Predictors of Burden and Infectious Illness in Schizophrenia Caregivers
DENNIS G. DYCK, PHD, ROBERT SHORT, PHD, AND PETER P. VITALIANO, PHD
Objective: The objective of the study was to test predictive models of schizophrenia caregiver burden and infectious
illness episodes for caregivers who had regular contact with their mentally ill family members. Methods: A nurse
interviewer, blind to the patient’s symptoms, caregiver burden, and psychosocial status, administered the Health
Review to 70 caregivers. A second family interviewer, blind to caregiver health status and patient symptoms,
assessed caregiver resources (eg, active coping and social support), vulnerabilities (eg, anger expression and passive
coping) and burden. Concurrently, independent patient raters, blind to caregiver health and psychosocial status,
assessed caregiver stressors. The Brief Psychiatric Rating Scale and the Modified Scale for the Assessment of
Negative Symptoms were used to assess the severity of positive (eg, hallucinations and delusions) and negative (eg,
anhedonia and asociality) symptoms, respectively. Results: Predictive models, including measures of stressors,
resources, and vulnerability factors for caregiver burden and for presence of infectious illness, were each highly
significant, accounting for 40% and 29% of the variance, respectively. However, the specific measures that
predicted burden and infectious illness differed. Greater burden was predicted by more severe patient negative
symptoms (stressor), greater anger control and blame self-coping (vulnerability), and decreased tangible social
support (resource). Presence of infectious illness episodes was predicted by more severe patient positive symptoms
(stressor) and less satisfaction with social support while controlling for the frequency of reporting on the Health
Review. When scores from the Brief Psychiatric Rating Scale (stressors) were categorized into quartiles, it was found
that the frequency of infectious illness in the highest quartile was four times that in the lowest quartile. Other results
indicated that even though burden was not associated with infectious illness, it was associated with “continuing
health problems,” perceived stress, and depression. Conclusions: These data indicate that although schizophrenia
caregiver burden and infectious illness are predicted by measures of patient stressors, vulnerabilities, and resources,
the specific measures predicting these outcomes differ. The results also call attention to the powerful influence of
patient symptoms as a predictor of burden and the presence of infectious illness among caregivers. Key words:
schizophrenia, caregiver, positive symptoms, negative symptoms, infectious illness, burden.
ADL ? activities of daily living; ANOVA ? analysis of
variance; AX ? Anger-Expression Inventory; BPRS ?
Brief Psychiatric Rating Scale; CES-D ? Center for
Epidemiological Studies in Depression Scale; DSM-
IV ? Diagnostic and Statistical Manual of Mental Dis-
orders, 4th edition; FBIS ? Family Burden Interview
Schedule; ISEL ? Interpersonal Support Evaluation
List; MSANS ? Modified Scale for the Assessment of
Negative Symptoms; PSS ? Perceived Stress Scale;
RWCCL ? revised Ways of Coping Checklist; SCID ?
Structured Clinical Interview for DSM-IV; SSQ ? So-
cial Supports Questionnaire.
Over the past 35 years, families of persons with
severe and persistent mental illness have increas-
ingly assumed caregiving responsibilities for their
adult family members. Recent estimates indicate
that between one-third and two-thirds of persons
with long-term psychiatric disabilities currently re-
side with family members. Studies indicate that fam-
ilies are significantly affected by the responsibility
of their caregiving function (1–4; Rheinhard et al.,
unpublished manuscript). Objective demands of and
subjective reactions to this responsibility have been
labeled burden (5). Factors affecting burden include
a number of patient illness variables, such as the
severity of symptoms, length of hospitalization,
number of previous hospitalizations, and length of
The family caregiving burden of persons with se-
vere mental illness includes financial responsibili-
ties, missed work, disturbance of domestic routines,
constraints on social and leisure activities, and re-
duced attention to other family members (15, 16).
The financial strain due to medical costs and the
patient’s economic dependency are considerable
(17). Often family members forego work outside the
home to provide care for the person with mental
illness. Siblings who must cope with the problem
may need therapy, which increases medical bills
(18). Gubman and Tessler (19) suggest that family
members may become ill and/or indebted and that
divorce may result from the chronic financial and
emotional strain. In addition, behavioral problems
of the patient contribute to the experience of burden,
including frequent and intense arguments, with-
From The Washington Institute for Mental Illness, Research and
Training, Washington State University and Eastern State Hospital,
Spokane (D.G.D., R.S.); and Department of Psychiatry and Behav-
ioral Sciences, University of Washington (P.P.V.), Seattle, Washing-
Address reprint requests to: Dennis G. Dyck, PhD, The Washing-
ton Institute for Mental Illness, Research and Training, Washington
State University, 601 W. First Ave., Spokane, WA 99201-3899.
Received for publication August 26, 1998; revision received April
411 Psychosomatic Medicine 61:411–419 (1999)
Copyright © 1999 by the American Psychosomatic Society
drawal, bizarre behavior that is disturbing to neigh-
bors, and threatened or actual harm to self or others.
Also identified as sources of burden are a variety of
household complaints about patient behavior, such
as being noisy at night, failing to adhere to a regular
time schedule, being generally uncooperative, refus-
ing to do household chores, being verbally abusive,
and making unreasonable demands (4). The care-
giver often suffers from severe mental and emotional
drain, feels utterly defeated, and has feelings of anx-
iety, resentment, and anger, with stress being cumu-
lative over time (20, 21). A study of unresolved grief
in families of persons with severe mental illness
found that levels of grief increased over time (22).
Models of Burden and Infectious Illness
Although the economic, social, and emotional im-
pacts of mental illness on families are well docu-
mented (5), there has been little systematic work eval-
uating the role and relative weights of predictive
variables within the context of a conceptual model.
Similarly, little is known about the predictors of care-
giver health status and the extent to which such effects
are mediated by caregiver burden. Although it has
been speculated that a lifetime of caregiving carries the
potential for increasing health risk (16, 18), this has
not been systematically evaluated (23). Because the
experience of caregiving burden is related to many
care recipient and caregiver variables, it is reasonable
to assume that it is a multivariate phenomenon. With
this guiding assumption, Vitaliano et al. recently
adapted a heuristic multivariate model to conceptual-
ize and measure the predictors of burden (23) and
physiological dysregulation (24) in spouse caregivers
of persons with Alzheimer’s disease. In this model,
subjective burden and distress (23) are assumed to
result from three independent sources: stressors, vul-
nerabilities, and resources. The model is represented
by the following expression: Burden ? (Exposure to
Stress ? Vulnerability)/(Psychological and Social Re-
sources). The predictors in the model are exposure to
stressors (care recipient characteristics and level of
positive and negative symptoms), vulnerability (anger
and passive coping: wishful thinking, avoidance,
blame of self, and blame of others), and resources
(active coping: problem focused, seeking support, and
religiosity; and social support: tangible support and
satisfaction with support). We used this conceptual
model to test and compare predictive models of bur-
den and infectious illness.
The sample consisted of 70 persons with a diagnosis of schizo-
phrenia or schizoaffective disorder and their primary family care-
givers. Patient participants were between 21 and 58 years of age, had
regular contact with their parents, and were receiving care at a large
community mental health center in Spokane, Washington.
Patient Symptom Assessment
The assessment of patient participants began with the SCID (25),
which permitted differential diagnosis of schizophrenia, schizo-
phreniform disorder, and schizoaffective disorder. For study inclu-
sion, patients had to have met DSM-IV criteria for schizophrenia or
schizoaffective disorder. The SCID also incorporates diagnostic
modules that allow exclusion of major depressive disorder, bipolar
disorder, brief reactive psychosis, delusional disorder, and psy-
chotic disorder not otherwise specified. After the diagnostic inter-
view, patients were administered their first of up to 28 monthly
assessments of severity of positive and negative symptoms (see
below). Independent raters trained in the use of the BPRS (26) and
MSANS (27) followed a semistructured interview to ensure that
rated phenomena were reviewed.
The 24-item version of the BPRS was used to measure positive
psychiatric symptoms. Positive symptoms are those that character-
ize schizophrenia during an acute exacerbation of the illness (ie,
relapse) and often lead to an intensification of mental health services
or hospitalization. Each symptom was rated on a seven-point scale
of severity ranging from 1 (“not present”) to 7 (“extremely severe”).
The BPRS (28) is the most commonly used measure for evaluating
psychotic symptoms and relapse in patients with severe and persis-
tent mental illness. In the present analyses, we used the mean of the
five psychotic items designed to define positive symptoms: concep-
tual disorganization, hallucinations, unusual thought content, gran-
diosity, and suspiciousness.
Negative symptoms were measured by the MSANS (27). Negative
symptoms are those more persistently stable characteristics of
schizophrenia and include the following five symptom clusters:
affective flattening, alogia, avolition/apathy, anhedonia/asociality,
and attention. The MSANS consists of 24 items that contain five-
point ratings (1 ? “not present”; 5 ? “severe”) of symptoms within
each symptom cluster and five items that provide a global rating for
each of the five symptom clusters. The mean of these five global
ratings served as a composite measure of negative symptoms.
In addition to measuring the severity and type of patient symp-
toms, we recorded several demographic variables (age, education,
and gender), length of illness, and number of psychiatric hospital-
Family Caregiver Assessments
Two interviewers (psychosocial and health), blind to the results
of each other’s assessment and to patient symptom status, assessed
family caregivers. After obtaining informed consent from the patient
and being given permission to talk with their family members, the
psychosocial interviewer provided a brief description of the study,
answered questions, and extended an invitation to meet with the
family to more fully explain the study and to obtain informed
consent. The family was given the option of meeting either in their
home or at an office. The initial family interview began with all
interested family members. A self-rating scale constructed for the
study asked about the frequency of contact, involvement with the
D. G. DYCK et al.
412 Psychosomatic Medicine 61:411–419 (1999)
patient, and caregiving responsibilities. The primary caregiver was
identified and the psychosocial interview initiated. The interview
included a baseline assessment of stressful life events, coping strat-
egies, sources of social support, emotional distress, depression, and
burden. The interview lasted approximately 2 hours. At the end of
the initial interview, a second interview was arranged to assess
Caregiver Health Assessment
The health interview was conducted by a registered nurse who
was blind to the psychosocial status of the caregiver and to the
symptom status of the mentally ill family member. A portion of this
interview used the Health Review (28), which included a checklist
of specific illness symptoms related to infectious disease (primarily
upper respiratory illness), isolated illnesses, and continuing prob-
lems. The tool was used by Kiecolt-Glaser et al. (29) in their assess-
ment of health problems among caregivers of patients with Alzhei-
mer’s disease. The first administration of the tool, on which the
current results were based, was conducted face to face. At this time,
subjects were asked about their symptoms in the past 6 months. It is
noteworthy that Jenkins et al. (28) showed that diagnoses based on
telephone administration were the same as those defined by a com-
puter algorithm using Health Review clusters in 77% of the cases
examined. Subjects were read the symptom list and asked to indi-
cate which symptoms occurred and how long they lasted. Addition-
ally, subjects were asked to note whether they saw a physician for
their problems as well as the number of days that they were unable
to perform their normal daily activities because of their illness. The
International Classification of Disease-9 (30) coding system, estab-
lished by the World Health Organization, was used by a rater as a
guide in determining whether the participant had an infectious
illness condition, an isolated event, or continuing health problems.
Interrater reliability of classifying symptoms as an infectious illness
episode was established on a subsample of 33 interviews. Interrater
agreement on the frequency of infectious illness episodes was 96%.
Caregiver Psychosocial Variables
Caregiver psychosocial variables included demographic vari-
ables (age, gender, education, and income), caregiver burden and
distress, caregiver vulnerabilities (anger expression and passive cop-
ing), and caregiver resources (social support and active coping).
Caregiver burden was assessed with the FBIS (31). This compre-
hensive instrument contains 13 subscales assessing economic bur-
den, assistance in daily living, supervision, and contributions to
household, as well as sections measuring caregiver attributes, such
as worry, attribution, stigma, and family affective response. For the
present study, a composite burden score was developed. Scoring for
each subscale followed the recommendations of the authors. Sub-
scale scores were entered into a principal component analysis. Ten
of the 13 scales were found to load on the first component. A burden
measure was formed from the 10 subscales (the inverse of gratifica-
tions, objective and subjective ADL care, objective and subjective
supervision, subjective financial expenditures, objective impact,
stigma, worry, and displeasure) by ranking subjects by magnitude on
each subscale and then summing the ranks.
Caregiver stress was measured using Cohen’s PSS (32). Designed
to measure the degree to which persons appraise their lives as
stressful, this 14-item scale required participants to rate, from zero to
four (“never” to “always”), the frequency of their daily stressors.
Cohen and Williamson (33) reported that the mean score in the
general population was 20.2 ? 7.8 for women and 18.8 ? 6.9 for
Caregiver depression was assessed using the CES-D scale (34).
The CES-D was developed as a measure of depression in the general
population. The instrument was normed on a sample of more than
3000 respondents from the general population and 105 psychiatric
patients. The scale was found to be predictive of several depressive
outcomes and to be linearly related to severity of a depressive
disorder. The 20-item scale assesses the dimensions of depressed
mood, appetite loss, sleep disturbance, and energy level. Internal
consistency is 0.85 for the general sample and 0.90 for the psychi-
atric group. Test-retest correlations range from 0.51 to 0.67 when
tested over 2 to 8 weeks and 0.32 to 0.54 when tested over 3 months
to 1 year. The CES-D is reported to have excellent concurrent valid-
ity, correlating significantly with several other depression and mood
scales. It also discriminates between the psychiatric and general
populations. Scores on the CES-D range from 0 to 60, with scores of
16 and above indicative of clinically significant depressive symp-
tomatology in adults.
Measures of caregiver resources, such as social supports and
coping, included the ISEL (35), the SSQ (36), and the RWCCL (37).
The ISEL is a 40-item self-report scale (38) specifically designed to
assess the extent to which emotional and tangible support functions
would be available if needed. The ISEL contains subscales of tangi-
ble support, self-esteem, belonging, and appraisal. Subjects indicate
how much they agree with a list of statements that describe the
presence or absence of support on a five-point Likert scale from
“strongly disagree” to “strongly agree.”
The SSQ has six items, each with the following two parts: the
number of available people the individual can turn to in times of
need in each of a variety of situations and the individual’s degree of
satisfaction with the perceived available support. Separate factor
analyses of the two SSQ scales have shown the number and satis-
faction scores to be composed of different, unitary dimensions with
only a moderate correlation (0.34) between the two components. The
number and satisfaction scales of the SSQ have test-retest reliabili-
ties of 0.75 and 0.79 (4-week interval) and coefficient ? values of
0.90 and 0.93, respectively (36).
The WCCL was originally developed by Folkman and Lazarus
(38) and revised by Vitaliano et al. (37). The revised scales were
shown to be more reliable for medical students, spouses of persons
with Alzheimer’s, and psychiatric outpatients than were the original
scales. The five revised scales that were developed were Problem
Focused (15 items; eg, “made a plan of action and followed it”),
Wishful Thinking (8 items; eg, “wished for a miracle to happen”),
Seeks Social Support (6 items; eg, “talked to others and accepted
their sympathy”), Avoidance (10 items; eg, “tried to forget the whole
thing”), and Blame Self (3 items; eg, “blamed yourself”). Other
coping subscales that proved informative in the present study were
Religiosity (3 items; eg, “prayed about it”) and Count Your Blessings
(6 items; eg, “focused on the good things in my life”). For each
coping item, respondents are asked to record the degree to which
they used that strategy to deal with their problem. Options include
“never used,” “rarely used,” “sometimes used,” and “regularly
used” and are scored from 0 to 3, respectively. In previous work with
caregivers of patients with dementia of the Alzheimer type, Vi-
taliano et al. (39) reported that distressed caregivers had lower
Problem Focused and higher Wishful Thinking scores than their
Anger and hostility, behavioral dispositions long regarded as
important factors in the pathogenesis of essential hypertension and
coronary heart disease (40, 41), were assessed with the AX (42). The
AX is a 24-item scale that, in addition to providing an overall
measure of the frequency with which anger is expressed, contains
three 8-item subscales: anger-in, anger-out, and anger control. An-
ger-in reflects the tendency to suppress or hold in angry feelings,
ILLNESS IN SCHIZOPHRENIA CAREGIVERS
413 Psychosomatic Medicine 61:411–419 (1999)
anger-out reflects the tendency to express anger toward environmen-
tal objects or other people, and anger control measures the frequency
with which an individual attempts to control the expression of
anger. The internal consistency, test-retest reliability, and conver-
gent and discriminant validity of these scales are quite acceptable
and are summarized by Spielberger et al. (42). In the present study,
internal consistency was 0.81 for total anger expression, 0.76 for
anger-in, 0.71 for anger-out, and 0.83 for anger control.
Results are presented in five sections: 1) description
of the sample; 2) summary statistics on measured vari-
ables; 3) partial correlations for the burden and infec-
tious illness measures with each measure of the pre-
dictive variable categories of the conceptual model (ie,
stressors, resources, and vulnerabilities), controlling
for under- and overreporting bias; 4) multiple regres-
sion analyses predicting burden and infectious illness
episodes from significant variables within each of the
measured categories of the conceptual model; and 5) a
categorical breakdown of the BPRS measure of positive
symptoms showing the associations with the fre-
quency of caregiver infectious illness.
The mean age of the 70 patients was 33 (SD ? 9)
years. They were predominantly men (74%) and white
(94%). The mean length of patient illness was approx-
imately 10 (SD ? 8) years, and the average number of
psychiatric hospitalizations over that time was 5 (SD ?
6). Less than half of the patients lived with their fam-
ilies (40%), but usually the caregiver was a parent
(77%) and married or living with a partner (67%). The
mean age of caregivers was 52 (SD ? 12) years. Unlike
the patients, caregivers were predominantly women
(83%). Approximately 91% of caregivers had com-
pleted high school, and 26% had completed college.
Median annual family income was approximately
Summary Statistics of Measured Variables
The stressor, vulnerability, and resource variables
all had acceptable levels of internal consistency (Table
1). The sample mean PSS scores (21.5) were similar to
those reported by Cohen and Williamson (33) for men
and women in the general population. However, mean
CES-D scores (14.1) were close to the level (16.0) es-
tablished as the cutoff for clinically significant depres-
sion. It is noteworthy that 31% of the sample scored 16
or higher on the CES-D. A review of a subsample of
caregiver medical charts for a 2-year period before
study entry revealed that 9 of 33 (27%) had been
prescribed antidepressant medications. Sample means
on the psychological vulnerability (anger expression)
and resource (social support) measures were within
the normal range.
The mean level of positive symptom severity on the
five-item BPRS (4) total was in the “very mild” range
(9.9), with scores ranging from “no symptoms” (5.0) to
“moderate symptoms” (22.0). Similarly, the mean
level of negative symptoms as reflected by the five-
item MSANS total was in the “mild” range (8.4), with
scores ranging from “no symptoms” (5.0) to “moderate
symptoms” (16.5). The mild to moderate level of
symptom severity observed reflects that patients were
recruited from an outpatient setting and that they had
their illness for an average of 10 years.
Caregiver infectious illness measures indicated that
during the 6-month baseline period, 44% of caregivers
were ill. Caregivers also reported an average of 2.7
isolated events and 2.5 continuing health problems,
and, in each case, 79% of the sample reported having
A number of important relationships were found
between measures of burden and the presence of in-
fectious illness and caregiver vulnerability, resources,
and stressors (Table 1). Some were likely measures or
sequelae of burden. For example, depression is likely
an expression of subjective burden. Other variables,
listed as stressors, were related to burden in a different
manner. For example, negative symptoms of the pa-
tient likely contribute to caregiver burden rather than
being merely a manifestation of burden. It was inter-
esting that perceived stress correlated with both bur-
den and illness. Positive patient symptoms and the
number of years since onset of patient illness corre-
lated significantly with caregiver illness as well.
Variables for social support tended to be inversely
related to burden. ISEL subscales listed under Re-
sources in Table 1 were inversely related to burden.
Social support satisfaction, measured by the SSQ, also
showed a negative relation.
It was counterintuitive that caregivers demonstrat-
ing a problem-focused coping strategy tended to report
high burden. The other forms of coping, listed as re-
sources, did not correlate significantly with burden. It
is significant, however, that caregivers expressing cop-
ing strategies involving faith in a deity and those sat-
isfied with their social support systems tended not to
have an infectious illness.
Of the variables listed under vulnerabilities, all cop-
ing strategies were found to correlate significantly
with burden. Controlled anger correlated with burden,
whereas outward anger correlated with illness.
D. G. DYCK et al.
414 Psychosomatic Medicine 61:411–419 (1999)
Age of the caregiver and patient correlated inversely
with burden, and male caregivers reported less burden
than did female caregivers. Older caregivers also had
fewer infectious illness episodes.
Two predictive models were investigated, one for
caregiver burden and one for the presence of infectious
illness (Table 2). Both regression equations yielded
significant overall probability levels and accounted for
40% and 29% of the variance in dependent variables,
respectively. However, the variable makeup of the pre-
dictor models of burden and infectious illness differed
from one another.
Candidate predictors for the multivariate models
were chosen on the basis of their significant partial
correlation with the dependent variables (Table 1).
Variables significantly correlated with burden were
selected for entry into a multiple stepwise linear re-
gression analysis. Depression and perceived stress
were considered manifestations of distress and were
closely linked with burden (r ? 0.42 and 0.54, respec-
tively). They were not, therefore, entered into the mul-
tiple regression equation for predicting burden.
The results of the regression analysis for burden
(Table 2) indicated support for the heuristic multivar-
iate model, with 40% of the variance accounted for by
the predictors. More severe patient negative symptoms
(stressor) and greater levels of anger control and
blame-self coping (vulnerability) were associated with
higher caregiver burden scores, and increased tangible
social support (resource) predicted lower burden.
Younger patient age predicted higher burden, and be-
cause age was not specified in the heuristic model, it
was broken out separately in the prediction model.
When entered into the model first, patient age ac-
counted for 16% of the variance in burden. The four
psychosocial variables accounted for the additional
24% of variance.
TABLE 1. Internal Consistency Reliability Coefficients and Partial Correlation Coefficients (Controlling for Frequency of Reporting
Continuing Problems and Isolated Events) for Burden and Infectious Illness
Social support satisfaction
Tangible social support
Seeks social support coping
Count your blessings coping
bNA ? not applicable.
ILLNESS IN SCHIZOPHRENIA CAREGIVERS
415Psychosomatic Medicine 61:411–419 (1999)
The method for predicting the presence vs. absence
of infectious illness episodes involved selecting the
variables that were significantly correlated with the
presence of illness (Table 1) and then entering them
into a stepwise multiple logistic regression analysis.
Stressor variables that were entered included per-
ceived stress, patient illness onset, and positive symp-
toms. Resource variables were religiosity and satisfac-
tion with social support. Vulnerability markers were
blame-others coping and anger-out. Finally, the sole
demographic variable significantly correlated with in-
fectious illness, caregiver age, was entered. The regres-
sion analysis for infectious illness attempted to control
for under- and/or overreporting biases by entering the
frequency of reporting continuing problems and iso-
The results indicated that positive symptoms (stres-
sor) predicted the presence of infectious illness (Table
2), and increased satisfaction with social support pre-
dicted the absence of infectious illness episodes. The
frequency of reporting isolated events was a significant
covariate, whereas the frequency of reporting continu-
ing problems was dropped in the stepwise procedure.
Together, the significant predictors explained 29% of
the variance in infectious illness episodes. The two
psychosocial variables accounted for 25% of the vari-
ance, and the frequency of reporting isolated events
accounted for another 4% of total variance.
Categorical Breakdown of Positive Symptoms
To appreciate the magnitude of the association be-
tween positive symptoms on the BPRS and infectious
illness, the BPRS distribution was divided into quar-
tiles, and an ANOVA was used to examine differences
in the frequency of infectious illness episodes. The
first BPRS group (N ? 10) had no positive symptoms,
the second group (N ? 35) had mild symptoms, the
third group (N ? 16) had mild-to-moderate symptoms,
and the fourth group (N ? 8) had moderate-to-severe
symptoms. Results of ANOVA for the linear trend were
highly significant (F(1,65) ? 6.54, p ? .015), with the
group having the highest BPRS score exhibiting about
four times as many infectious illness episodes as the
group having the lowest BPRS score (Fig. 1). In the
group with the highest score, almost 80% of the care-
givers had an infectious illness episode during the
There is now a considerable body of research sup-
porting the idea that chronic exposure to stressors is
associated with alterations in host resistance through
Chronic stress can also lower health status indirectly
through alterations in health behaviors, such as in-
creased smoking and alcohol consumption (43). Al-
though mechanisms of stress-induced changes on in-
fectious illness have not been fully characterized, host
resistance to acute respiratory illness or infection may
decrease with increased stress (44, 45), and suscepti-
bility to the common cold may be increased (46).
Moreover, evidence links interpersonal relationships,
immune function, and health (29, 47). Prospective
Fig. 1. Proportion of schizophrenia family caregivers (CGs) who
reported at least one infectious illness episode during a
6-month period by the severity of the care recipient’s pos-
TABLE 2. Regression Analyses Predicting Burden and
Infectious Illness Episodes
Burden, adjusted R2? 0.40
0.348 3.44 .001
Satisfaction with support
?0.453 5.28 .022
aPresence vs. absence of an infectious illness episode during a
D. G. DYCK et al.
416Psychosomatic Medicine 61:411–419 (1999)
studies controlling for baseline health status have
shown increased mortality among individuals with
fewer relationships (48), and connections between nat-
ural killer cell function and social support have been
reported among breast cancer patients (49). Thus,
stress and interpersonal relationships are linked to
suppressed immunity and health (47). Building on
previous research with caregivers of patients with de-
mentia, the present study demonstrates significant re-
lationships between caregiving stressors in family
members of patients with schizophrenia, burden, and
the presence of infectious illnesses. More importantly,
consistent with Vitaliano et al.’s (23) conceptual
framework, the results indicated strong support for
predictive models involving measures of stressors,
vulnerability factors, and personal and social re-
The predictive model for caregiver burden indicated
that the most potent stressor variable was the care recip-
ient’s negative symptoms. This observation is consistent
with those of several other studies that have indicated
that the negative symptoms (eg, lack of conversation,
underactivity, slowness, and few leisure interests) are
to speculation, but it may be that the more temporally
stable negative symptoms, relative to the episodic posi-
tive symptoms, translate into increased caregiving de-
mands for family members. The negative symptoms in-
clude a variety of functional and cognitive difficulties
that have burden implications for caregivers (eg, in-
creased need for assistance in daily living and household
chores). Also, consistent with the findings of at least one
other study (53) on schizophrenia caregivers, the model
for burden included increased tangible support as a re-
source variable. Similarly, Vitaliano et al. (23) found that
increased satisfaction with support networks was asso-
ciated with decreased objective and subjective burden.
Finally, the model for burden included two vulnerability
measures, anger control and blame-self coping. It is con-
ceivable that the frequency with which an individual
attempts to control the expression of anger is a marker of
subjective burden. Interestingly, in a longitudinal study
of caregivers of persons with dementia, increases in an-
ger control were associated with weight gain (54). The
sistent with its classification as a vulnerability factor and
its frequently reported association with depression.
Although not included as potential predictors of
burden in the stepwise procedure, wishful thinking
coping, blaming others, and avoidance were all signif-
icantly correlated with burden. All of these coping
measures were highly correlated with each other and
with blaming self, and their positive correlation with
burden was consistent with findings of previous work
(23). Although there are few published data on the
relationship between coping skills and burden in care-
givers of patients with schizophrenia, the finding that
these measures of passive coping were associated with
burden is consistent with the observation that passive
coping is associated with psychological distress in
other studies (23).
The model for predicting the presence of infec-
tious illness episodes among schizophrenia caregivers
once again supported Vitaliano et al.’s (23) overall
conceptualization, but variable makeup was somewhat
different than it was for burden. First, the most potent
stressor variable was the care recipient’s positive
symptoms. Although the stable negative symptoms
contributed to caregiver burden, the episodic positive
symptoms (which typically predict or accompany psy-
chiatric crises, such as relapse and hospitalization)
were predictive of the presence of caregiver infectious
illness. On the other hand, increased satisfaction with
social support predicted decreased frequency of infec-
tious illness episodes. The beneficial health effects of
satisfaction with support have been frequently re-
It is noteworthy that burden was not significantly as-
sociated with infectious illness. Thus, burden did not
seem to represent a final common pathway to acute in-
events was included in the model predicting infectious
illness, thereby statistically controlling and reducing the
chances that the predictive relationships observed were
due to a generalized “physical complaint bias.”
The findings linking patient symptoms with ill-
nesses of caregivers were consistent with those of
many studies that have linked the chronic stress of
caregiving for patients with dementia to compromised
immune function and decreased physical health. Al-
though there are undoubtedly a number of common
elements that emerge in the course of coping with
dementia and schizophrenia, there are also a number
of differences, as noted by Biegel et al. (55). First, the
illness trajectory for persons with schizophrenia tends
to be less predictable and more cyclical in nature than
the trajectory for dementia. This is particularly true of
the positive symptoms (eg, hallucinations, delusions,
and disorganized thinking). In contrast, Alzheimer’s
disease demonstrates a steady downward path, which
may give caregivers an opportunity to anticipate and
adjust to the decline in the care recipient’s functioning
over time. Second, Biegel et al. have noted that care-
givers of Alzheimer’s disease patients may be more
likely than caregivers of schizophrenia patients to in-
stitutionalize the patient when the caregiving de-
mands become unmanageable. Also, placements of
ILLNESS IN SCHIZOPHRENIA CAREGIVERS
417Psychosomatic Medicine 61:411–419 (1999)
persons with dementia are more likely to be perma-
nent as compared with persons with schizophrenia,
who may go back and forth between living with the
caregiver and being an inpatient. Indeed, in recent
years, the shift to managed care has made it increas-
ingly difficult to hospitalize psychiatric patients, plac-
ing even more demands on the family.
Caring for a disabled family member over an extended
time period can exert a negative impact on a caregiver’s
physical health (56). In a recent review, it was concluded
that physical health (as assessed by self-report), health-
care utilization, and immune function were negatively
affected among dementia caregivers (56). Although phys-
iological measures were not obtained in the present
study, the increase in reported infectious illness associ-
ated with positive symptoms is certainly consistent with
the possible involvement of physiological alteration. In
fact, it is known that immune compromise (29), as well
as metabolic (24) and cardiovascular risk factors, are
higher in caregivers of persons with dementia than in
matched control subjects (57).
In the past, research on caregivers of schizophrenia
patients was confined to describing the level, nature,
and predictors of caregiver burden. Greenberg et al.
(58) have presented preliminary evidence that stigma
and worry may be associated with lower levels of
self-reported physical well-being among mothers of
persons with a schizophrenic illness. The current
study is the first to provide evidence for a relationship
between patient stressors, schizophrenia caregiver
burden, and infectious illness. Future data from the
current study will provide information on the extent to
which the observed relationships between psychiatric
patient symptoms, family burden, and health can be
modified by a family education and support interven-
tion. This intervention is designed to improve infor-
mal and formal supports and to reduce caregiver iso-
lation as well as psychiatric symptoms and relapse
among patients (59).
There are several study limitations that must be
noted. First, the sample of the present study may not
be representative of the larger population of caregivers
of schizophrenia patients. More specifically, the care-
givers in the sample elected to participate in a 2-year
intervention study and may not represent eligible non-
participants. In a similar vein, external validity is
threatened because the study sample does not ade-
quately represent minority populations or male care-
givers. Second, the current study is correlational in
nature and did not follow a controlled manipulative
design; thus, inferences concerning causal hypotheses
are limited. Finally, it is possible that other chronic
medical conditions, such as hypertension and diabetes
mellitus, could have been selected that would perhaps
have shown stronger relationships than our measure of
continuing health problems. Because the medical
charts of all caregivers for the 2 years before the inter-
vention, as well as the 2 years during the intervention,
will be obtained, it will be possible to examine other
chronic medical conditions in a subsequent report.
This research was supported by the National Insti-
tute of Mental Health (Grant R01MH52259). Some of
the results presented in this article were presented as a
citation paper at the 1998 Annual Meeting of the
American Psychosomatic Society, Clearwater Beach,
Florida, March 12–14, 1998.
1. Clausen JA, Yarrow MR. The impact of mental illness on the
family. J Soc Issues 1955;4:6–10.
2. Grad J, Sainsbury P. The effects that patients have on their
families in a community care and a control psychiatric service:
a two-year follow-up. Br J Psychiatry 1968;114:265–78.
3. Norbeck JS, Chafetz IL, Skodol-Wilson J, Weiss SJ. Social sup-
port needs of family caregivers of psychiatric patients from three
age groups. Nurs Res 1991;40:208–13.
4. Fisher GA, Benson PR, Tessler RC. Family response to mental
illness: developments since deinstitutionalization. In: Greenley
JR, editor. Mental disorder in social context. Greenwich, CT: JAI
Press; 1990. p. 203–36.
5. Maurin JT, Barmann-Boyd C. Burden of mental illness on the
family: a critical review. Arch Psychiatr Nurs 1990;2:99–107.
6. Brown G, Bone M, Dalison B, Wing J. Schizophrenia and social
care. London: Oxford University Press; 1966.
7. Davis AE, Dinitz S, Pasaminick. Schizophrenics in the new
custodial community. Columbus, OH: Ohio State University
8. Doll W. Family coping with the mentally ill: an unanticipated
problem of de-institutionalization. Hosp Community Psychiatry
9. Group for the Advancement of Psychiatry. The chronic mental
patient in the community. New York: Mental Health Materials
Center; 1978. p. 281–379.
10. Freeman H, Simmons O. The mental patient comes home. New
York: Wiley and Sons; 1963.
11. Herz MI, Endicott J, Spitzer RL. Brief versus standard
hospitalization: the families. Am J Psychiatry 1976;133:
12. Hoenig J, Hamilton MW. The schizophrenic patient in the com-
munity and his effect on the household. Int J Soc Psychiatry
13. Myers J, Bean L. A decade later: a follow-up of social class and
mental illness. New York: Wiley and Sons; 1968.
14. Thompson E, Doll W. The burden of families coping with the
mentally ill: an invisible crises. Fam Relations 1982;31:379–88.
15. Kreisman DE, Joy VD. Family response to the mental illness of a
relative: a review of the literature. Schizophr Bull 1974;10:
16. Hatfield AB. Coping and adaptation: a conceptual framework for
understanding families. In: Hatfield AB, Lefley HP, editors.
Families of the mentally ill. New York: Guilford; 1987. p.
17. Franks DP. Report on economic expenses of families of the
D. G. DYCK et al.
418 Psychosomatic Medicine 61:411–419 (1999)
chronically mentally ill. Washington, DC: National Institute of
Mental Health; 1987.
18. Lefley HP. Research directions for a new conceptualization of
families. In: Lefley HP, Johnson DL, editors. Families as allies in
treatment of the mentally ill: new directions for mental health
professionals. Washington, DC: American Psychiatric Press;
1990. p. 127–62.
19. Gubman GD, Tessler RC. The impact of mental illness on
families: concepts and priorities. J Fam Issues 1987;8:226–45.
20. Hatfield A. Psychological costs of schizophrenia in the family.
Social Work 1978;13:727–36.
21. Noh S, Turner RJ. Living with psychiatric patients: implications
for the mental health of family members. Soc Science Med
22. Miller FM, Dworkin J, Ward M, Barone D. A preliminary study
of unresolved grief in families of seriously mentally ill patients.
Hosp Community Psychiatry 1990;41:1321–5.
23. Vitaliano PP, Russo J, Young HM, Teri L, Maiuro RD. Predictors
of burden in spouse caregivers of individuals with Alzheimer’s
disease. Psychol Aging 1991;6:392–402.
24. Vitaliano PP, Scanlan JM, Siegler I, McCormick W, Knoop R.
Coronary heart disease moderates the relationship of chronic
stress with metabolic syndrome. Health Psychol 1990;17:520–9.
25. Spitzer R, William J, Gibbon M, First M. Structured clinical
interview for DSM-IV. New York: Biometrics Research Depart-
ment, New York State Psychiatric Institute; 1989.
26. Overall J, Gorham D. The Brief Psychiatric Rating Scale. Psychol
27. Andreasen NC. Modified Scale for the Assessment of Negative
Symptoms. Washington, DC: Department of Health and Human
Services, Public Health Service, Alcohol, Drug Abuse, and Men-
tal Health Administration; 1984.
28. Jenkins CD, Kraeger BE, Rose RM, Hurst MW. Use of a monthly
health review to ascertain illness and injuries. Am J Public
29. Kiecolt-Glaser JK, Dura JR, Speicher CE, Trask OJ, Glaser R.
Social caregivers of dementia victims: longitudinal changes in
immunity and health. Psychosom Med 1991;53:345–62.
30. Puckett CG. The educational annotation of ICD-9 CM. 4th ed.
Reno, NV: Channel Publications; 1993.
31. Tessler RC, Fisher GA, Gamache G. The family burden interview
schedule. Amherst, MA: Social and Demographic Research In-
stitute, University of Massachusetts; 1992.
32. Cohen S, Kamarck T, Mermelstein S. A global measure of per-
ceived stress. J Health Soc Behav 1983;24:385–96.
33. Cohen S, Williamson G. Perceived stress in a probability sample
of the United States. In: Spacapan S, Oskamp S, editors. The
Social Psychology of Health. Newbury, CA: Sage; 1988. p.
34. Radloff LS. The CES-D scale: a self-report depression scale for
research in the general population. Appl Psychol Measure 1977;
35. Cohen S, Mermelsteen R, Kamarch T, Hoberman HM. Measuring
the functional components of social support. In: Sarason IG,
Sarason BR, editors. Social support, theory, research & applica-
tions. Boston: Martenus Nijhoff and Dordrecht; 1985. p. 73–94.
36. Sarason IG, Sarason BR, Shearin EN, Pierce GR. A brief measure
of social support: practical and theoretical implications. J Soc
Pers Relationships 1987;4:497–510.
37. Vitaliano P, Russo J, Carr JE. The ways of coping
checklist–revision and psychometric properties. Multivariate
Behav Res 1985;20:3–26.
38. Folkman S, Lazarus RS. Analysis of coping in a middle aged
community sample. J Health Soc Behav 1980;21:219–39.
39. Vitaliano P, Maiuro RD, Russo J, Katon W, DeWolfe D, Hall G.
Coping profiles associated with psychiatric, physical health,
work, and family problems. Health Psychol 1990;9:348–76.
40. Barefoot JC, Dahlstrom WG, Williams RB. Hostility, CHD inci-
dence, and total mortality: a 25-year follow-up of 255 physi-
cians. Psychosom Med 1983;45:59–63.
41. Shekelle RB, Gale M, Ostfield AM, Paul O. Hostility, risk of
coronary heart disease and mortality. Psychosom Med 1983;45:
42. Spielberger C, Johnson E, Russell S, Crane R, Jacobs G, Worden
T. The experience and expression of anger: construction and
validation of an anger expression scale. In: Chesney MA, Rosen-
man RH, editors. Anger and hostility in cardiovascular and
behavioral disorders. Washington, DC: Hemisphere; 1985. p.
43. Belloc N, Breslow L. Relationship of physical health status and
health practices. Prev Med 1972;1:409–21.
44. Graham NMH, Douglas RB, Ryan P. Stress and acute respiratory
infection. Am J Epidemiol 1986;124:389–401.
45. Boyce WT, Jensen EW, Cassell JC, Collier AM, Smith AH, Ramey
CT. Influence of life events and family routines on childhood
respiratory tract illness. Pediatrics 1977;60:609–15.
46. Cohen S, Tyrrell DAJ, Smity AP. Psychological stress and sus-
ceptibility to the common cold. N Engl J Med 1991;325:606–12.
47. House JS, Landis KR, Umberson D. Social relationships and
health. Science 1987;241:540–4.
48. Oxman TE, Berkman LF. Assessment of social relationships in
elderly patients. Int J Psychiatry Med 1990;20:65–84.
49. Levy S, Herberman R, Lippmow M, d’Angelo T. Correlation of
stress factors with sustained depression of natural killer cell
activity and predicted prognosis in patients with breast cancer.
J Clin Oncol 1987;5:348–53.
50. Maurin JT, Barmann Boyd C. Burden of mental illness on the
family: a critical review. Arch Psychiatr Nurs 1990;2:99–107.
51. Creer C, Wing J. Living with a schizophrenic patient. Br J Hosp
52. Raj L, Parmanand K, Ajit A. Social burden of positive and
negative schizophrenia. Int J Soc Psychiatry 1991;37:242–50.
53. Potasnik H, Nelson G. Stress and social support: the burden
experienced by the family of a mentally ill person. Am J Com-
munity Psychol 1984;12:589–607.
54. Vitaliano PP, Russo J, Scanlan J, Greeno C. Weight changes in
caregivers of Alzheimer’s care recipients: psychobehavioral pre-
dictors. Psychol Aging 1996;11:155–63.
55. Biegel D, Milligan SE, Putnam PL, Song LY. The role of race in
family caregiving with persons with mental illness: burden,
support systems and use of self-help. Cleveland, OH: Center for
Practice Innovation, Case Western Reserve University; 1992.
56. Schulz RA, Visintainer P, Williamson G. Psychiatric and phys-
ical morbidity effects of caregiving. J Gerontol 1990;45:181–91.
57. Vitaliano P, Russo J, Barley S, Young HM, McCann BS. Psycho-
social factors associated with cardiovascular reactivity in older
adults. Psychosom Med 1993;55:164–77.
58. Greenberg JS, Greenley JR, McKee D. Mothers caring for an adult
child with schizophrenia: the effects of subjective burden on
maternal health. Madison, WI; Mental Health Research Center,
University of Wisconsin; 1992.
59. McFarlane WR, Link B, Dushay R, Marchal J, Crilly J. Psycho-
educational multiple family groups: four-year outcome in
schizophrenia. Fam Processes 1995;34:127–44.
ILLNESS IN SCHIZOPHRENIA CAREGIVERS
419 Psychosomatic Medicine 61:411–419 (1999)