The Mental Health of Informal Caregivers
in Ontario: An Epidemiological Survey
Jeanette J. Cochrane, BA(Hons), Paula N. Goering, PhD, and Joy M. Rogers, RN, MSc
Introduction
Over the past 2 decades, Canadian and
American social policy has placed increas-
ing emphasis on home care of the
chronically ill, elderly, disabled, and men-
tally disturbed. Rooted in concerns about
escalating health care costs and the damag-
ing effects of institutionalization, this policy
direction has created massive reverbera-
tions throughout all levels of the health and
mental health systems. One unintended and
little-noticed impact of this change is the
increased burden of care on informal sup-
ports-that is, families and friends who
function as primary caregivers.
Large-scale community surveys docu-
menting the prevalence of problems
associated with caregiving by informal
sources are rare. According to one Ameri-
tion of can source, at least 80% of those needing
tex: m long-term care receive it from informal
sources.' A study sponsored by the Ameri-
can Association of Retired Persons and the
Travellers Foundation estimated that up to 7
million Americans, 75% of them women
and more than 50% holding full-time jobs,
are currently providing care for an average
of 12 hours a week and have been doing so
for at least 2 years.' The US Census Bureau
National Long-Term Care Survey conducted
in 1984 found that approximately 2.2 mil-
lion caregivers were caring for 1.6 million
disabled elderly persons and that 37.4% of
-all caregivers were adult children, with
spouses constituting another 35.5%.2 How-
ever, these studies have not investigated the
prevalence of psychiatric disorder, physical
illness, utilization of services, or disability in
their survey populations.
The literature provides a large amount
of information about the circumstances,
tasks, and effects of caring for a relative.
However, the information is predominantly
suppon servies. based on small studies that used conve-
Health. 1997;87:20 nience or clinical samples.?20 These studies
m. 2fA .i.
have documented that caregiving can be an
arduous and debilitating experience and that
it constitutes a potential threat to physical,
psychological, and social health.21 Findings
suggest that the toll can be especially high if
the care recipient has a severe and pro-
longed illness or disability.22 But with few
exceptions,23'24 outcome indicators have
been restricted to measures of general psy-
chological distress, symptoms of depression,
and physical illness.
Assessment of potential unmet needs
also requires information about whether
caregivers are receiving services. Few stud-
ies of caregivers report on their health
service utilization, and those that do have
produced inconsistent results. Haley et al.25
reported that caregivers made more fre-
quent visits to their physicians and used
more prescription drugs than a matched
group of noncaregivers. In contrast, care-
givers studied by George and Gwyther26 did
not use more medical services than older
adults living in the community.
The first province-wide epidemiologi-
cal surveys of the health and mental health
status of Ontario residents, conducted in
1990 and 1991, have provided a unique
opportunity to investigate the psychiatric
and physical health status and service uti-
lization of caregivers. This paper provides
the first descriptive overview of the psychi-
atric correlates of informal caregiving in the
general population.
The authors are with the Health Systems Research
Unit, Clarke Institute of Psychiatry, and with the
Department of Psychiatry, University of Toronto,
Canada. Dr Goering is also with the Faculty of
Nursing at the University of Toronto.
Requests for reprints should be sent to Jeanette
J. Cochrane, Health Systems Research Unit, Clarke
Institute of Psychiatry, 250 College St, Toronto,
Canada M5T 1R8. Internet correspondence may be
directed to jcochraneghsru.clarke-inst.on.ca.
This paper was accepted December 30, 1996.
December 1997, Vol. 87, No. 12

Methods
The Ontario Health Survey
and Supplement
The Ontario Health Survey and the
supplement to it are community-based epi-
demiological surveys. The survey was
designed to collect physical health and
health risk data. The supplement was
developed to provide estimates of the
prevalence of psychiatric disorders, disabil-
ity, and service utilization among residents
of Ontario.
Sample
The Ontario Health Survey was con-
ducted between January 1 and November
30, 1990. Some 32 000 household respon-
dents were interviewed in their homes by
trained Statistics Canada interviewers. The
sample design used to identify respondents
for the survey included stratification and
multistage probability sampling to select
defined geographic areas and household
dwellings. Eligible respondents were non-
institutionalized civilians aged 15 years
and older. The population of Ontario was
stratified by the 42 public health units in
the province, and each unit was further
divided into urban and rural strata. House-
holds were randomly selected from the
total list of habitable residences; one indi-
vidual in each household was interviewed.
The response rate for the survey was 88%.
The supplement interviews took place
between December 1, 1990, and April 15,
1991. Interviews were conducted by expe-
rienced Statistics Canada field staff who
were rigorously trained in the administra-
tion of the University of Michigan
Composite International Diagnostic Inter-
view.27 The supplement sample was based
on the households participating in the
August to November segment of the sur-
vey. One individual aged 15 years or older
was randomly chosen from each house-
hold; persons between the ages of 15 and
24 were oversampled. A total sample of
9953 household respondents over the age
of 15 completed the supplement interview.
The response rate was 77%.
Persons aged 65 years and older were
given a shortened version of the diagnostic
portion of the supplement to minimize the
burden of the interview, so information for
some psychiatric diagnoses is not available
for this older group. This paper reports
results for all respondents under the age of
65 (n= 8116).
December 1997, Vol. 87, No. 12
Interview and Questionnaire
Psychiatric status was assessed in a
face-to-face interview by means of the
University of Michigan Composite Intema-
tional Diagnostic Interview, 27 which is a
modification of the Composite Intemational
Diagnostic Interview developed by the
World Health Organization.28 The latter is a
structured diagnostic interview that was
designed to be used in epidemiological sur-
veys and administered by trained lay
interviewers. The University of Michigan
version was adapted for use in Ontario by
the Supplement Working Party in collabora-
tion with Dr R. Kessler (Institute for Survey
Research, University of Michigan) and Dr
H.-U. Wittchen (Max Planck Institute,
Munich). For this report, respondents were
classified as having a past-year diagnosis if
they met Diagnostic and Statistical Manual
ofMental Disorders, 3rd edition, revised, 1-
year criteria for Axis I or lifetime criteria for
Axis II disorders. Axis I disorders included
affective (unipolar, bipolar, and dysthymia),
anxiety (social and simple phobias, agora-
phobia, panic disorder, and generalized
anxiety disorder), and substance disorders
(abuse of or dependence on alcohol or other
substances). Axis II disorders were antiso-
cial personality disorder and adult antisocial
disorder. Our summary variable "any diag-
nosis" included any affective disorder, any
anxiety disorder, and any Axis II disorder,
but it excluded substance use disorders.
Physical health status was assessed
with the use of a series of questions from
the Ontario Health Survey. The questions
covered a multitude of problems ranging
from headaches and colds to more serious
disorders such as cardiovascular disease
and cancer. Each health problem was
assigned a disease classification code from
the International Classification ofDiseases,
9th edition.
Disability questions in the supplement
referred to the number of days in the past
month when respondents experienced diffi-
culty in performing normal everyday
activities. Subjects were asked to report on
the number of partial days (limited disabili-
ty) or full days (total disability) in which
they were unable to perform usual activi-
ties, and on the number of days when
extreme effort was required to perform
these activities.
Areas of inquiry regarding service uti-
lization included inpatient episodes,
outpatient services (by general practitioners,
psychiatrists, other specialty physicians,
nurses, psychologists, social workers, occu-
pational therapists, and nonmedical helpers
such as clergy and faith healers), and
Caregivers' Mental Health Status
"other" services, including self-help groups,
telephone hotlines, and work-related ser-
vices designed to help the person get or
keep a job. The questions on utilization
were not associated with any specific symp-
toms and depended on the respondents'
ascription of use to the presence of a self-
defined mental health problem.
The following caregiver question was
used to identify caregivers and to provide
key information about circumstances related
to the caregiving role: "In the last 6 months
have you been involved in providing care
for a close friend, relative or family mem-
ber who needs help because of chronic
illness, old age, disability, mental retarda-
tion or mental illness?" The caregivers were
also asked to indicate the age, sex, and rela-
tionship of each person being cared for and
the place where the care recipient lived.
Finally, caregiver burden was assessed
by two questions, which concerned the
amount of time spent providing care and
caregivers' perceptions of the degree to
which caregiving responsibilities interfered
with their own lives.
Analysis
Three objectives are outlined below,
along with the statistical methods used to
address each objective.
1. To describe the prevalence and
characteristics of caregivers and care
recipients and compare caregivers with
noncaregivers in the community. Propor-
tions and 95% confidence intervals (CIs)
were used to describe characteristics of the
three groups. Differences between care-
givers and noncaregivers were tested with
the chi-squared statistic and the t test.
2. To determine whether being a care-
giver is associated with the following
dependent variables: mental disorder,
physical health problems, disability, and
the use of mental health services. Crude
odds ratios (ORs) and 95% confidence
intervals around these estimates were cal-
culated for each potential association.
Proportions, 95% confidence intervals
around the proportions, and tests of statisti-
cal significance using the chi-squared
statistic are also reported.
3. To estimate the magnitude of the
association between being a caregiver and
each of the dependent variables, once con-
founding or modifying variables are
controlled. On the basis of the literature
and the available data, age, sex, employ-
ment status, and economic disadvantage
were identified as potential confounding or
modifying variables for each of the crude
associations examined in objective 2.
American Journal of Public Health 2003
Cochrane et al.

The potential for confounding was

determined by comparing the magnitude of TABLE 1 -Prevalence of Informal Caregiving in Ontario, Canada: Results from
a 1991 Province-Wide Household Survey
the crude association with that of the
adjusted association, with a difference of In Home Out of Home Total
10% or more considered evidence of con-
founding29 (provided there was no Care Recipients (95% Cl) (95% Cl) (95% Cl)
statistically significant interaction). For Sample n 415 1097 1512
each dependent variable (diagnosis, disabil- Estimated n 387600 794200 1181800
ity, utilization), the following steps were Estimated % 32.8 (29.3, 36.3) 67.2 (63.7, 70.7) 100.0
conducted, using logistic regression: (1) the Age of care recipients, %
<19 y 12.1 (12.1,16.5) 3.3 (1.6, 4.9) 6.2 (4.4, 7.9)
crude associations between caregiving and 19-64 y 52.2 (45.4, 58.9) 33.7 (29.2, 38.2) 39.8 (35.9, 43.6)
the dependent variables were individually >64 y 35.8 (29.6, 41.9) 62.0 (57.9, 65.9) 54.0 (49.7, 56.9)
determined and reported; (2) potential inter- Relationship to caregiver, %
actions between caregiver status and each Spouse 17.2 (12.0, 22.3) .. .a 5.8 (3.9, 7.6)
of the potential confounders were exam- Parent or sibling 33.8 (24.5, 40.2) 36.1 (31.3, 40.9) 35.4 (31.6, 39.1)
Child 16.2 (11.1, 21.2) 3.6 (1.9, 5.3) 7.8 (5.8, 9.8)
ined; (3) the associations between care- Friend ... .a 29.4 (24.9, 33.9) 20.8 (17.7, 23.9)
giving and the dependent variables were Other (in-law, other
adjusted for age, sex, employment status, relative, etc.) 29.3 (22.9, 35.8) 30.8 (25.5, 34.9) 30.3 (26.7, 33.9)
and economic disadvantage separately; and
(4) as it is possible that these variables may Note. Cl = Confidence interval.
aAn absence of estimate means low precision and no accurate estimate possible.
not act as confounders separately but may
do so jointly, a model was built for each
dependent variable, which included being a
caregiver and all the potential confounder
variables. As before, the crude and adjusted caregiver. Among those living elsewhere, ing 1 to 7 hours per week and the remainder
associations were compared for each model. approximately 40% lived alone, 45% lived spending 8 or more hours. "In-home" care-
Statistical significance was set at an a with someone else, and 14% lived in an givers spent significantly more time (z =

level of .05, and all tests were two-sided.
Further, all data were weighted to compen-
sate for the probability of selection and for
nonresponse at household and individual lev-
els, and to reflect the age and sex distribution
institution. (Information is missing for the
remaining 1%.)
As Table 1 illustrates, few of the care
recipients (6.2%) were younger than 19
years old, and not surprisingly, most of
3.63, P < .001) than "out-of-home" care-
givers-a mean of 35.5 hours vs 9.7 hours
per week. In addition, caregiving provided
in the home interfered more with care-
givers' other activities (X2 10.5, 2 dfJ P <
=

of the population of Ontario in 1990. Analy-
ses were conducted with the SUDAAN
statistical package, which takes into account
weighted data and the complex stratified
sampling procedures used in this survey.
Results
Objective 1
Prevalence of caregivers. In this
study, 15.0% of the respondents between
the ages of 15 and 64 reported that they
provided care to persons with serious or
chronic physical, mental, or developmen-
tal problems. The number of caregivers
who were providing care to individuals
who lived elsewhere was approximately
double the number who were caring for
persons living with them (9.4% vs 4.7%).
A small number of caregivers provided
care for more than one person. However,
less than 1% were providing care for per-
sons who lived elsewhere in addition to
caring for someone in their homes.
Prevalence and characteristics of
care recipients. More than 1 million indi-
these lived with their caregivers. Most of
the recipients (54.0%) were over 64 years
of age, and the majority of them did not live
with their caregivers.
The largest group of care recipients
(35.4%) were parents or siblings of the
caregivers. Parents-in-law and other rela-
tives made up the second largest group of
recipients (30.3%). Spouses and children
formed a relatively small proportion of care
recipients (5.8% and 7.8%, respectively);
almost all lived with their caregivers.
Sociodemographic comparison of care-
givers and noncaregivers. Caregivers and
noncaregivers were similar in most of the
social and demographic characteristics
measured. However, differences showed up
with respect to sex, age, and economic fac-
tors. The finding that women were more
likely than men to be caregivers (x2 = 17.8,
1 dfg P < .001) is consistent with previously
reported findings. Caregivers were also
older (X2 = 14.3, 2 df, P < .001) and more
likely to be receiving some form of public
financial assistance (X2 = 10.8, 1 df, P <
.001) than noncaregivers.
Caregiver "burden." Forty-one percent
of all caregivers stated that caregiving tasks
.006). Altogether, 48% of in-home care-
givers reported a moderate or great amount
of interference, compared with 37% of
caregivers who provided assistance to indi-
viduals living outside the caregiver's home.
Although the burden experienced by
the in-home caregivers was much greater
than that for the out-of-home group, tests
revealed no between-group differences on
any of our outcome measures (mental and
physical disorder, disability, and mental
health service utilization). Therefore, in the
following analyses, the two groups of care-
givers are combined.
Objective 2
Mental and physical health status.
Prevalence estimates for University of
Michigan Composite International Diagnos-
tic Interview affective, anxiety, and
substance use disorders for the year prior to
the interview are presented in Table 2. Care-
givers had significantly higher prevalences
for each category of psychiatric disorder
except substance use. Anxiety disorders
were more prevalent among caregivers (OR
1.7; 95% CI 1.3, 2.3) than among non-
= =

viduals in Ontario were estimated to be interfered somewhat or a great deal with caregivers; similarly, more caregivers had a
receiving care from informal sources. their other activities. Time spent providing diagnosis of affective disorder (OR 1.5; =

Approximately one third lived with their care varied, with 75% of caregivers spend- 95% CI 1.0, 2.2).
=

2004 American Journal of Public Health December 1997, Vol. 87, No. 12

High proportions of both caregiver and
noncaregiver groups (68.8% and 62.7%,
respectively) reported at least one physical
problem in the past year. The odds of
reporting a physical health problem were
somewhat higher for caregivers (OR = 1.3,
95% CI = 1.0, 1.6).
Disability. There was little difference
between the groups in the number of days
during which individuals were totally
unable to perform their usual daily activi-
ties. The odds of reporting limited-activity
days (OR = 1.4; 95% CI = 1.1, 1.8) and
days that required "extreme effort" (OR =
1.7; 95% CI = 1.3, 2.3) were higher for
caregivers than for noncaregivers. When
the proportions of caregivers who reported
limited-activity days and extreme effort
days were compared with those of noncare-
givers, they were also found to be elevated
(23.8% vs 18.2%; P < .02) and (21.7% vs
13.9%; P < .001), respectively.
Mental health service utilization. The
use of inpatient services for mental health
problems was very low and did not differ
between the two groups. Approximately
12% of the caregivers had used outpatient
and other services in the past year. The odds
of using services were approximately twice
as high for caregivers as for noncaregivers
(OR= 1.9; 95% CI = 1.4, 2.5) (see Table 2).
Objective 3
No significant interactions were found
between any of the potential confounders
(age, sex, employment status, and economic
disadvantage) and caregiver status for any
of the dependent variables (psychiatric dis-
order, disability, and utilization). The
possible confounding of the relationship
between caregiver status and the dependent
variables by the four variables separately
was then explored by comparing the crude
and adjusted odds ratios. The magnitude of
the difference between these two measures
(adjusted for age, sex, employment status,
and economic disadvantage) was less than
10% and therefore did not provide evidence
for confounding. Finally, when the four
potential confounders were entered jointly,
the adjusted association between caregiving
and the dependent variables did not differ
from the results obtained when they were
entered individually.
Discussion
We estimate that close to 1 million
Ontario residents, or approximately 15% of
the adult population aged 15 to 64, provide
care for an infirm relative or friend. Infor-
December 1997, Vol. 87, No. 12
Caregivers' Mental Health Status
TABLE 2-1-Year Prevalence of Psychiatric Disorders, Disability, and Health
Care Utilization for Mental Health Problems among Caregivers and
Noncaregivers In Ontario, Canada: Results from a 1991 Province-
Wide Household Survey
Caregivers Noncaregivers
(n = 1219), %a (n = 6897), %a
(95% Cl) (95% Cl) Odds Ratio
Psychiatric diagnosis
Any diagnosis (except
substance use disorder) 20.6 (16.9,24.3) 14.9 (13.5,16.3) 1.5 (1.1,1.9)
Any affective disorder 6.3 (4.2,8.3) 4.2 (3.4,4.9) 1.5 (1.0,2.2)
Any anxiety disorder 17.5 (13.9,21.1) 10.9 (9.8,12.2) 1.7 (1.3,2.3)
Any substance use disorder 4.1 (2.4,5.8) 5.3 (4.5,6.0) 0.8 (0.5,1.2)
Disabilityb
Limited disability 23.8 (19.2,28.4) 18.2 (16.7,19.7) 1.4 (1.2,1.6)
Extreme effort 21.7 (17.1, 26.2) 13.9 (12.6,15.3) 1.7 (1.5,2.0)
Total disability 14.2 (11.5,16.9) 14.4 (13.0,15.7) 1.0 (0.8,1.2)
Type of service use
Any use 12.2 (9.2,15.2) 6.9 (6.1, 7.9) 1.9 (1.4,2.5)
Outpatient use 11.4 (8.5,14.4) 6.5 (5.6,7.4) 1.9 (1.3,2.6)
Otherc use 2.8 (1.6,4.0) 1.3 (1.0,1.7) 2.2 (1.3,3.5)
Note. Cl = Confidence interval.
aPercentage of weighted data.
bProportion of individuals who reported at least 1 day in the past month on which they
experienced difficulties performing usual activities.
cincludes self-help groups, telephone hotlines, and work-related services.
mal caregivers are found in all age and both another potential source of stress for those
sex groups. Although caregivers are pre- who are providing care.
dominantly female and over 24 years of Caregiving is associated with higher
age, 41% are male and 18% are young rates of most psychiatric disorders. These
adults under age 24. The majority of care findings are consonant with previous stud-
recipients are parents, in-laws, or siblings ies that have found elevated symptoms of
and are over 64 years of age. Only a few distress and depression, and they add new
live in institutions. information about the prevalence of clinical
The nature of the caregiving situation disorders. Among caregivers, the odds of
varies by setting. Two thirds of care recipi- having an anxiety disorder were 1.7 times
ents live outside the caregiver's home. This the odds among noncaregivers, with 18% of
type of caregiving is typically part-time and caregivers classified in this group. For close
creates somewhat less interference for the to one in five caregivers, symptoms of anxi-
caregiver than that provided in the home. ety coexist in patterns that would be
Caregiving in the home averages more than diagnosed as phobias, panic disorder, or
35 hours per week-that is, the equivalent generalized anxiety. Such syndromes are
of a full-time job-and half of these care- likely to be more severe than stress-related,
givers report that their responsibilities transient experiences of tension, and they
interfere with their lives. Caregiving is not may require psychiatric treatment. Clinical
the only activity or obligation for most of depression is also more prevalent among
those who provide it. Many caregivers are caregivers than among noncaregivers,
also married and working for pay. However, although it affects a smaller proportion,
a notable portion of them are economically with a rate of 6% and an increased odds
disadvantaged. Compared with noncare- ratio of 1.5. Clinical depression is usually
givers, caregivers are less likely to be more disabling than anxiety disorders,30'3'
employed and more likely to be on public so its recognition and treatment are particu-
assistance. The relationship between fewer larly important.
financial resources and caregiving can be We considered the implications of
interpreted in more than one way. It may using different time periods for the care-
reflect limited options for altemative forms giver questions (in past 6 months) and the
of care among the poor, the extent to which diagnostic questions (in past year). There
having a sick relative may deplete financial are two possibilities. First, we may have
resources, or both. Whatever the nature of missed some caregivers who only provided
the relationship, however, limited funds are care in the first 6 months of the year. If the
American Journal of Public Health 2005
Cochrane et al.
prevalence of a disorder among these indi-
viduals is similar to that among the
individuals in our sample, these inclusions
would strengthen the association. Second,
we may have included in the disorder
group some caregivers who had a disorder
only in the period prior to caregiving,
thereby incorrectly classifying these indi-
viduals as having disorders and being
caregivers. Their deletion would weaken
the association. Neither threat is very large
because of the enduring nature of both care-
giving and psychiatric disorder. Two other
community surveys of caregiving provided
some evidence that caregiving is a relatively
long-term task. The length of time spent in
the caregiver role was reported as "at least
two years" in the American Association of
Retired Persons and Travellers Foundation
study'; and in the Long-Term Care Study,
79% of those providing care at the time of
the survey had been caregivers for at least
1 year.2
Caregivers have not only higher odds
of experiencing psychiatric disorders but
also higher prevalence rates of disability
and physical illness. However, the reasons
for these findings cannot be demonstrated
in a cross-sectional study of this type. It is
quite possible that increased illness and
disability are the consequences of a stress-
ful life situation. These indicators may
reflect the severe end of a continuum of
negative response, which ranges from mild
somatic and psychological symptoms to
the development of serious psychiatric and
physical illness. But it is also possible that
the health status of both the caregiver and
the care recipient is influenced by other
predisposing factors. Illnesses of various
kinds may cluster within families, who
share common genetic and environmental
conditions. Longitudinal investigations are
necessary to disentangle such questions of
cause and effect. But regardless of the rea-
sons for the higher prevalence rates, the
magnitude of these problems within the
caregiver group has implications for treat-
ment and service delivery.
Among caregivers, the odds of service
use for mental health reasons are almost
twice those among noncaregivers.
Increased prevalence of psychiatric illness
and disability appears to be related to
increased demand for and use of services.
Even though the odds ratios for having a
disorder and for using services are of simi-
lar magnitude, it cannot be concluded that
the needs of the caregiving group are being
met. One in five of the caregiver group has
a psychiatric disorder. Only one in eight
seeks services for mental health reasons.
2006 American Joumal of Public Health
We also know that there is not a good
match between the two groups.32 So there
is probably a substantial subgroup of care-
givers who need mental health services but
are not receiving them. Because the quality
and adequacy of service use were not
assessed, it is also possible that there are
unmet needs among those who are using
services but not getting appropriate inter-
ventions for sufficient periods of time.
The emotional and social problems
associated with caregiving have been
widely discussed in a growing literature on
the caregiving burden, but small samples
and a focus on particular diseases have lim-
ited the conclusions that can be drawn. The
Ontario Health Survey supplement provides
unique information because of the large
sample, the use of a standardized diagnostic
instrument, and the noncaregiver compari-
son group. Limitations of the study include
the restriction in the number of questions
that could be added to a lengthy interview
schedule. Little information was gathered
about the health problems of the care recipi-
ents and the tasks of caregiving.
Since comparable data about mental
disorders were unavailable for respondents
over age 65, these individuals were ex-
cluded from the analysis, preventing the
examination of an important segment of the
caregiver population.
The results of this study have impor-
tant implications for policy, service
delivery, and clinical practice. Documenta-
tion of the prevalence of caregiving and its
associated mental health problems is of crit-
ical importance as governments and
hospitals continue to move toward commu-
nity-based care. Clinical and technological
advances in psychiatry and medicine allow
many people to survive chronic and life-
threatening illnesses and injuries and to live
in the community. Recent health and social
policies have capitalized on these advances
in the interest of containing health care
costs through minimizing institutionaliza-
tion and maintaining people in their own
communities for as much of their lives as
possible. But to accomplish these worthy
goals, the needs of caregivers must be
acknowledged and met by the establish-
ment of appropriate and readily accessible
support services. As Lefley points out,
"Caregiving for one group at the expense of
another can scarcely be considered a desir-
able mental health objective." 33(pI069)
Furthermore, the desired result in terns of
cost-effectiveness is unlikely to be achieved
unless community-based service systems
attend to the mental health needs of infor-
mal caregivers. O
Acknowledgments
The caregiver study was supported by a grant from
the National Health Research and Development
Program (6606-5069-63B).
We gratefully acknowledge the help of Dr
David Streiner, professor, Department of Psychia-
try and Department of Clinical Epidemiology and
Biostatistics, McMaster University, Hamilton,
Ontario, and Ms Anne Rhodes, doctoral candidate,
Preventive Medicine and Biostatistics, Faculty of
Medicine, University of Toronto, and research fel-
low, Mental Health Systems Program, Department
of Psychiatry, University of Toronto.
The mental health supplement of the Ontario
Health Survey was sponsored by the Ontario Men-
tal Health Foundation with the support of the
Premier's Council on Health, Well-Being, and
Social Justice; the Ministry of Health; and the
Ministry of Community and Social Services,
Ontario. A multisite working group chaired by Dr
Dan Offord, McMaster University, was responsi-
ble for the study design and implementation.
References
1. Weinstein GW. Crisis of elder care. Ms. Octo-
ber 1989:73-79.
2. Stone R, Cafferata GL, Sangl J. Caregivers of
the frail elderly: a national profile. Gerontolo-
gist. 1987;27:616-626.
3. Lawton MP, Moss M, Kleban MH, Glicks-
man A, Rovine M. A two-factor model of
caregiving appraisal and psychological well-
being. JGerontol. 1991;46:P181-P189.
4. Schulz R, Williamson GM. A 2-year longitu-
dinal study of depression among Alzheimer's
caregivers. Psychol Aging. 1991 ;6:569-578.
5. Pruchno RA, Potashnik SL. Caregiving
spouses: physical and mental health in per-
spective. JAm Geriatr Soc. 1989;37:697-705.
6. Arey S, Warheit GJ. Psychosocial costs of liv-
ing with psychologically disturbed family
members. In: Lee N, Rubino PJ, Clayton J,
Wind K, eds. The Social Consequences of
Psychiatric Illness. New York, NY: Brun-
ner/Mazel; 1980:158-175.
7. Thompson EH, Doll W. The burden of fami-
lies coping with the mentally ill: an invisible
crisis. Fam Relations. 1982;31:379-388.
8. Noh S, Turner RJ. Living with psychiatric
patients: implications for the mental health of
family members. Soc Sci Med. 1987;25:
263-272.
9. Jed J. Social support for caretakers and psy-
chiatric rehospitalization. Hosp Community
Psychiatry. 1989;40:1297-1299.
10. Maurin JT, Boyd CB. Burden of mental ill-
ness on the family: a critical review. Arch
Psychiatr Nurs. 1990;4:99-107.
11. Miller B, McFall S, Montgomery A. The
impact of elder health, caregiver involvement,
and global stress on two dimensions of care-
giver burden. J Gerontol. 1991;46:S9-S19.
12. Baillie V, Norbeck JS, Barnes LEA. Stress,
social support and psychological distress of
family caregivers of the elderly. Nurs Res.
1988;37:217-222.
13. Deimling GT, Bass DM. Symptoms of mental
impairment among elderly adults and their
December 1997, Vol. 87, No. 12

effects on family caregivers. J Gerontol.
1986;41:778-784.
14. Widerstrom AH, Dudley-Marling C. Living
with a handicapped child: myth and reality.
ChildEduc. 1986;62:359-367.
15. McCubbin HI, McCubbin MA, Patterson JM,
Cauble AE, Wilson LR, Warwick W. CHIP-
Coping Health Inventory for Parents: an
assessment of parental coping patterns in the
care of the chronically ill child. J Marriage
Fam. 1983;45:359-370.
16. Calder SA, Ebmeier KP, Stewart L, Crawford
JR, Besson JAO. The prediction of stress in
carers: the role of behaviour, reported self-care
and dementia in patients with idiopathic
Parkinson's disease. Int J Gerontol Psychiatry.
1991;6:737-742.
17. Schulz R, Visintainer P, Williamson GM.
Psychiatric and physical morbidity effects of
caregiving. J Gerontol. 1990;45:P181-PI91.
18. Hall BL. The role of adult children in helping
chronically ill hospitalized parents. Can J
Aging. 1989;8:68-78.
19. Farkas S. Impact of chronic illness on the
patient's spouse. Health Soc Work. 1980;
5:39-46.
20. Bardach JL. Group sessions with wives of
aphasic patients. Int J Group Psychother.
1969;19:361-365.
December 1997, Vol. 87, No. 12
21. Gallagher DE. Intervention strategies to assist
caregivers of frail elders: current research sta-
tus and future research directions. In: Lawton
MP, Maddox G, eds. Annual Review of
Gerontology and Geriatrics. New York, NY:
Springer Press; 1985:5.
22. Haley WE, Pardo KM. Relationship of severity
of dementia to caregiving stressors. Psychol
Aging. 1989;4:389-392.
23. Drinka TJ, Smith J, Drinka PJ. Correlates of
depression and burden for informal caregivers
of patients in a geriatrics referral clinic. JAm
Geriatr Soc. 1987;35:522-525.
24. Gallagher D, Wrabetz A, Lovett S, Del Mae-
stro S, Rose J. Depression and other negative
affects in family caregivers. In: Light E,
Lebowitz BD, eds. Alzheimer's Disease Treat-
ment and Family Stress. Directions for
Research. Rockville, Md: National Institute of
Mental Health; 1989. DHHS publication ADM
89-1569.
25. Haley WE, Levine EG, Brown SL, Berry JW,
Hughes GH. Psychological, social and health
consequences of caring for a relative with senile
dementia. J Am Geriatr Soc. 1987;35:
405-411.
26. George LK, Gwyther LP. Caregiver well-
being: a multidimensional examination of
family caregivers of demented adults. Geron-
tologist. 1986;26:253-259.
Caregivers' Mental Health Status
27. Kessler RC, McGonagle KA, Zhao S, et al.
Lifetime and 12-month prevalence of DSM-
III-R psychiatric disorders in the United
States: results from the National Comorbidity
Survey. Arch Gen Psychiatry. 1994;51:8-19.
28. Wittchen H-U. Reliability and validity studies
of the WHO-Composite International Diag-
nostic Interview (CIDI): a critical review. J
Psychiatr Res. 1994;28:57-84.
29. Moldonado G, Greenland S. Simulation study
of confounder-selection strategies. Am J Epi-
demiol. 1993;138:923-936.
30. Hecht H, Wittchen H-U. The frequency of
social dysfunction in a general population
sample and in patients with mental disorders:
a comparison using the Social Interview
Schedule (SIS). Soc Psychiatry Psychiatr Epi-
demiol. 1988;23:17-26.
31. Kouzis AC, Eaton WW. Emotional disability
days: prevalence and predictors. Am J Public
Health. 1994;84: 1304-1307.
32. Lin E, Goering P, Offord DR, Campbell D,
Boyle M. The use of mental health services in
Ontario: epidemiological findings. Can JPsy-
chiatry. 1996;41:572-577.
33. Lefley HP. Aging parents as caregivers of
mentally ill adult children: an emerging social
problem. Hosp Community Psychiatry.
1987;38:1063-1070.
American Joumal of Public Health 2007