p¼.021; model R2¼.36).
Table 1. Duration and Burden of Caring for Dementia Sufferers
Time spent caring, h/wk*
Memory loss, mo*
Need for care, mo*
Supervision Only % Assistance/Dependence %
Proportion needing IADL assistance
Notes: * Median (interquartile range).
ADL ¼ activities of daily living; IADL ¼ instrument activities of daily
Table 2. SF-12 Mental and Physical Component Summary Scores in
91 Dementia Caregivers Compared With Western Australian
Population Data by Age Groups
CaregiversState PopulationCaregiversState Population
Notes: * Mean (95% confidence interval).
yAge 30–64 years, n ¼ 6326; age 65þ years, n ¼ 1427.
BRUCE ET AL.
Correlates of caregiver stress.—We used multiple logistic
regression analysis to explore the relative contributions of
mental and physical health to caregiver stress, entering the
following variables: MCS scores, PCS scores, mental health
problem,taking medications for nervous
physical health problem, and taking any medications. In
this model, PCS scores, MCS scores, and reported mental
health problem were each independently associated with
caregiver stress. Odds ratios (95% confidence interval)
were PCS, 1.11 (1.01–.21), p¼.023; MCS, 1.13 (1.04–.23),
p ¼ .003; and mental health problem, 87.5 (2.37–3221.22),
p ¼ .015. This relationship was not altered by substituting
specific medical conditions for physical health problems.
We studied a representative sample of caregivers of
persons with dementia at the time of first referral to formal
community support services to explore the potential impact
of medical illness on caregiver stress using self-reported
health status and results of the SF-12 health survey. As
a group, they had SF-12 MCS scores lower than population
levels, and a substantial proportion (20%) had MCS scores
consistent with moderate or severe mental health problems
(14). These data are consistent with the findings from
studies that show that caregivers are often stressed when
they first ask for help (15) and with results of our previous
studies using similarly identified participants (7,8).
Most caregivers, especially older ones, reported having
physical health problems and were taking medications.
There was no excess physical disability in caregivers
compared with population levels, although 40% had some
degree of disability and many had low PCS scores
indicating more severe degrees of disability. The major
cause of physical limitation was the category ‘‘arthritic
conditions,’’ which included self-reported osteoarthritis,
previous joint replacement, osteoporosis, and previous
fractures. Importantly, we found strong independent asso-
ciations between caregiver stress and PCS score (i.e., with
physical disability). Although it is hazardous to impute
causation from cross-sectional associations, it seems likely
that physical disability contributed to caregiver stress rather
than vice versa (3).
In contrast, the presence of medical conditions was not
associated with stress or poor MCS scores. This is consistent
with a recent literature review in which the authors
concluded that the evidence linking caregiving with physical
health problems was weak and inconsistent (3). To our
knowledge, physical disability has not been evaluated in this
regard. It has been shown that after the death or
institutionalization of their relative, caregivers subsequently
visit their physicians more often, presumably to attend to
previously neglected health needs (16). Most caregiver
studies focus on psychosocial issues (17), and recent
management guidelines for professionals do not mention
the physical health of caregivers (2,18,19). It is conceivable
that attempts at lessening the impact of physical disability
could benefit the mental health of caregivers.
The presence of reported physical health problems
appeared to be associated with better MCS scores in the
univariate analysis (Table 3). This counterintuitive result
could possibly be explained by the range of ages studied,
because older caregivers reported substantially more
physical health problems and older persons are less likely
to have low MCS scores (6,13). However, the result appears
to be due to differences in the younger caregivers only,
because younger caregivers who did not report physical
health problems had significantly lower MCS scores than
did those with physical problems, whereas there was no
difference among older caregivers (data not shown).
Possibly some younger caregivers are especially vulnerable
to mental health problems resulting from caregiving.
A striking feature of our data is the low rate of self-
reported mental health disorders (only four cases) and the
contrasting high rate of MCS scores indicating serious
mental health problems. More than 25% of caregivers had
MCS scores less than 42, a level suggested to indicate
clinical depression (6) and similar to other estimates of
depression prevalence in caregivers of persons with de-
mentia (3). We know that many persons with mental health
problems do not consult their family physicians (20). We
have reported that caregivers of persons with dementia often
have difficulty discussing their problems with their
physicians even when they have a good relationship and
see them regularly (8). Caregivers of persons with dementia
could be at risk for double jeopardy: having a high risk for
mental health problems that remain undiagnosed.
The SF-12 appears to be an appropriate assessment tool to
assist with assessing caregivers of persons with dementia
because of the breadth of the information it provides. It is
brief, easy to administer, and has excellent performance
parameters, although its utility in the clinical setting may be
limited because of the need to apply weights and to calculate
Our study has several important strengths. We used
a sampling method that we have successfully employed
previously (7,8) and were able to recruit most eligible
Table 3. Mean (6SD) Mental Component Summary (MCS) and Physical Component Summary (PCS) Scores in 91 Dementia Caregivers
Depending on Whether They Reported Feeling Stressed, Having Health Problems, or Taking Medications
Proportion Reporting, %
MCS ScoresPCS Scores
Physical health problems
Mental health problems
Taking any medications
Taking medications for nervous condition
45.7 6 10.0y
48.7 6 9.3*
30.0 6 4.1z
48.7 6 9.5
41.8 6 12.1*
52.7 6 9.5
43.3 6 11.9
48.1 6 9.8
44.4 6 11.3
48.0 6 9.7
45.8 6 12.7*
44.3 6 12.2z
44.6 6 13.8
44.3 6 12.2z
44.3 6 12.2
52.9 6 7.0
56.0 6 5.9
47.7 6 12.0
53.4 6 8.9
53.4 6 8.9
Notes: *p , .05,yp , .01,zp , .001 by t test. [Scores of 50 are close to the population average and lower scores denote poorer health.]
SF-12 IN DEMENTIA CAREGIVERS
participants, indicating that the study sample is representa-
tive of local caregivers of persons with dementia at the time
of initial referral to services in our health care system. In this
as in previous studies (7,8), we found that caregivers who
declined to participate were usually too stressed or burdened
to become involved in research studies, and thus our data
probably underestimate the true mental and physical health
burden of these persons. Similarly, we studied caregivers
relatively early in their caring ‘‘career,’’ and it is conceiv-
able that we underestimated the impact of physical health
problems, because it has been suggested that health effects
from physical disorders take longer to develop (3).
The weaknesses of this study include the lack of controls,
and thus our use of published SF-12 data, but we lack data
comparison data for the other important medical variables.
The study sample is relatively small, with a heterogeneous
group of caregivers of different ages and caregiver types.
We relied on self-reported medical illnesses, although we
would have preferred to have assessed the caregivers’
psychological, medical, and functional status independently.
Finally, the study is cross-sectional in nature, which
precludes any definitive conclusions about causality.
Funded by a Health Promotion Project Grant from Healthway, Western
The authors thank Alison Daly from the Western Australian Health
Surveillance Unit for providing comparative SF-12 data.
Address correspondence to D. G. Bruce, School of Medicine and
Pharmacology, Fremantle Hospital, P. O. Box 480, Fremantle, Western
Australia 6959, Australia. E-mail: firstname.lastname@example.org
1. Donaldson C, Tarrier N, Burns A. The impact of the symptoms of
dementia on caregivers. Br J Psychiatry. 1997;170:62–68.
2. Eccles M, Clarke J, Livingstone M, et al. North of England Evidence
Based Guidelines Development Project: guideline for the primary care
management of dementia. BMJ. 1998;317:802–808.
3. Schulz R, O’Brien AT, Bookwala J, et al. Psychiatric and physical
morbidity effects of dementia caregiving: prevalence, correlates and
causes. Gerontologist. 1995;35:771–791.
4. Vedhara K, Cox NGM, Wilcock GK, et al. Chronic stress in elderly
caregivers of dementia patients and antibody response to influenza
vaccination. Lancet. 1999;353:627–631.
5. Schulz R, Newsome JT, Fleissner K, et al. Caregiving as a risk factor
for mortality. The caregiver health effects study. JAMA. 1999;282:
6. Ware JE, Kosinski M, Keller SD. The 12-item short-form health
survey: construction of scales and preliminary tests of reliability and
validity. Med Care. 1996;34:220–233.
7. Bruce DG, Paterson A. Barriers to community support for the de-
mentia caregiver: a qualitative study. Int J Geriatr Psych. 2000;15:
8. Bruce DG, Paley GA, Underwood PJ, et al. Communication problems
between dementia caregivers and general practitioners delay commu-
nity support services. Med J Aust. 2002;177:186–188.
9. Hickie IB, Andrews G, Davenport TA. Measuring outcomes in patients
with depression or anxiety: an essential part of clinical practice. Med J
10. Bell CM, Araki SS, Neumann PJ. The association between caregiver
burden and caregiver health-related quality of life in Alzheimer disease.
Alzheimer Dis Assoc Disord. 2001;15:129–136.
11. Australia Bureau of Statistics (ABS). Mental Health and Wellbeing
Profile of Adults. Australia 1997, Catalog no. 4326.0; ABS, Canberra,
12. McCallum J, Anderson I. The shorter short form: analysis of SF-12
items in Australian data. In: Australian Institute of Health and Welfare
(AIHW) 1996 Integrating Health Outcomes in Routine Health Care.
Conference Proceedings, AIHW cat no. HOC-1, Canberra, Australia
13. Daly A, Saunders D, Roberts L. 2000 Collaborative Health and Well-
being Survey: an overview. Health Department of Western Australia;
Perth, June 2001.
14. Sanderson K, Andrews G. Prevalence and severity of mental health-
related disability and relationship to diagnosis. Psychiatr Serv. 2002;
15. Audit Commission. Forget Me Not: Mental Health Services for Older
People. London: Audit Commission, 2000.
16. Grasel E. When home care ends—changes in the physical health of
informal caregivers caring for dementia patients: a longitudinal study.
J Am Geriatric Soc. 2002;50:843–849.
17. Perel VD. Psychosocial impact of Alzheimer disease. JAMA. 1998;
18. Doody RS, Stevens JC, Beck C, et al. Practice parameter: management
of dementia (an evidence-based review). Report of the Quality
Standards Subcommittee of the American Academy of Neurology.
19. Cummings JL, Cherry D, Kohatsu NG, et al. Guidelines for managing
Alzheimer’s disease: Part II. Treatment. Am Fam Physician. 2002;65:
20. Andrews G, Henderson S, Hall W. Prevalence, comorbidity, disability
and service utilisation: overview of the Australian National Mental
Health Survey. Br J Psychiatry. 2001;178:145–153.
Received April 16, 2003
Accepted August 4, 2003
BRUCE ET AL.