1 V
LIBRARY 7
mount: State

University

'\ __._

 

 

 

_’

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.

 

DATE DUE DATE DUE DATE DUE '

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

ll

MSU Is An Affirmative Action/Equal Opportunity lnditution
cWWJ

 

 

 

 

 

 

 

 

 

A PATH MODEL ANALYSIS OF
THE CANCER EXPERIENCE: PATIENT DEPRESSION AND
CAREGIVER RESPONSES AMONG HUSBANDS AND WIVES
CARING FOR A SPOUSE WITH CANCER

BY
Timothy F. Dwyer

A THESIS

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

MASTER OF ARTS

Department of Family and Child Ecology

1990

“use

(9

'1

I
I/—.
,. ,4 5...

ABSTRACT

A PATH MODEL ANALYSIS OF
THE CANCER EXPERIENCE: PATIENT DEPRESSION AND
CAREGIVER RESPONSES AMONG HUSBANDS AND WIVES
CARING FOR A SPOUSE WITH CANCER

BY
Timothy F. Dwyer

The purpose of this study is to describe husbands and
wives responses to caring for a spouse with cancer using a
path model analysis. Specifically, the focus of this
research is to examine patients' functional physical status,
duration of care, and depression and analyze the causal
relationships on caregivers' depression and their responses
to the impact of care on their health, daily schedule, and
finances.

Data utilized were collected among 238 caregivers and
their spouses with cancer over the phone and self-
administered questionnaires. Using ordinary-least squares
regression, a three-equation model was employed to reflect
the relationships in models depicting the three burdens
under study. Findings revealed patients' functional status
affects caregivers' burden responses primarily because it
influences patients' and then caregivers' mental health.
Only in the case of impact on schedule does there appear to
be a direct effect of patient's functional status on

caregiver burden.

ACKNOWLEDGEMENTS

I wish to acknowledge my sincere thanks and
appreciation to Drs. Barbara and Bill Given for the
opportunity to work with them. During my two and a half
years as a graduate assistant with the Family Care Studies,
it has been a pleasure to learn and develop and grow with
the support and mentorship of Drs. Given.

To my distinguished and laudable thesis committee, I
owe much appreciation and thanks: my thesis chair, Dr.
Lawrence Schiamberg, for his guidance and steadfast support;
Dr. Manfred Stommel, for his always thoughtful and astute
comments and suggestions, and especially for his guidance in
conceptualization and analysis; Dr. Sharon King, for her
encouragement and feedback, and a special thanks also for
introducing me to the Family Care Studies; and Dr. Donald
Melcer, for the insight and inspiration I feel he roused in
me, both in and out of the classroom. I wish to acknowledge
the great respect I hold for each of them.

Special acknowledgement is due to my friend and co-
worker, Fady Tawil, for his patience and assistance with the
data runs for my thesis.

Finally, special credit and thanks to Laura, my wife,
for understanding my occassional obsessive ruminations and

for encouraging my perseverance.

iii

TABLE OF CONTENTS

iv

EQQQ
LIST OF TABLES vi-vii
LIST OF FIGURES viii
CHAPTER I - INTRODUCTION 1
Background of the Problem 1-3
Purpose of the Study 3
Research Questions 4
General Conceptual Framework 4-5
Definitions of Variables 6-7
Assumptions 7
Limitations 8
Organization of Chapters 9
CHAPTER II - REVIEW OF RESEARCH AND
THEORETICAL FRAMEWORK 10
Family Cancer Care Concepts & Issues 10-19
Theoretical Framework 19-22
Research Hypotheses 22-25
CHAPTER III - METHODS AND PROCEDURE 26
Research Design 26
Sample Selection 27-28
Sample Description 28
Data Collection 28-29
Operationalization of Major Variables 29-32
Analysis Strategy 32-33
CHAPTER IV - PRESENTATION OF DATA AND ANALYSIS 34
Overview 34
Sample Characteristics 34-36
Characteristics of the Cancer Experience 36-38
Description and Reliability of Measures 39-40
Recursive Path Model Analysis 40-43
Regression Equations for Path Analysis 44-57
Analysis of Research Hypotheses 57-61
Summary 62

CHAPTER V - SUMMARY, CONCLUSIONS, AND DISCUSSION

Overview

Summary of Study

Summary of Research Questions
Primary Conclusions

Implications for Intervention
Implications for Future Research
Summary

APPENDICES
Appendix A - CES-D Scale

Appendix B - Caregiver Response Inventory
Appendix C - Karnofsky Performance Scale

REFERENCES

74
75
76

77-82

LIST OF TABLES

19212

1 Caregiver Sociodemographic Data

2 Patient Sociodemographic Data

3a Cancer Specific Data

3b Cancer Specific Data

4 Characteristics of the Cancer Experience1

5 Characteristics of the Cancer Experience2

6 Reliabilities of Burden and Depression
Measures

7 Scale Means for Burden and Depression
Measures

8 Pearson Linear Correlation Coefficients

9 Regression of Patient Functional Status
with Dependent Variable Patient
Depression

10 Regression of Patient Functional Status
and Patient Depression with Dependent
Variable Caregiver Depression

11a Regression of Patient Functional Status,
Patient Depression, and Caregiver
Depression with Dependent Variable
Impact on Health

11b Regression of Patient Functional Status,
Patient Depression, and Caregiver
Depression with Dependent Variable
Impact on Schedule

11c Regression of Patient Functional Status,

Patient Depression, and Caregiver
Depression with Dependent Variable
Impact on Finances

vi

35

36

36

37

38

38

39

4O

41

45

45

46

50

53

 

13

Effects of Patient Functional Status
on the Caregiver Burden Response to
Impact on Health, Schedule, Finances

Summary Table of Caregiver Burden

Responses Regressed on Cancer
Experience and Depression Variables

vii

Page
56

57

LIST OF FIGURES

Predicted Path Model Depicting the Cancer
Experience, Intervening (Depression), and
Caregiver (Burden) Response Variables

Recursive Path Diagram of Caregiver Burden
Regression Equations for Path Coefficients

Path Model Coefficients for Impact on
Health

Path Model Coefficients for Impact on
Schedule

Path Model Coeffients for Impact on
Finances

viii

43

44

47

51

54

CHAPTER I

INTRODUCTION

c o e

The incidence of cancer in America has become a
startlingly common occurrence. Consider the following
facts: About 985,000 people were newly diagnosed as having
cancer in 1989 (not including the incidence of non-melanoma
skin cancer, which is estimated at 500,000 new cases
annually). About 75 million Americans now living will
eventually have cancer, or about 30% of the present total
population. This means that cancer can be expected to
strike in approximately three out of four families (American
Cancer Society, 1989). Present trends in the delivery of
health care services, such as shorter hospital stays, are
likely to necessitate the need for increasing family
involvement in the care for ill family members. The family
is usually the first line of defense and the mainstay of
support throughout the course of the disease. Exactly how
caring for a family member with cancer affects the family
member providing care is not fully understood at this time.
The need to examine how spouses react to the caregiving role
when their partner has cancer, as well as some of the

contingencies which may affect their reaction (e.g. the

2
patient's response to the cancer experience), is the focus
of this research.

Family care for cancer patients will be discussed in
terms of the patient's need for care and the impact of care
by family caregivers. When illness such as cancer strikes
in a family it commands a response from the entire family.
Family stress theory (McCubbin, Joy, Caudle, Comeau,
Patterson, & Needle, 1980) focuses attention on the ways in
which the family is able to manage the illness and cope with
their responses. Olsen (1970) described the impact of
serious illness on the families as one of crisis and
resulting stress. This event can throw the family system
into disorder and disequilibrium as one member may no longer
be able to function at his or her usual ability, thus
placing new demands on the family as a unit.

The family care is conceptualized in terms of the
informal support that caregivers provide. Indeed, the
family is a critical element in the delivery of health care
services for cancer patients (Northouse & Northouse, 1987).
The prevailing use of prospective reimbursement schemes and
Diagnostic Related Groupings (DRG's) in inpatient medical
care has prompted earlier hospital releases and a greater
dependence on the utilization of outpatient services. One
effect of this is that health care professionals must rely
on the availability of family members to provide many of the
services which are necessary for adequate cancer treatment

and follow-up. Transportation to outpatient cancer

3
treatments, rehabilitative and assisted self-care efforts in
the home, emotional support, and physical care are just some
of the major activities carried out by families providing
care in the home (Stetz, 1987; Lewis, 1983; Grobe, Ilstrup,

& Ahmann, 1982; Welch, 1980).

W

In view of what has been discussed in this brief
introduction, the present study is designed to examine the
factors which contribute to caregiver burden responses, with
specific attention to the effect of patient and caregiver
depression, physical dependency, and the duration of need
for care. In this study, caregiver burdens are viewed as
the subjective response to specific dimensions of impact to
the role of providing care. These responses are also
conceived to be influenced by the affect of caregiver
depression.

The primary goal of this research is to describe and
analyze the effect of those factors which influence the
outcomes of caregiver burden to better understand the effect
of cancer caregiving on the spouse. A secondary goal, to be
discussed following the analysis, includes identifying those
factors which may prove most salient to possible
intervention designs which might assist those burdened

caregivers.

4
W

The general question that is being addressed in this
research is: "how do selected factors of the cancer
experience and patient depression affect the outcomes of
spouse-caregiver burden responses?" More specifically the
present study attempts to deal with the following questions:
1) to what extent do specific characteristics of the cancer
care experience influence caregiver burden responses; and
2) do patients' emotional responses to the cancer
experience, such as depression, help to explain the
relationship to better predict caregivers' burden responses?
Figure 1 depicts the variables under study as they are

conceptualized in the path model.

Ge Con tu ewo k

A family systems framework (Broderick & Smith, 1979)
will serve to guide the development of a path model of
cancer caregiver burden responses. Following a brief
introduction to the theoretical perspective guiding this
research, the major variables to be examined will be
discussed along with the proposed path model illustrating
the relationship of the variables under study.

The cancer experience, as it will be referred to here,
is framed in a family systems perspective which suggests
that a variety of cancer care characteristics will influence
caregiver burden. Specifically, the path model (Figure 1)

posits particular characteristics of the care experience

5
which are likely to influence patients' and caregivers'
levels of depression which, in turn, are suggested to
influence caregivers' burden responses. The negative
impacts of care upon the primary caregiver are
conceptualized in terms of the role strains associated with
living and coping with the physical demands of direct care,
disruption of the caregivers' daily schedule in attending to
the physical and psychological needs of the patient, and
financial strain likely to result from lost opportunity
costs and additional expenses related to medical care

activities, equipment and supplies.

 

"CANCER 7 INTERVENING "CAREGIVER

 

 

 

EXPERIENCE" EMOTIONAL BURDEN"
VARIABLES RESPONSE RESPONSE
VARIABLES VARIABLES
AREGIVER
EPRESSION 1. IMPACT ON
3. DURATION OF CARE v HEALTH

 

- IN MONTHS
2. IMPACT ON

 

 

 

 

 

 

2. PATIENT FUNCTIONAL SCHEDULE
STATUS
3. IMPACT ON
ATIENT FINANCES
EPRESSION

 

 

 

 

 

£I§QRE_1 - Proposed Path Model Depicting the Cancer
Experience, Intervening (Depression), and Caregiver (Burden)
Response Variables

6
A family systems perspective of the cancer care
experience implies that a variety of inputs to the family
system are viewed as influencing patients' and caregivers'
affective intervening states of depression which, in turn,
is posited to influence caregivers' outcome responses to

their role in care.

e ' ' ' V ' es
The following conceptual definitions will help guide
the reader through the selected review of research and
further elaboration of the family systems framework guiding

this research.

Patient. Patient is defined as the identified spouse with
new or recurrent cancer diagnosis.

Caregiver. Caregiver is defined as the spouse who is self-
identified as the person who provides the most care for a

family member with cancer.

Depression. Depression is defined as a general mood state
of non-psychiatric, depressive symptomatology, which
includes cognitive, affective, and behavioral features.
Depression in this study will be measured by the Center for
Epidemiologic Study - Depression scale (CES-D 20), for both

patient and caregiver.

7
geneez_§e;e_§xne;ienee. The cancer care experience is
defined as characteristics of the cancer which account for
the need for care. Specifically, a) duration of care,
measured in months, and b) the physical functional status
of the patient, measured by the Karnofsky Performance Scale,
will serve to operationalize the concept of cancer care

experience for both patient and caregiver.

Canegiyez Bergen Responsee. Caregiver burden responses are

defined as caregivers' perceptions and self-appraisal of the
impact of their role in care on the following dimensions:

A) physical health; B) daily schedule: and C) finances.

W
Assumptions made in this research include:

1. The measures are appropriate to the concepts identified
and sensitive to accurately appraise patients' and
caregivers' depression and caregiver burden responses.

2. The causal ordering in the proposed model is logical
and appropriate to test an exploratory predictive model

of caregiver burden.

finiteness
Limitations of this research include:
1. Subjects who agreed to participate in this study may be
very different from those who chose not to participate.

Therefore, it is quite likely that the findings are not

8
representative of all cancer caregiver dyads. Reliance
on a convenience sample may conceivably select out
those truly depressed or burdened patient/caregiver
dyads who elect not to participate in research studies.

2. All characteristics of the cancer experience are not
explored (i.e. specific patient dependency, patients'
symptom experience and distress, or specific caregiver
attributes such as care involvement).

3. Additional characteristics of the dyad and family
system, which may influence both patient and caregiver
depression and the caregivers' perception of burden in
care, are not explored (i.e. history and quality of the

relationship, or other sociodemographic variables).

A final issue that may be considered a limitation of
this study concerns the exclusion of variables related to
the family and individuals' social-psychological and life-
cycle development. It bears noting that inclusion of these
potentially critical variables was beyond the scope of the

present research.

0 ni ' Cha te s
The presentation of this study in organized into five
chapters. Information in Chapter I is the introduction,
purpose of the study, general research questions and
orientation to the conceptual framework, definitions of the

variables under study, proposed path model, and the basic

9
assumptions and limitations of this research. In Chapter II
a selected review of research and conceptual framework will
be presented. The review of research will link this
analysis with the work others have done regarding the impact
of cancer in families and their responses to the cancer
experience. Following this brief review, concepts relevant
to family systems theory are integrated in a conceptual
framework that is basis for generating the proposed model
and hypotheses for this study. In Chapter III the methods
of research used to conduct this study are described. Data
and analyses are presented in Chapter IV. Finally, in
Chapter V a summary and discussion of the findings,
implications for family therapy, and directions for future

research are discussed.

CHAPTER II

REVIEW OF RESEARCH AND THEORETICAL FRAMEWORK

W
The purpose of this chapter will be to present a

selected review of the current research on cancer care and
the responses of both the patient and family caregiver.
Following a description of the patient's need for care and
the caregiver's involvement with that care, a family systems
perspective will be employed to frame the issues of patient
and caregiver emotional responses to the cancer experience,
as well as caregivers' burden responses to the role of
caregiving. The ways in which family caregivers are
involved with and affected by the needs of family members
with cancer, as well as the occurrence of depression among
cancer patients and caregivers, will be surveyed in the
following review of research.

In the following review of research, caregiver may be
concurrently defined as 'family.‘ Caregiver burden
responses are considered in light of the general caregiving
research since very little is known about the burdens
specific to cancer care.

Typically, family members provide over eighty percent

of the necessary in-home care and health services (Hooyman &

10

11
Lustbader, 1986). The types of assistance that families
provide range from frequent or episodic supportive contact
to more routine aid for a family member's instrumental and
physical daily living needs. While the stage and site of
the disease typically determines the amount and degree of
needed assistance, family response to diagnosis and
treatment information seems to vary and often determines
their plan of action (Gotay, 1985; Welch, 1980). The
services that families provide for cancer patients, the care
needs they meet, and the stresses they experience in
shouldering a variety of family roles are of crucial
importance to the total care picture when considering the
adequacy of cancer care delivery (Oberst, Thomas, Gass, &
Ward, 1989; Oberst & Scott, 1988; Northouse & Northouse,
1987).

Cassileth (1979) notes that families serve as primary
sources of support for patients, and also as important
resources to health care professionals. Family members
often act to coordinate the necessary contacts with
appropriate care agencies, as well as to monitor and carry
information about the patient's day to day physical and
psychological status. Researchers have attempted to examine
and describe the process of care in terms of the emotional
strain and psychosocial needs of family members in cancer
care situations (Parkes, 1975): the informational needs
(Googe & Varrichio, 1981), as well as the educational and

support needs of caregivers (Skorupka & Bohnet, 1982).

12

Other studies have indicated that spouse caregivers
report feeling alone and stressed (Cooper, 1984), needing
extended family support (Chekryn, 1984), or experiencing
"exhaustion crisis" (Oberst & James, 1983). Whether these
types of strains lead to caregivers' feelings of burden in
cancer caregiving is not clearly known, but would seem
likely to contribute to such negative outcomes as impacts
on health, daily schedule and finances. Also, other
outcomes of care have been studied relative to the negative
impact on the family caregiver's work role (Wellisch,
Jamison, & Pasnau, 1978), the physical demands of care
(Lewis, Ellison, Woods, 1985) and other reactions to
lifestyle changes (Welch, 1981: Gotay, 1984) associated the
impact of cancer on the family.

Researchers have shown that families want to be
involved in the process of home care for a family member
with cancer (Vess, Moreland, Schwebel, & Kraut, 1988;
Edstrom & Miller, 1981) but often lack the skills necessary
to provide care, and are ill prepared for the impact that
cancer and home care may entail (Holmes, 1987; Krouse &
Krouse, 1982). Most importantly, families want to know how
to effectively provide the needed care and comfort for the
patient (Skorupka and Bohnet, 1982). In Grobe, Ahmann, &
Ilstrup (1982), families expressed a desire to be involved
in such activities as monitoring symptoms and progress,
basic caring for activities, providing equipment, and

transporting their family member to cancer treatments and

13
general transportation. Welch (1981) found that families
often thought that formal supportive services were
inadequate in meeting the care needs of the family. With
this brief review, a portrait emerges as to the tasks and
activities which families are engaged in providing for a
family member with cancer.

The negative impacts of care upon the primary
caregivers are generally conceptualized in terms of the role
strains associated with living and coping with the physical
demands of direct care, financial strain, medical
emergencies, and the psychological and social consequences
of providing emotional support for the patient and perhaps
other family members. Very few researchers have
specifically examined the impacts of family caregiving for
cancer patients, therefore only a brief review of the cancer
caregiver stress and burden is available. A description of
the more general family caregiving literature (i.e for
elderly or Alzheimers' patients) will help to guide the
reader through the concepts under study.

The appraisal of stress associated with the emotional
strains and demands of physical care among the identified
primary caregiver, are noted in a few family cancer
investigations (Oberst & Scott, 1988: Oberst, Thomas, Gass,
and Ward, 1989; Northouse, 1988). Lewis (1986), in her
review of the impact of cancer on the family, identified
eleven major issues which have been empirically associated

with cancer families experiences. Issues relevant to the

14
present study include emotional strain and physical demands
of care, altered roles and family lifestyle, and financial
press.

The most commonly reported responses to caregiving are
the subjective burdens and negative psychological reactions.
In a study of caregivers for the physically impaired
elderly, depression was found to dominate perceptions of
burden across five dimensions (Stommel, Given & Given,
1990). Poulshock and Deimling (1984) argue that burden
should only refer to subjective perceptions of and reactions
to the experience of caregiving. Thus, the event activity
or circumstance of caregiving should be considered as a
burden only when the caregiver views it that way. Whether
depression is an antecedent emotional state or condition in
caregiving or if it emerges from the context of, is not
clear. Again, the present study posits depression as an
intervening emotional response to cancer, and not simply a
negative outcome of caregiving.

Specific predictors of caregiver burden have generally
focused on characteristics of the patients (i.e. physical
and cognitive health status), characteristics of the
caregiving process (i.e. activities of involvement and
available support), and characteristics of the caregiver
(i.e. sociodemographic background, attitudes, and health
status). While depression could be viewed as a burden or
outcome of caregiving, it is difficult to assess when or if

depression emerges from involvement in the caregiving

15
environment because no studies exist which document pre-
caregiving measures.

There is a great deal of inconsistency therefore, in
the identification of relevant predictors for caregiver
burden responses. Whether those same burden models of
caregiving hold for cancer families is unknown. Questions
yet remain as to whether a caregiver burden framework
developed for the elderly is appropriate for the cancer care
experience. This exploratory study aims to shed light on
that question.

Because of its symbolic meaning, as well as its threat
of dire physical consequences, cancer can impose major
changes in the family role structure, unsettle the emotional
balance, and disrupt the family's patterns of interaction
and relationships (Vess, Moreland & Schwebel, 1985) Thus,
generalizations about the impact of cancer and the responses
of family members remain quite tenuous. There are, as yet,
no studies that have examined the family reactions across
different disease attributes, or conclusively linked patient
specific disease characteristics with particular family
strains or burden reactions.

There are many important psychosocial factors in the
complex web of family relations and influences which could
be examined to understand patient and family caregivers'
response to cancer (Spiegel, Bloom, & Gottheil, 1983). The
role of depression, however, seems most compelling in terms

of its influence in the potential responses to the event of

16
cancer as well as the role of care (Farber, Weinerman, &
Kuypers, 1984; Friedenbergs, Gordon, Hibbard, Levine, Wolf,
& Diller, 1981-82).

Depression among cancer patients has been a critical
variable under study in several psychosocial oncology
investigations (Frank-Stromberg & Wright, 1984; Plumb &
Holland, 1977). Goldberg & Cullen (1985) examined the
evidence concerning psychosocial adjustment in cancer and
noted certain specific cancer sites which seem to be
associated with psychiatric problems that have a biological
basis. Holland (1987) reported depression as the most
common psychiatric problem encountered among cancer
patients. In a study examining the prevalence of
psychiatric disorders among cancer patients (Derogatis,
Morrow, Fetting, Perman, Piasetsky, Schmale, et. al., 1983),
eighty-five percent of the cancer patients who received a
DSM-III diagnosis evidenced some type of psychiatric
disorder with depression or anxiety as the major symptom,
with the majority of those diagnoses (68%) constituting
adjustment disorder related to the cancer. Thus, it appears
likely that patients' depression would conceivably influence
caregivers' mental health and their response to the role of
care.

Oberst & Scott (1988) examined the temporal facets of
patient and spouse responses to hospital discharge after
cancer surgery and adjustment at home. They found that

spouse anxiety and distress persisted or was elevated over

17
time, well after the patient's mood states and adjustment
patterns had stabilized. Other researchers (Cassileth,
Lusk, Brown, & Cross, 1985) have documented similarity in
the mood state patterns of cancer patients and their
attending relative seen in an outpatient clinic. Patients
and family members showed the greatest mood disturbance when
they received palliative, as opposed to active treatment.
Also, complete bedridden status was associated with greater
mood disturbance in patients and their matched relative.
The finding of functional impairment and parallel mood
disturbance provide some support for the direction for the
present research and underscore the importance of examining
the influence of mood states (i.e. depression) in cancer
patients and family members. Determining the mutuality of
patients' and caregivers' depression is not possible in this
study.

There are, of course, significant clinical differences
between reactive and major depression, and between
psychogenic and biogenic causes of depression. Indeed, some
degree of an acute stress response is normal to the cancer
patient. Snaith (1988) describes the states of grief and
loss, hopelessness, or demoralization in the face of illness
and adversity, and frustration at one's thwarted ambition,
and the like as likely psychogenic causes of depression. He
suggests these should not be viewed as homogeneous disorders
of mood, but rather as reaction and response to the

diagnosis and threat of life disruption. Since the focus of

18
this research is not directed to identifying the causes or
concomitant influences on depression, the present review is
restricted to noting the relevance of depression as an
intervening and more enduring variable in the causal path
model of caregiver burdens.

Goldberg (1988) notes depression as one of the two most
common psychiatric disorders in cancer patients (delirium.
being the other), and delineates the medical basis for
evaluating depression and psychosocial distress. He notes
that the physical consequences of cancer and treatment can
account for disturbance of mood and thoughts that are often
indistinguishable from other concomitant sources of distress
(e.g. psychological, social, or existentially mediated).
Thus, with regard to the patient, it becomes difficult to
disentangle what might be biologically based depression and
what is reactive despair from the complex web of antecedent
physical assault of cancer and the concomitant psychosocial
and emotional experiences. For example, an aggressive
chemotherapy regime could impact a patient‘s mental state
thus intensifying their concern or perhaps despair over job
loss, family role disruption, or a host of other social
impacts. Patients' mental state of depression is then
posited to strongly influence the caregivers' overall
emotional response to the cancer experience.

In summary, there is empirical evidence to suggest
that cancer affects more than just the person with the

diagnosis, and that in family caregiving situations the

19
primary care provider is often at risk to be negatively
affected by responses of the family member with cancer, and
by the demands of care. Exactly what factors account for
the contingencies of this relationship is not clearly
understood in the cancer family care process.

In spite of the considerable abundance of research on
the attributions of family caregiver burden and the impact
of caring for the dependent elderly and family members with
Alzheimer's Disease, very little is known about the
reactions and burden responses to cancer caregiving. It
seems plausible that reports of caregivers' experience of
burden or impact will be colored by their subjective
perceptions and personal interpretations of the contexts of
care (Poulshock & Deimling, 1984). Thus, the cancer
caregiver's report of burdens will be influenced by the
contextual issues of the cancer experience and by the
caregiving responsibilities with which they are faced.

It was the purpose of the preceding review to highlight
some of the cancer family care issues and concepts and to
provide a background from which the present analysis will
emerge. A family systems framework will serve to guide the
development of the proposed path model of cancer caregiver

burden responses (See Figure 1).

W
The health, age, sex, relationship, and a whole host of

characteristics can affect the outcome of reactions to the

20
caregiving role. Thus, while family caregiving is more or
less viewed as a normative family process (Brody, 1985), it
remains poorly understood. There are few descriptions of
the home care process and caregiver reactions in the cancer
literature. Investigation of the impact of family
caregiving for cancer patients has only recently become a
salient focus in cancer research. Cassileth, et. al. (1985)
assert the impact of cancer on the family is of major
concern in psychosocial oncology, and also in clinical
family therapy practice (Quinn & Herndon, 1986).

Most of the current research is lacking a theoretical
model which can link the study results beyond descriptive
analysis into a broader interpretive structure (Weisman,
1979). Vess, et. al., (1985), examined the impact of cancer
on the family's psychosocial environment and the
reallocation of roles from a developmental perspective.
However, such an explicit conceptual framework is not common
in current psychosocial cancer research. The present
research proposes a broadly interpreted general systems
family framework (Broderick & Smith, 1979). Since the
conceptualized variables place emphasis on caregivers'
perceptions and self-appraisal of the burden impacts
associated with the role of caregiving, elements of role
strain and interactionist perspective (Burr, Leigh, Day, &
Constantine, 1979) are encompassed in the framework.

A general systems framework provides a conceptual

linkage by focusing the present research questions on the

21
influence of the cancer patient's functional status and
depression on the spouse caregivers' depression and their
responses to the impact of the caregiving role upon their
health, daily schedule, and finances. Characterized as a
stress event (Giacquinta, 1977), cancer in the family can
produce a change in the system's boundaries, structure,
goals, purposes, rules, and values. A family systems
perspective then posits that a change in one component of
the family will affect a change in that entire family system
(Minuchin, 1974).

Specific questions and hypotheses to be examined in
this exploratory research are related to the theoretical as
follows. The first question concerns the description of the
specific cancer experience variables and the relationship
between that experience and patient's and their caregiving
spouse's depression. The second question relates to the
cancer experience variables, patient's and caregiver's
depression, and the potential contributions each make to the
prediction of burden responses among spouses' providing
care. The specific caregiver burden responses to be
examined are the perceived impacts of caregiving on
caregivers' health, daily schedule, and finances. The
focus of this research is to propose a predictive model of
caregiver burden responses. Specifically, the analysis may
clearly depict the causal relationship of patient's mental

state of depression upon caregiver's depression and,

22
consequently, the effect of depression upon the specific

caregiver burden responses to cancer care.

es H 0 ese

Based on the review of literature and predicted path
diagram, the following hypotheses are proposed for the
present analysis of husbands and wives. The researcher
wanted all subjects to be from comparable dyadic
relationships, thus, all of the following hypotheses will
center on the variable relationships among husbands and
wives.

The first hypothesis is intended to examine the causal
relationship between the cancer experience and caregiver
burden. Six sub-hypotheses were developed to examine the
relative importance of the cancer care experience variables
on the three separate caregiver burden responses.
Theoretically, the following hypotheses relate to cancer as
an input in the family system which the well-spouse must
respond to and accommodate the role as family caregiver.
The first set of sub-hypotheses follows from the view that
as duration of care persists over time caregivers' will

perceive greater burdens.

1.3: There is a positive causal relationship
between duration of care in months and the burden response
(A) impact on health, (B) impact on daily schedule, and (C)
impact on finances.

23
Similarly, the next series of sub-hypotheses posits
that as patients' functional status declines, a probable
need for more care would result, and caregivers' would

perceive greater burdens.

e ' : There is a positive causal relationship

between patients' functional dependency and caregiver's
burden response (A) impact on health, (B) impact on daily
schedule, and (C) impact on finances.

The second hypothesis proposed emerges from the
theoretical consideration of the systemic assault of the
cancer experience on patients' emotional mood state. In
order to assess the extent and strength of the relationship
between the selected characteristics of the cancer
experience and patient depression, two sub-hypotheses were
developed. Again, these hypotheses are based on the view
that as duration of care (or need for care as result of the
cancer disease) persists over time patients' will likely
become or remain depressed, and that as patients' functional
status declines they are more likely to be depressed. The

two sub-hypotheses are stated as follows:

fixpg£n§§i§_24= There is a positive causal relationship
between the duration of care and patients' depression.

H!EQLD§§1§.2¢: There is a positive causal relationship
between patients' functional dependency and patients'
depression.

24
The systems theory framework put forth in this study

holds that a change in one component of the system will
affect a change in the entire system. Thus, if cancer
patients are prone to depressive moods, for whatever
reasons, then their mood state would consequently affect the
mental health of their caregiving spouse. The third
hypothesis, proposed to examine the correlation between

patients' and caregivers' depression, is stated as follows:

gyneeneeie_;: There is a positive causal relationship
between patients' depression and caregivers' depression.

The effect of caregiver's depressed mood state on their
perception of the burdens associated with cancer care is the
focus of the next hypothesis. Caregivers are more likely to
perceive greater burden as a result of their depressed mood.
Thus, in a systems theory framework, the final hypothesis
emerges from the influence of the cancer experience upon the
patients' and then caregivers' mental health will influence
caregivers' perception of burdens. The fourth and final
hypothesis then is designed to examine the relationship
between caregivers' depression and the three specific

caregiver burden responses.

Hypotnesie 4: There is a positive causal relationship
between caregiver depression and the specific caregiver
burden responses of the impact on: A) health, B) schedule,
and C) finances.

25
Based upon the selected review of research and
conceptual framework, the above stated hypotheses emerge
from the proposed path diagram. The research design,
methods, and data analysis strategy for the present stated

hypotheses will be discussed in the following chapter.

CHAPTER III

METHODS AND PROCEDURE

Description of the methods, including study design,
sampling, data collection, and analysis strategy are

presented in this chapter.

W

In order to describe the family cancer experience and
determine the causal relationships between patient
depression, caregiver depression, and caregiver burden
responses of married cancer patient/caregiver dyads, a
secondary analysis of cross-sectional data was conducted.
This study was carried out in conjunction with, and support
from, Dr. Barbara A. Given and Dr. Charles W. Given,
Principal Investigators for the following grants: NCNR,
"Family Homecare for Cancer - A Community-Based Model" (#1
R01 NR01915-01) and ACS, "Family Home Care For Cancer
Patients" (#PBR-32). Both are one year longitudinal panel
studies.

The data reported here are taken from the first wave of
a panel study of family caregivers of new and recurrent

cancer patients.

26

27
Men

The researchers identified available and willing family
cancer dyads for the proposed study design resulting in a
convenience sample. Patients and their family caregivers
were introduced to the study by nurses in community-based
cancer care settings. Subjects were recruited via card back
system and then screened according to clinical, cancer
specific, and functional health criteria. The cancer
patient/caregiver dyads were selected for screening using
the following criteria: adult patients between 20 and 85
years of age; diagnosed with new or recurrent solid tumor
or lymphoma: symptomatic or with some physical
impairment/limitation in Activities of Daily Living or
Instrumental Activities of Daily Living: and currently
under medical treatment. The primary family caregiver was
the person identified as the individual who provides the
most care in the home for their family member with cancer.

Subject rights were guaranteed the protection accorded
them under the human subjects review approval of the study
from which it is drawn from. This included voluntary
participation, withdrawal from the study at any time without
penalty, and all information obtained to be held in
strictest confidence, with no individual identity linked in
any way to the results.

Two hundred thirty-eight married cancer-caregiver dyads
were drawn from the first wave data set. Analysis was

conducted with 190 cases for which responses of both the

28
patient and caregiver were obtained. Spouse caregivers were
self-identified as the person who provided the most care.

Cancer patient selection is described below.

W

Data from 238 cancer patients and their caregiving
spouses residing in the lower peninsula of Michigan were
collected. The age of the patient and caregiver respondents
ranged from 22 to 83 years old, with a mean age of 58 and 57
years for patients and caregivers, respectively (See Table 1
for sample description). The primary sites of cancer in
this sample (e.g. female breast, colon-rectum, lung, and
lymphoma) reflect the highest prevalence of reported cancer
sites elsewhere in the U.S. (ACS, 1989).

The cancer patient/caregivers were selected for the
study by use of a screening instrument which employed the
following inclusion criteria: patients diagnosed with new
or recurrent cancer within the last two years; some
physical impairment/limitation, as evidenced by an activity
performance status rating of 2 or more: and the caregiver's
report of the patient's symptoms which necessitate his/her
receipt of care and assistance in the home.

mm

Data for the major variables under study were collected
via a self-administered mailed questionnaire booklet to each
patient and caregiver dyad. Data on the patient's

diagnosis, tumor site, and the duration of caregiving were

29
gathered from the caregiver over the telephone using a study

eligibility screening instrument.

W

In the following discussion, the major variables under
study will be described and operationally defined.
Depression, as noted previously, is conceptually defined as
a non-clinical and primarily psychogenic state, with
cognitive, affective, and behavioral depressive features.
Patient and caregiver depression were measured by the (CES-
D) scale (Radloff, 1977).

The CES-D is a 20 item, Likert-type, self-report scale
designed to assess depressive symptoms experienced over the
past month. (Appendix A). This scale has been widely used
as a state measure of current emotional distress and
depressive mood (symptomatology) within a general, non-
psychiatric population (Devins & Orme, 1985). Found to
correlate only modestly with clinical diagnoses of
depression, this instrument can be used for screening
persons for symptomatology related to depression (Meyers and
Weisman, 1980), or at risk for clinical depression (Roberts
and Vernon, 1983).

Responses are measured on a four point scale describing
the frequency of occurrence (i.e. "almost all of the time"
to "rarely or none of the time") for each depressive symptom
item. The scale has shown consistently good psychometric

properties, with moderate convergent validity with other

30
depression measures. High internal consistency has been
noted in previous reports (Radloff, 1977; 1986), with
Cronbach's alphas reported between .84 to .90.
Reliabilities in this study show similarly high alpha
coefficients (.91 to .88, for caregivers and patients,
respectively) in the total sample (N=268) from which the
present study subjects were drawn. Reliability and validity
will be discussed further in the following chapter.

The other major outcome variables are collectively
described as Caregiver Burden Responses, also noted
previously, and conceptually defined as caregivers' self
appraised impacts of care on: A) physical health; B) daily
schedule: and C) finances. Operationally, these burden
responses to caregiving will be measured by the Caregiver
Responses Inventory (Given & Given, 1987). (Appendix B)
This inventory was constructed to tap the subjective impact
reactions to the specific processes and situations of
caregiving. An initial pool of 111 items was reduced to 40
items with 5-point Likert response scales. From these items
five distinct subscales were found measuring perceived
caregiving burdens in the following areas: impact on
health, impact on schedule, impact on finances, feelings of
family abandonment, and a sense of entrapment. The factor
structure of these scales has been confirmed on two
additional independent samples of caregivers of the
physically impaired elderly and of Alzheimer's patients.

All five scales are reliable with Cronbach's alpha ranging

31
from .76 to .84 in the present sample. The current research
focused only on the three burden subscales of impact (e.g.
health, schedule, and finances) because of their empirical
relevance to possible interventions.

The cancer care experience is operationalized by asking
two specific questions which address first, the length of
time care has been provided to the patient and second, the
functional physical dependency of the patient. Duration of
care is measured by asking the caregiver to report in months
"how long have you been providing care?" for this cancer
patient.

Patients' functional physical dependency is measured by
the Karnofsky Performance Status Scale (KPS) (Appendix C).
The KPS is a multidimensional construct reduced to a
unidimensional one. This scale, a standard measure in
oncology research and practice, is a functional performance
tool that appraises a patient's symptomatology and ability
to carry on normal activity (Heinrich & Schag, 1987). It is
the most widely used method of quantifying functional status
of cancer patients. It is reported to possess good
construct validity and has been compared well with Katz's
ADL Index and other quality of life measures. High inter-
rater reliability (.97) was found among non-clinicians using
the scale.

In this study, the measure was administered over the
telephone by asking the caregivers to report, based on the

five-point scale, the gross functional performance or

32

activity status. Conceptually, this measure may evidence
the patient's need for care. 1

Other data, such as the type of cancer and whether
patients were undergoing active cancer treatments, were
collected to further illustrate the cancer experience and
patients' likely need for care. However, these are nominal
data and were not employed in the path model but only
reported here as background variables and descriptors of the
sample under study.

In the next section the strategy for analyzing the

proposed path model is discussed.

mm

The researcher analyzed the variables under study by
the following methods. The first stage of the analysis
involved calculating the descriptive statistics which
describe the characteristics of the sample, the responses on
the major variables (means, standard deviations), frequency
distributions, psychometric properties of the major scales
used, and the inter-correlations of the major variables and
outcome measures utilizing Pearson product-moment
correlation coefficient.

The second stage of analysis employed a recursive path
model, derived from the theoretical considerations outlined
earlier. The proposed path model presents a graphic
depiction of the direct and indirect effects of the

variables on the outcome measures and predictive inferences

33
related to the strength of the effects of the variables.
While the path model yields a method for decomposing and
interpreting linear relationships among a set of variables
that are assumed to have a prior causal ordering, an initial
step in the path analysis was to indicate the causal
ordering of the variables. The focus of such causal
ordering was as follows: a) the selected cancer experience
variables and depression; b) depression and the specific
caregiver burden reactions; and c) the cancer experience
and caregiver burden reactions. The statistical technique
employed for the path analysis was ordinary-least-squares
regression.

Estimates of the standard error of the observed sample
correlation allowed the researcher to judge if the predicted
population correlation could have generated the observed
sample correlation. This allowed one to test the goodness-
of-fit of the model for predicting cancer caregivers' burden
responses. Presentation of data and analysis will be

examined next.

CHAPTER IV

PRESENTATION OF DATA AND ANALYSIS

W

In this chapter data analysis and interpretation of the
study findings are presented. Reliability measures and
descriptive statistics are presented for the measures and
subscales used in this research. Data relevant to the study
questions and hypotheses are presented to examine the
strength of the relationships between the variables.
Regression equations were formulated to examine the variance
that can be explained and to derive the standardized beta
coefficients and depict the direct and indirect effects of

the variables in the path model.

W
The sample consisted of 238 caregivers and their
spouses with cancer. Table 1 presents the socio-demographic
characteristics of the subjects, including caregiver age,
gender, race, and eduction. Table 2 presents patient age
and gender. The subject characteristics are not entered
into the later analysis, but are only presented for

description of the study sample.

34

35

1en1e_1. Caregiver Sociodemographic Data

 

W Sade: u 8
i = 57 Years Male 98 41
SD = 11 Female 140 59
Min. = 23
Max. = 81 Total 238 100
31 i
RAGE
Caucasian 229 96
Black 5 2
Hispanic 1 .5
Other 1 .5
Total 236 99

Missing Cases = 2

mm H 36.
Grade School or Less 11 5
Some High School 31 13
High School Graduate 83 35
Some College 72 30
College Graduate 20 8
Graduate/Prof. Degree 21 9
Total 238 100

 

 

 

The sample of caregivers consist of 140 women and 98
men, with a mean age of 57 years (S.D. = 11). Ninety-six
percent of these caregivers are white, 86% have graduated
from high school, and 48% have some college or higher
education experience. Since only spouse caregivers are
included, patient gender complements that of the caregivers.

Mean age of patients is more than 58 years (S.D. = 11).

36

I§h12_2- Patient Sociodemographic Data

 

Pagieng Age Gender 3 i

X = 58 Male 140 59
SD = 11 Female 98 41
Min. = 22 ___

Max. = 83 Total 238 100

 

 

 

Table 3a and 3b presents cancer specific data regarding
primary tumor site and primary mode of treatment. The
cancer related data reveal the most frequent tumor sites to
be the female breast (23%), lung (16%), lymphoma (13%), and
colon-rectal (12%). Most patients were undergoing active
treatment for their cancer, the most frequent being

chemotherapy (83%).

Teble 3a. Cancer Specific Data

 

 

 

 

WI! 3
Bladder 5 2
Breast 55 23
Colon/Rectal 28 12
Gastro-intestinal 23 10
Gynecological 9 4
Lung 39 16
Prostate 16 7
Lymphoma 31 13
Head/Neck 6 2
Other 26 11
Total 238 100
's ' e ' nce

The mean, standard deviation, and frequency

distribution of the two identified independent variables,

37
patient physical functional dependency and the duration of

care, are presented in Table 4 and 5, respectively.

Ienle_3b. Cancer Specific Data

 

IIQQLEQDL N i
Chemotherapy 196 83
Radiation 9 4
Hormone 5 2
Surgery/Other 12 5
None 14 6

Total 236 99

Missing Cases = 2

 

 

 

Table 4 presents data for the Karnofsky Performance
Scale. Mean scores for this measure of patients' physical
functional dependency, is 2.44 (S.D. = .87) on a 5 point
scale. Only 10% of the caregivers reported their patients
were fully active in all activities. Fifty-two percent of
the caregivers said their patient was fully active in only
light activities, indicating some functional impairment,
while 23% say their patient is symptomatic but in bed less
than half the day, and 15% report physical functional
limitation to the extent that their patient is in bed more
than half the day.

Table 5 shows data on caregivers' duration of
involvement in care, reported in months, and reveals that

61% have only been providing care for twelve months or less.

38
The mean number of months of duration in care is nearly 25
(S.D. = 47), but the median duration of care is 9 months.

Tanle 4. Characteristics of the Cancer Care Experience1

 

 

Actixitx_§tatu§* Frequency £22223:
(1) Fully active 24 10
- all activities
(2) Fully active 123 52
- light activities
(3) In bed < 50% of day 54 23
(4) In bed > 50% of day 36 15
(5) Bedridden/unable to 1 --
care for self
Total 219 100
Mean = 2.44 8.0. = .88

* Karnofsky Performance Scale

 

The case reporting 480 months duration was excluded
from the analysis of the major variables. The groupings of
months in Table 5 is presented for descriptive purposes
only. Analysis for was performed on duration as a

continuous interval variable.

Ianle . Characteristics of the Cancer Care Experience2

 

 

Duratign_2f_sars (in months)

E 3

1-3 45 19

4-6 53 22

7-9 29 12

10-12 15 6 _

13-24 30 14 x = 25 months
25-36 21 10 Med. = 9 months
37-48 8 4 SD = 47
49-60 8 4 Min. = 1
61-62 2 -- Max. = 480
73 or more 18 7

Total 237 99. Missing cases = 1

 

 

 

39
5 ° '0 ° ' ' e 5 es

Reliability coefficients for the CES-D 20 depression
measure and caregiver burden subscales are presented in
Table 6. Reliability refers to the internal consistency of
the scale and the extent to which all the items in the scale
are equally good indicators of the concept or attribute in
question. It is generally understood, the higher the
coefficient alpha score the stronger the internal
consistency of the scale, meaning that the items in the
scale are consistently measuring the same concept. The
alpha coefficient scores presented in Table 6 reflect
acceptable to high levels of internal consistency of the
measures. When a scale is shown to possess repeatedly high
reliability scores it possesses reproducibility and scale

scores are less likely to be influenced by error.

Ianle g. Reliabilities of Caregiver Burden Subscales and
(CES-D) Depression Measures (N = 268)

 

Seals W W
Caregiver Depression 20 .91
Patient Depression 20 .88

Impact on Health 5
Impact on Schedule 5 .84
Impact on Finances 3

 

 

 

Descriptive statistics for the depression measures and
he caregiver burden subscales are presented in Table 7.

Examination of the caregiver depression measure reveals a

4O
‘moderate positive skew and somewhat higher kurtosis,
reflecting a slightly asymmetric and more peaked
distribution than that of a normal curve. While most
caregiver depression scores cluster around the sample mean,

a few have extreme high values.

WW
Analysis of the path model is presented in two stages.
First, the bivariate correlations among all variables are
presented. Second, three recursive path models will be
presented to depict each of the three caregiver burden
outcomes as the dependent variable: impact on health, impact

on schedule, and impact on finances.

Table 7. Scale Means for Patient and Caregiver Depression
and Caregiver Burden Responses (N = 190)

 

l. 1.71 .41 1.00 3.20 .648 .591 192
2. 1.72 .42 1.00 3.55 .911 1.993 220
3. 2.07 .62 1.00 4.20 .619 .870 220
4. 2.93 .94 1.00 5.00 -.051. -.669 220
5. 2.33 .86 1.00 5.00 .823 .415 220
1. Patient Depression (1-4)
2. Caregiver Depression (1-4)
3. Impact on Health (1-5)
4. Impact on Schedule (1-5)
5. Impact on Finances (1-5)

 

 

 

Table 8 contains the Pearson r linear correlation

coefficients for all predictor and outcome variables in the

41
proposed path model. Since duration of care did not
correlate with the other predictor variable, patient
performance status (r = -.10), the intervening variables of
patient and caregiver depression (r = .04, -.03,
respectively), nor with any of the three burden outcome
measures of impact on health, schedule, and finances (r =
.09, .05, 06, respectively), it was not expected to
contribute significantly to the model explaining caregiver
burden reactions. Thus, duration of care was dropped from

the subsequent path models.

Ieble . Pearson Linear Correlation Coefficients of All
Predictor and Outcome Variables (N = 190)

 

 

1 2 3 4 5 6 7
1 1.0
2 -.098 1.0
3 .040 .289***1.0
4 -.025 .224*** .399*** 1.0
5 .088 .151* .304*** .547*** 1.0
6 .047 .424*** .481*** .518*** .567*** 1.0
7 .055 .062 .233*** .317*** .273*** .196** 1.0

 

1. Duration of Care

2. Functional Status (Karnofsky)
3. Patient Depression

4. Caregiver Depression

5. Impact on Health

6. Impact on Schedule

7. Impact on Finances

 

* p 5 .05
** p g .01
*** p g .001

 

 

42

The finding of non-significant correlations for the
duration of care variable may have resulted from possible
problems in measurement, likely to result from respondents
having to remember an often ill-defined onset of illness.
The finding of non-significance and subsequent decision to
exclude this variable from the model will be discussed
further in the next chapter.

Patients' functional status is shown to moderately
correlate with patients' reported levels of depression (p <
.001). Correlations among the dependent variables indicate
that the impact of caregiving on health, schedule and
finances are highly related to each other. They also
correlate with patients' and caregivers' levels of
depression. Thus, assuming that depression is the more
enduring state, these correlations appear to suggest that
caregivers' level of depression may influence their
perception of how caregiving impacts their lives.

Two of the three caregiver burden responses (excluding
finances) appear to be moderately related (p 5 .05) to the
measures of patients' physical functional status and their
reported level of depression. Whether these effects are
direct or mediated through caregivers' affective state (i.e.
their own level of depression) will be explored through the
path models. The recursive path diagram (Figure 2) depicts
the proposed model and how the variables are thought to be
related to each other, with caregiver burden reactions

conceptualized as the outcomes. Three separate path figures

43
will be analyzed to examine the three specific caregiver
burden reactions (i.e. impact on caregiver health, schedule,

and finances, respectively).

 

 

  
      
  

 

 

 

IMPACT ON:

 

PATIENT FUNCTIONAL. 1. HEALTH
STATUS ‘ 2. SCHEDULE
(KARNOFSKY) 3. FINANCES

 

 

 

 

 

 

b2 ATIENT
EPRESSION

 

 

 

 

Eignze_g. Recursive Path Diagram of Caregiver Burden

The proposed path model was generated to depict the
relationships between the variables suggested by the
research questions. Patient depression is treated as an
intervening variable, influenced by patients' physical
functional status. Caregiver depression is also
conceptualized as intervening, influenced by patients'
functional dependency and by patients' depression. All the
proposed paths are recursive. Which is to say, that all
causal impacts are assumed to be unidirectional and without

feedback loops (Land, 1969).

44
R ' t'o P An sis
The path coefficients are derived using a three

equation model for each of the three caregiver burden

 

impacts.
1. Pdep = b2(pfs)
2. Cdep = c1(pdep) + c2(pfs)

3. Impam,c = d1(cdep) + d2(pdep) + d3 (pfs)

 

 

 

Eignne 3. Regression Equations for Path Coefficients

The first two regression equations calculate the beta
coefficients and amount of variance explained in the
intervening variables. The third regression computes the
proposed direct effects of all predictor and intervening
variables on the caregiver burden outcome variables. In
Figure 3, the three regression equations computed in the
path are presented.

Results of the first stated regression equation are
presented in Table 9. The first regression analysis
revealed that patient functional status was significant (p <
.001) in explaining the relatively small proportion (R2==
.08) of the variance found in patient depression. Recall
that in Table 8 the Pearson Correlation coefficients for the
two variables were only moderately correlated (.289, p <
.001).

Table 10 presents results of the second regression run

with patient functional status and patient depression

45
entered in the equation to predict caregiver depression.

Table 10 reveals a significant beta coefficient, .358

(p.<.001), for patient depression. Nearly 17 percent (R2
.1684) of the variance in caregiver depression can be
explained and accounted for by the variable patient

depression.

Ienle . Regression of Patient Functional Status with
Dependent Variable Patient Depression

 

 

Pgeeieto; Veziable ESE; I Signif
Patient Functional Status .28942 .0000 *
Multiple R .2894
R2 .0333
F 17.2785
F Significance .0000

 

 

 

The results of the first two regression equations are
reflected in the beta coefficients for patient functional

status and patient depression.

Tenle 10. Regression of Patient Functional Status and
Patient Depression with Dependent Variable
Caregiver Depression

 

 

W Bess . Tits—15' ni
Patient Functional Status .12203 .0807
Patient Depression .35805 .0000*

Mpltiple R .4103

R .1684

F 19.0325

F Significance 40000

 

 

 

46

The beta coefficients for patient depression on
patient functional status (.289, in Table 9) and for
caregiver depression on patient depression (.358, in Table
10) will be displayed in each of the three path figures
depicting their influence on the specific caregiver burden,
impact on health (Figure 4).

Table 11a presents the results of the third regression
equation examining the amount of variance explained by the
predictor and intervening variables for the caregiver burden
response of impact on health. When entering the predictors
of patient functional status and the intervening variables
of patient and caregiver depression into the regression run,
it appears that only caregiver depression (.503) is
significant (p < .001) in explaining the variation in the

burden response impact on caregiver health.

Ienle_11e. Regression of Patient Functional Status, Duration
of Care, Patient Depression and Caregiver
Depression with Dependent Variable Impact on

 

 

Health
W m: uisnif
Patient Functional Status .00821 .8984
Patient Depression .10296 .1315
Caregiver Depression .50257 .0000*

Mpltiple R .5534

R .3063

F 27.5230

F Significance .0000*

 

 

 

47

The beta coefficient (.503) for caregiver depression
is then added to the path diagram depicting the causal
relationships of all variables with impact on health. As
shown in Figure 4, there are no direct effects of patient
functional status on impact on caregiver health, but only
the indirect effects through patient depression and
caregiver depression. Nor are there direct effects of
patient depression on caregivers' perceived impact on
health. The reader should note that only significant beta
coefficients are depicted. Nearly thirty-one percent of the
variance (R?== .306) in the dependent variable, impact on

health, can be explained in this path model.

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

E =.83
GIVER .503**

EPRESSION
PATIENT FUNCTIONA
DEPENDENCY .358** IMPACT A
(KARNOFSKY)

.289** ATIENT

EPRESSION

** < .01 IE for Impact on Health = 30.6%
Fignze 4. Path Model Coefficients for Impact on Health

The next phase of the path analysis then is to

determine if the causal model can help to better predict the

48
relationship between patient functional status and caregiver
impact on health. By taking into account the additional
factors of patient and caregiver depression, which were
posited to influence caregiver burdens, the path model
allows the researcher to determine the combined effects of
the variables on the ultimate dependent variable, impact on
health.

The combined indirect effects of the significant
variables in the path equation must be less than or equal to
the observed correlation (r = .151) between patient
functional status and impact on health, as expressed in the

following equation:

r 2 (d1x c1x b2) or, r (.151) 2 .052

 

(Recall) Figure;-

1. Pdep = b2(pfs)
2. Cdep = c1(pdep) + c2(pfs)
3. Impa = d1(cdep) + d2(pdep) + d3(pfs)

 

 

 

In the three-equation model for the impact on
caregiver health, patient functional status is only
indirectly related to caregivers' perceived impact on
health. This relationship is depicted by the multiplicative
term of (11 x c1 x b2, and amounts to .052. This is less than
“the observed correlation (r = .151) between patient
:functional status and the impact on caregiver health. In

carder to accept the model as reasonably fitting the observed

49

data, the standard error on the sample correlation was
computed. This allows one to test if the observed
correlation in the sample differs from the one hypothesized
on the basis of the path model to such a degree that it is
unlikely that the proposed path model fits the sample data.

The formula to test for a significant difference
between the observed and predicted r was calculated in the

following way:

observed sample r - hypothesized population r
or
(.151) - (.052)

SE .0719

The result, 1.38, indicates the observed correlation
is not significantly different (i.e. more than 2 standard
deviations away) from the correlation hypothesized for the
population, and the model predicting caregiver impact on
health can be accepted.

The second caregiver response, impact on schedule, was
then analyzed in the same manner as the impact on health.
Table 11b presents the regression results when all variables
are entered upon the second outcome measure.

As the findings from Table 11B show, there are a
:number of significant beta coefficients to be added in the

:second path figure (Figure 5) First, there appears to be a

50
significant (p < .0001) direct effect of patient functional
status on impact of caregivers' schedule (beta coefficient =
.269). Second, there is a significant effect of patient

depression (beta coefficient = .263).

Tabls_llb- Regression of Patient Functional Status, Patient
Depression and Caregiver Depression with Dependent
Variable Impact on Schedule

 

 

Pgedicto; Variebles Beta T Signif
Patient Functional Status .26890 .0000*
Patient Depression .26284 .0000*
Caregiver Depression .35340 .0000*

Multiple R .6501

R2 .4226

F 45.6273

F Significance .0000

 

 

 

The direct and indirect effects of patient functional
status on impact on schedule can be derived from Figure 5
which illustrates the direct paths (or direct causal
relationships) upon the second outcome, perceived impact on
caregivers' schedule.

As with the first path, the next phase of analysis
involved determining if the causal model can help to better
predict the relationship between the predictor variables and
caregiver impact on schedule. The combined effects of the
variables in the path equation must be less than or equal to
the observed correlation (r = .424) between patient

functional status and impact on schedule.

51

 

 

    

 

 

 

 

 

 

 

 

 

 

 

 

 

 

PATIENT FUNCTIONAL__e1529*
DEPENDENCY .358** IMPACT B
(KARNOFSKY)
.289** ATIENT .263**
DEPRESSION
** < .01 I? for Impact on Schedule = 42.2%
Eign;e_§. Path Model Coefficients for Impact on Schedule

By adding the combined effects of the variable beta
coefficients, the formula can be expressed in the following

equation:
r 2 d1+ (d2 x b2) + (d1x c1x b2) or, r (.424) 3 .382

Similarly with the test of the model for impact on
health, the standard error of the observed correlation was
computed. Again, the calculated standard error was entered

in the following equation:

sample r(.424) - hypothesized population r (.382)

SE r (.0659)

 

52

The result, .64, reveals no significant difference
(i.e more than 2 standard deviations away) from the
correlation found hypothesized for the population.
Therefore, based on the standard error estimate for the
predicted r, the path model predicting caregiver impact on
health can be accepted.

Table 11c presents the regression results for the
third outcome measure, impact on finances. As with the
regression results for impact on health, among all the
variables only caregiver depression appears to have a
significant effect in explaining the variance on the
caregiver burden of impact on finances.

.There is not a large amount of variation in the

dependent variable that can be explained by the other

variables entered in the equation (R2 .117, p < .001).
While caregivers' perceptions of the impact of care on
finances may be influenced by other relevant factors, such
as socioeconomic resources, it does appear that some effect
can be explained by their affective state of depression,
which in turn appears to be influenced by patient
depression.

The beta coefficient for caregiver depression (.275)
is then added to the next path diagram (Figure 6) as a
direct effect of caregiver depression on impact on finances
(and an indirect effect of patient depression). Figure 6

depicts the causal path of the variables on the third

outcome under investigation. Similar to the model for

53
impact on health, there are no apparent direct effects of
the predictor variables on impact on finances, but only the
indirect effects through the intervening variables of
caregiver and patient depression.
Tenle_11_. Regression of Patient Functional Status, Patient

Depression and Caregiver Depression with Dependent
Variable Impact on Finances

 

 

Patient Functional Status -.04132 .5709
Patient Depression .13766 .0746
Caregiver Depression .27472 .0003*

Multiple R .3417

R2 . 1167

F 8.1932

F Significance .0000

 

 

 

As with the two previous path models, the next phase
of analysis involved determining if the causal model can
help to better predict the relationship between the
predictor variables and caregiver impact on finances.

The combined effects of the variables in the path
equation must be less than or equal to the observed
correlation (r = .062) between patient functional status and
impact on finances. By adding the combined effects of the
variable beta coefficients, the formula can be expressed in

the following equation:

r 2 (d1 x c1x b2) or r (.062) 2 .028

54

 

 

 

 

 

 

 

 

 

 

 

 

E =.94
AREGIVER .275**
DEPRESSION
ATIENT FUNCTIONAL
EPENDENCY .358** IMPACT C
(KARNOFSKY)
.289** ATIENT
EPRESSION
** < .01 13 for Impact on Finances = 11.7%

 

 

 

Figure 6. Path Model Coefficients for Impact on Finances

Again, in order to test whether the model can help to
predict the depicted causal relationships in the population
involves the calculating the standard error score for the
predicted correlation. Thus, the following equation is
calculated to determine the standard error of the predicted

correlation, based on the model:

sample r (.062) - hypothesized population r (.028)

SE r (.0726)

The result, .47, indicates that the path model
prediction for the population correlation does not differ
significantly from the observed sample correlation.
However, since the observed correlation was not significant
at the .05 level, it is unlikely that the population

correlation differs from zero. Correspondingly, despite the

55
significance of individual paths in the model, the path
chain from patient functional status to impact on finances
does not differ from zero.

In review of the preceding analysis, recall that in
the path figure for impact on health, the path model
depicting the causal relationships provide the reader with
additional information not readily apparent by the Pearson r
correlation analysis. That is, that the predictor variables
do not by themselves explain a significant amount of the
variance in the outcomes. However, indirectly, through
patient depression and subsequently through caregiver
depression, they do color the perception of caregivers'
response in the impact of the care role on health.

As demonstrated in the preceding analysis, the impact
on schedule is directly affected by patient functional
status and also by patient depression. This seems fairly
logical since the patients' physical health and
psychological adjustment is likely to have implications for
the caregivers' involvement in care, thus causing some
impact on their daily schedule.

Table 12 presents a synopsis of the direct and
indirect effects for each of the outcome caregiver burden
response variables. As noted earlier, there are no apparent
direct effects of patient functional health on caregiver
impact on health or impact on finances, but the indirect
effects through patient depression and consequently through

caregiver depression. The hypothesized population

56
correlations for the model are presented as the total
effects. Standard error scores for the observed sample r

are also shown in Table 12.

Teble 12. Effects of Patient Functional Status on the
Caregiver Burden Response to Impact on Health,
Schedule, Finances

 

 

m Sebedulefimmses
DE --- .269 ---
IE 1 .052 .113 .028
TE2 .052 .382 .028
Observed r .151 .424 .062
SE Estimate3 1.38ns .64ns .47ns

 

1 - Indirect Effects via Patient and Caregiver Depression

2 - Hypothesized total effect based on Path Model

3 - Deviation of the observed from hypothesized effects in
terms of the standard error of the sample correlation

ns - non-significant

 

 

 

Table 13 presents a summary of the regression results
of all three outcome burden responses on the independent
variable, patient functional status, and the intervening
variables, patient depression and caregiver depression.

This table displays the beta coefficients for all variables,
including the predicted amount of variance that can be
explained by the independent and intervening variables for

each caregiver burden.

57

Table 1;. Summary Table of Caregiver Burden Responses
Regressed on Cancer Experience and Depression

 

 

 

 

 

 

Variables

DEZEND£!I_!AEIABL§§
INDEPENDENT IMPACT ON IMPACT ON IMPACT ON
ELRIABLEQ HEALTH SCHEDULE FINANCES
Patient Functional .01 .27* -.04
Status
Pt Depression .10 .26* .14
Cg Depression .50* .35* .27*
TOTAL R2 30.6% 42.6% 11.7%
*signtficant at p < .001

Ana 5' e heses

Given the results of the foregoing analysis, the first
series of hypotheses may be discussed. Hypotheses 11,6
reflect the causal relationships between the cancer
experience variables, duration of care and patient
functional status, and the three specific caregiver burden
responses. Each hypothesis will be discussed in light of
the correlations and the third regression equation
determining the relative contribution of the predictor
variable on the outcome of each of the caregiver burdens.
The hypotheses concerning the relationship of the duration
of care and burden can be summarized in the following:
E2222h§§i§ 1%3:There is a positive causal relationship
between duration of care in months and the burden response

(A) impact on health, (B) impact on schedule, and (C) impact
on finances.

58

Since only non-significant correlations were found
relating duration and all other variables in the model, this
variable could neither have direct nor indirect effects on
the burden variables. Therefore, hypotheses 1%; stating the
causal relationship between duration and the specific
caregiver burden responses must be rejected.

The next hypotheses to be discussed relates to another
dimension of the cancer experience, patients' physical

functional status, on the impact on caregiver health.

_6:There is a positive causal relationship
between patients' functional dependency and caregiver's
burden response (A) impact on health, (B) impact on
schedule, and (C) impact on finances.

The correlation for the cancer experience variable
patient functional status reveal significant associations
with the burden responses of caregiver impact on health and
impact on schedule, Pearson r .151 (p < .05)) and .424 (p <
.001), respectively. The impact on finances, however,
revealed non-significant correlations. All three burden
responses were regressed on patient functional status with
patient and caregiver depression. Only impact on schedule
yielded significant beta coefficients suggesting a direct
causal link between the two variables. There were no direct
causal relationships between the other two burden variables
and patient functional status. But in the case of impact on

health, patient functional health status has an indirect

59
effect on the burden response variable mediated through the
intervening variables of patient and caregiver depression.
The second set of hypotheses were intended to assess
the extent and strength of the relationship between the
selected characteristics of the cancer experience and

patient depression. Two sub-hypotheses were developed:

Hypothesis 2,: There is a positive causal relationship
between the duration of care and patients' depression.

As with the Pearson correlation between duration and
the caregiver burden responses, there was no significant
relationship found between the variable duration of care and
patients' depression (r .04 ns). Consequently, analysis did
not reveal any effect (either direct or indirect) of
duration of care.

The next stated hypothesis was intended to examine the
causal effect of the second dimension of the care
experience, patient functional status, on patients'

depression.

Hypothesis 22: There is a positive causal relationship
between patients' functional dependency and patients'
depression.

The simple regression on patients' functional status
for patients' depression revealed a significant beta

coefficient (.289, p <.0001), evidencing a causal

relationship. Since only 8% of the variation of patient

60
depression can be explained by the variable patient
functional status, other additional factors are likely to
also influence patients' depression. However cautiously,
the stated hypothesis can be accepted.

The third hypothesis proposed to examine the
relationship between patient and caregiver depression.
Pearson's r revealed a significant correlation (.399, p <
.0001), suggesting the possibility a causal relationship may

exist.

EXEQLA§§i§_;: There is a positive causal relationship
between patients' depression and caregivers' depression.

Regressions were run for caregiver depression on
patient depression and patient functional status and yielded
a significant beta coefficient (.358, p <.0001) only on
patient depression. Nearly 17% of the variation of
caregiver depression could be accounted for by patient
depression.

The fourth proposed hypothesis was designed to examine
the relationship between caregivers' depression and specific

caregiver burden responses.

flypepneeie_5: There is a positive causal relationship
between caregiver depression and the specific caregiver
burden responses of the impact on: A) health, B) schedule,
and C) finances.

61

Results from the correlation and regression analyses
on caregiver depression revealed significant and direct
relationships, respectively, for each of the three caregiver
burden responses. It appears that depression clearly
influences the caregivers' perception of burden with regard
to the impact on their health, schedule and finances. The
largest caregiver depression beta coefficient was found for
impact on health (.504, p <.0001), followed by impact on
schedule (.353, p <.0001). The causal relationship between
caregivers' depression and perceived impact on finances,
while still significant (p < .0001), was the weakest in
terms of the amount of variation explained (R? 11.7%).

The test for the goodness-of-fit of each of the models
was calculated by comparing the observed to the hypothesized
correlation (derived from the path model) in terms of the
respective standard errors. In each of the models
predicting caregiver burden responses, the difference
between observed and hypothesized correlations did not
exceed two standard errors. Thus, all path models were
accepted as reasonably fitting the data. Furthermore, while
impact on finances had a non-significant observed
correlation with patient functional status, a path chain
also indicated that the effect of patient functional status
on the burden response impact on finances was not likely to

differ from zero.

62
Summary

To summarize, patient functional status affects
caregiver burden responses primarily because of its
influence on patient and then caregiver health. Only the '
burden response of impact on schedule reveals a direct
effect from patient functional status. This finding may
have relevance to the design of intervention strategies to
assist in easing caregiver burdens by directing attention to
the mental health state of the patient, as well as the
caregiver.

In conclusion, other variables might be considered
when formulating future models for which to predict specific
burden responses, particularly for impact on caregivers'
health and finances. A summary of the study and findings
will be discussed in the next chapter. Implications for
clinical intervention and possible directions for future

research will also be discussed.

CHAPTER V

SUMMARY, CONCLUSIONS, AND DISCUSSION

Mes:

A summary of the present study and research questions
addressed in the hypotheses will be presented in this
chapter. Conclusions derived from the study findings and
implications for intervention will also be discussed. A
discussion of directions for future research will conclude

this chapter and study.

S a S u

The primary emphasis of the present research was to
describe and analyze the effect of selected characteristics
of the cancer experience, and the influence of patient and
caregiver depression, on caregiver burden responses.
Caregiver burdens were defined as the perceived impact of
the caregiving role on caregivers' health, daily schedule,
and finances. The present study was designed as an
exploratory recursive path model analysis in an attempt to
better understand the factors which may prove most salient
as targets for intervention to assist burdened caregivers.

Based on examination of the research literature, the

predicted path model of caregiver burden responses was

63

64
proposed to depict the relevant variables under study.
Since there was no specific theory of caregiver burden
responses for cancer for which to test, the exploratory path
analysis was designed to examine the effects of selected
predictor and intervening variables on caregivers' perceived
impact of care.

Nonetheless, causal ordering in the model was devised
by the investigator and suggested by the review of research.
Pearson linear correlations were used to determine the
relationships among all the variables. Path coefficients
were derived by employing three ordinary-least-squares

regression equations.

Su 0 ' s

The specific questions addressed in this research were:
to what extent does duration of care and patients'
functional status influence caregiver burden responses, and
does patients' and caregivers' emotional-mental state of
depression help to clarify the relationship to predict
caregiver burden responses? The first series of hypotheses
addressed the relationship between duration of care and each
of the three burden responses, impact on caregivers' health,
daily schedule, and finances. All hypotheses relating to
duration of care were rejected because no significant
relationship was found between it and the other variables

under study.

65

While duration of care did not appear as a significant
variable in the correlation and regression analysis,
questions arise as to problems of measurement and
conceptualization of the duration variable. Measurement
problems were likely a result of subjects having to recall
when care actually began, or the often ill-defined onset of
the disease. Conceptually, it is quite possible that the
further out in time patients and their caregivers get from
the date of new or recurrent cancer, the more apt they are
to adjust, normalize, or otherwise develop coping strategies
to deal with the impact and stress of cancer and care.
Thus, an important need exists to reassess and examine the
relationship between duration of care and caregiver burden.

The second series of hypotheses were stated to assess
the relationship between each of the burden responses and
patients' functional status. There were significant
correlations found between patients' functional status and
two of the burden outcomes, as well as both intervening
depression variables. Thus, patients' functional status was
retained in the causal model and subsequent analysis.
Results from the ordinary-least-squares regression analyses
revealed only the burden response of impact on schedule was
directly effected by the predictor, patients functional
status.

The relationship between patients' functional status
and patients' depression was the focus the next hypothesis.

A significant correlation was found between the two

66
variables and the beta coefficient (in this case, beta = r)
derived from the simple regression was entered in the path
figure.

A causal relationship was found between patient's
depression and caregiver's depression, which attests to
acceptance of the third hypothesis. A significant beta
coefficient was derived from the second regression and added
to the path figure.

The final hypotheses predicted a causal relationship
between caregiver depression and each of the three burden
responses. Significant beta coefficients were derived from
each of the regressions in the third equation and entered in
the path. Thus, the final hypotheses were accepted.

Testing of the path model was accomplished by
calculating the standard error of estimate for the sample
correlations. In order to assess the predictive power of
the path model for the population a test of goodness-of-fit
for each of the three burden models was calculated. It was
found that, indeed, each of the three path models predicted
population correlations consistent with the observed
correlations, i.e. the sample correlations did not differ by
more than two standard errors from the predicted population
correlation. Therefore, each model could be accepted as
partially predictive of the specific caregiver burdens under
investigation.

Only the model for impact on caregivers' schedule was

found to have both direct and indirect effects. It also had

67
revealed a greater amount of explained variation than each
of the models for impact on health and impact on finances.
Thus, the model for impact on schedule could be accepted as
better predictive than the models for impact on health and
finances.

In summary, based on the path model, patient's
functional status affects caregivers' burden responses
largely because it influences patients' and then caregivers'
mental health. Only in the case of impact on schedule does
there appear to be a direct effect of patient's functional
status on caregiver burden. An important implication of
this finding relates to the aims in design of interventions
to ease caregiver burdens, i.e. focussing on patients' and

caregivers' mental health.

r' Co 8'
The following conclusions may be made based upon the

findings from the present study.

1. Duration of care, measured in months, is not a relevant
variable in the proposed path model for caregiver
burdens among spouse caregivers for cancer patients.

2. There is a direct causal relationship between patients'
functional status and caregivers' burden impact on

schedule.

68

3. Patients' functional status is an important factor
which, if not predictive of, directly contributes to
cancer patient's depression.

4. Among husbands and wives, there is a significant causal
relationship between patients' depression and
caregivers' depression.

5. There is a causal relationship between caregivers'
depression and their burden responses to the impact on
their health, daily schedule, and finances.

6. The causal effect of patients' functional status on
caregiver burden responses is largely through its'
influence on patients' and then caregivers' depression.

7. The observed sample correlations in this study did not
differ significantly from predicted population
correlations in each of the three path models. Thus,
the models are found to be acceptable in delineating a
limited number of predictive influences to cancer

caregiver burden.

s te ’ ns
As depicted in each of the path models, caregiver
burden responses appear to be influenced primarily through
patient's depression and then caregivers' depression. This
finding has important implications to ease caregiver
burdens. While caution should be noted to not deny the
effect of patients' functional status, the psychosocial

aspect of depression appears to exert a greater influence on

69
caregivers' perception of the negative impacts of care.
Since care is often characterized as a family event or
family illness, strategies to address family reactions (i.e.
spouse caregivers' burden or depression) should therefore
take into account the family issues that are relevant to
specific responses incurred by the family's experience of
cancer. At the least, interventions should be designed to
address the mental health needs and well-being of both the
patient and his/her caregiver.

Identifying the goal to lower depression may be more
efficacious to the easing burdens than would the design of
specific programs to relieve the caregiver of specific
activities of care (i.e. respite or in-home help). The
critical point is that depression will likely color the
perceived impact even if the specific burden impact is
objectively relieved. Thus, caregivers may still feel the
effects of the burden.

While this study did not seek to address the causes of
depression, some of the intervention approaches may be
appropriately aimed concurrently at the causes and symptoms
of depression. For example, nursing strategies may promote
activities which advance greater health locus of control,
particularly with regard to symptom management and the
control over side effects of cancer treatment protocols
which could be influencing patients' depression.

Psychosocial interventions to address and treat family

and marital problems related to the role disruption brought

70
upon by cancer are another relevant area to broach in the
ease of cancer care related stress and burden. Attention to
the stress associated with the complementary roles of
patient (care receiver) and the well spouse (caregiver) may
be an important avenue to consider for clinical
interventions with cancer caregiver dyads.

An additional issue not addressed by the present study,
but which deserve mention as potentially critical factors in
assessing the impact of cancer on the family, relates to the
dynamic issues within the family system. Specifically,
problems related to interruption of the family life-cycle or
future family plans may influence both patients' and spouse
caregivers' emotional reactions to cancer and care.
Intrapsychic and world views toward death and spirituality
may also come into play when clinical intervention is called
for or initiated. The last issue, noted as both a
clinically relevant area of concern and a direction for
future research, involve the history and quality of the
relationship as the potentially overriding influence upon
depression and negative responses to cancer and care.

In summary, evidence from this study would suggest that
caregiver depression and burden responses are separate
dimensions of the family cancer experience. They must be
treated separately in the design of interventions to ease
the stress and burden of caregiving. Each of the burden
responses may requires distinct and precise approaches to

enable caregivers to experience any relief of burden.

71
However, as findings from this study indicate, depression
appears to be the overriding variable influencing the
perception of impact from care. Easing caregiver depression
then should be considered as a primary aim of interventions

to ease burden.

Web

Given that there is no arguably distinct or consistent
theoretical model of family caregiver burden, nor a specific
framework to guide an investigation of the stress .
encountered in family cancer care, the present study should
be viewed as an exploratory and descriptive path model
analysis of the cancer experience. The present model
predicting cancer caregiver burden, while not rejected as
significantly different from the population, does not cast
an exceptionally persuasive theoretical breakthrough. The
relevance of patients' depression, and then caregivers'
depression, was identified as mediating variables in the
causal path of caregiver burden responses. Thus, the modest
aim of the present study was achieved.

The selection of different predictor "need for care"
variables would be recommended to test for stronger and more
direct causal relationships to the burden outcomes.
Specifically, by selecting temporal ranges of the duration
of care variable may help in selecting those families newly
diagnosed, or diagnosed as newly recurrent, and reveal

variant effects of duration upon burdens.

72

Other specific measures of patient functional status
should be utilized for finer,‘more clearly distinct
descriptions of patient dependency. The measure used in the
present study, while presumably sensitive enough, did not
provide remarkably thorough detail. While the measure
served adequately for the current study, a more elaborate
multidimensional indicator would have perhaps served the
model better.

The current state of research in cancer caregiving is
reflected in the relative infancy with which the theoretical
groundwork has been laid. This study represents an attempt
to examine only some of the relevant concepts germane to
cancer family care. Further research in the general area of
psychosocial aspects of cancer families, and caregiving for
cancer patients in particular, is urgently needed. Specific
research directions should address variable issues of the

pre-caregiving environment.

fiBEEflIY

In final summary, the present study stands as an effort
to develop and test an exploratory model predicting the
specific caregiver burden responses of impact on health,
schedule, and finances. This study was designed to examine
the influence of patient and caregiver depression as a
mediating variable in the causal relationship between
patient functional status and caregiver burden. Findings

from this research reveal that, among husbands and wives,

73
cancer patients' functional status affects caregiver burden
largely because it influences patient depression and then
caregiver depression. Only in the case of impact on
schedule was there a direct effect of patients' functional

status.

APPENDICES

APPENDIX A

74

§E§:D_§QALE

During the PAST MONTH, how much of the time ...

Rarely or Some of Most of Almost None

of the Time the Time All of

the Time the Time

(1) (2) (3) (4)

1. were you bothered by things that usually don't
bother you?

2. have you not felt like eating; had a poor appetite?

3. have you felt that you could not shake off the blues,
even with the help from family or friends?

4. have you felt that you were just as good as other
people?

5. have you had trouble keeping your mind on what you

were doing?

6. have you felt depressed?

7. have you felt that everything you did was an
effort?

8. have you felt hopeful about the future?

9. have you thought your life has been a failure?

10. have you felt tearful?

11. has your sleep been restless?

12. were you happy?

13. have you talked less than usual?

14. have you felt lonely?

15. were people unfriendly?

16. have you enjoyed life

17. have you had crying spells?

18. have you felt sad?

19. have you felt that people disliked you?

20. could you not get "going?"

APPENDIX B

75

QABEQIYEB_EQBDEN_BE§EQN§E§

 

Strongly Agree Neither Agree Disagree Strongly
Agree Nor Disagree Disagree
(1) (2) (3) (4) (5)

A. I!Pé£&.22.§£§l£h=

1. I am healthy enough to care for . (reflect)

2. It takes all my physical strength to care for .

3. My health has gotten worse since I’ve been caring
for .

4. I have enough physical strength to care for
(reflect)

5. Since caring for it seems like I’m tired all
the time.

B. Impaet en Qaily acnegnle:

1. My activities are centered around care for .

2. I have to stop in the middle of my work to provide
care .

3. I have eliminated things from my schedule since
caring for .

4. The constant interruptions make it difficult to find
time for relaxation.

5. I visit family and friends less since caring for

C. Impacg on zineneee:

1. My financial resources are adequate to pay for
things that are required for caregiving. (reflect)

2. Its difficult to pay for 's health needs and
services.

3. Caring for has put a financial strain on the

family.

APPENDIX C

76
O S R O S A
Which of the following been describes your relative’s
physical functional performance status within the past

month?

No symptoms, fully active and able to carry out all
daily activities without restrictions (1)

Some symptoms, fully active and able to carry out
light activities or sedentary activities (house or
office work) (2)

Symptomatic, unable to carry out work activities and
in bed less than 50% of the day (3)

Symptomatic, unable to care for self, in bed 50% or
more during the day (4)

Symptomatic, unable to care for self, bedridden (5)

LIST OF REFERENCES

77

REFERENCES

(1989). Ceneez Feeps & Fignges -

American Cancer Society,
ieee. (Report 89-450M No. 5008-LE) Atlanta, GA: American

Cancer Society.

& Smith, J. (1979). The general systems
In W. R. Burr, R. Hill, F. I.

Broderick, C.,
approach to the family.
Nye, & I. L. Reiss (Eds.), Cenpenpgzeny tneozies epgut
pne fenily: Vol, 11, (pp.112-129). New York: Free Press.

Baider, L & DeNour, A. K. (1984). Couples'
adjustment to mastectomy: a preliminary report.
' Me ' ' e, 15, 265-

Int t'o Jou n s c i

276.
E. M. (1985). Parent care as a normative family

Brody, ‘
stress. Serentglegist. 25 (1). 19-25

Burr, W. R., Leigh, G. K., Day, R. D.,
(1979). Symbolic interaction and the family.

Burr, R. Hill, F. I. Nye, & I. L. Reiss (Eds.),
' ° ° V . I.

Co o a t o
(pp.42-111). New York: Free Press.

Cassileth, B. R., Lusk, E. J., Strouse, T. B., Miller, D.S.,
L. & Cross, P. A. (1985). Psychological

Brown, L.
analysis of cancer patients and their next-of-kin.

Caneen, 55, 72-76.
B., Miller, D.S.,

Cassileth, B. R., Lusk, E. J., Strouse, T.
A. N. (1984).

Brown, L. L., Cross, P. A., & Tengalia,
Psychosocial status in chronic illness. Nen_£ngleng

leurnal_2f_nsdisins. 31. 506-511-

Cassileth B. R. & Hamilton, J. N. (1979). The family with
Cancer. In B. R. Cassileth (Ed.), Tne Ceneer Patient,
Soeial and negieel Aepecgs pf gene. (pp. 233-247)
Philadelphia: Lea & Febiger.

Cancer recurrence: personal meaning,
Ca ce ursi ,

reactions and

& Constantine, J.
In W. R.

Chekryn, J. (1984).
communication, and marital adjustment.

1, 491-498

Cooper, E. T. (1984).
diagnosis of lung cancer on family relationships.

Nunsing, 1, 301-308.

A pilot study on the effects of the
Cancer

78

Derogatis, L. R., Morrow, G. R., Fetting, J., Penman, D.,
Piasetsky, R., Schmale, A. M., Heinrichs, M., & Carnicke,
L. M. (1983). The prevalence of psychiatric disorders
among cancer patients. 0 e ' Me 'ca

Assesiatien. 2&2: 751-757-

Devins, G. M., & Orme, C. M. (1985). Center for
Epidemiologic studies depression scale. In D. J. Keyser

& R- 0- Sweetland (Eds.), Tsst_2ritigues1_yelume_II.
Kansas City, MO: Test Corp. America, pp. 144-160.

Edstrom, S., & Miller, M. W. (1981). Preparing the family
to care for the cancer patient at home: a home care

course. Qanee; Nursing, 3, 49-52.

Farber, J. M., Weinerman, B. H., & Kuypers, J. A. (1984).
Psychological distress in oncology outpatients. Journal

ef_£sxshesesial_gnselesx. 2(3/4) 109-118-

Frank-Stromberg, M., & Wright, P. (1984). Ambulatory cancer
patients pereceptions of the physical and psychosocial
changes in their lives since the diagnoses of cancer.

Canee; Nunsing, 1, 117-130.

Friedenbergs, I. Gordon, W., Hibbard, M., Levine, L, Wolf,
C., & Diller, L. (1981-82). Psychosocial aspects of
living with cancer: a review of the literature.

integnetionai Jenznei of Esyenietzy, ii, 303-329.

Giacquinta, B. (1977). Helping families face the crisis of

cancer. Amsrisan_leurnal_2f_nur§in9 77 1585- 1588-

Given, B. A., Given, C. W. (1987). Family nemecepe fig;

ssnser.:_a_22mmunitx_bs§ed_med_l. National Center for
Nursing Research, Grant # R01 NR01915-02.

Given, B. A., Given, C. W. (1987). Enniiy homeeepe f0;
eenee;_pe§ien§e. American Cancer Society, Grant # PBR-
32.

Given, C. W., Given, B. A., and Stommel, M. (1987). Family
members reactions to caring for the elderly: a
conceptual and methodologic perspective. Unpublished
manuscript.

Goldberg, R. J. (1988). Psychiatric aspects of psychosocial

distress in cancer patients. Jenrnei of Eeycnosociel
QEEQLQQY: 5(1/2): 139-153-

79

Goldberg, R. J., & Cullen, L. O. (1985). Factors important
to psychosocial adjustment to cancer: a review of the

evidence. Sesial_§sisnse_nsdisine. 29. 803-807-

Googe, M. C., & Varrichio, C. (1981). A pilot investigation
of home health care needs of cancer patients and their

families. gnsplpsx_nursing_rprum. 2. 24-28-

Gotay, C. C. (1985). Why me? attributions and adjustment by
cancer patients and their mates at two stages in the

disease process. Sesial_§sienss_nedisine. 29. 825-831.

Grobe, M. E., Ahmann, D. L., & Ilstrup, D. M. (1982). Needs
assessment for advanced cancer patients and their

families. Qnsplegx_nursins_£9rum 9 (4). 26- 30-

Heinrich, R. L. & Schag, C. C. (1987). The psychosocial
impact of cancer: cancer patients and healthy controls.

Jougnei of Psyengeociai Oneoiogy, 5(3), 75-91.
Holmes, B. C. (1987). Psychological evaluation and

preparation of the patient and family. C , e9, 2021-
2024.

Holland, J. C. (1987). Managing depression in the patient
with cancer. C-A Can 0 ' 'c'ans, pl, 366-
371.

Hooyman, N. R., & Lustbader, W. (1986). Ieking Ceze. New

York: Free Press.

Karnofsky, A. D. & Burchenal, J. H. (1949). The clinical
evaluation of chemotherapy agents in cancer. In C. M.

Macleod (Ed ). E_a1uati2n_2f__hem2_hsrapsutis_ag_nts (pp
191- 205). New York: Columbia University Press.

Krant, M. J., & Johnson L. (1978). Family members'
perceptions of communications in late stage cancer.

e ou o n e e
Krouse, H. J., & Krouse, (1982). Cancer as crisis: The
critical elements of adjustment. Np;eing_geeee;en,

96-101.

Land, K. C. (1969). Principles of path analysis. In E.F.
Borgatta & G. W. Bohrenstedt (Eds.), Sgcioiogy
Mepnodelogy (pp. 3-37). San Francisco: Jossey-Bass.

Leiber, L., Plumb, M. M., Gerstenzang, M. L., & Holland, J.
(1976). The communication of affection between cancer

patients and their spouses. Esyenesonagie negicine, ee,
379-389.'

80

Lewis, F. M. (1986). The impact of cancer on the family: a
critical analysis of the research literature. Pepiene

Educatien_and_gpun_elins. 2. 269-289-

Lewis, F. M. (1983). Family level services for the cancer
patient: critical distinctions, fallacies, and

assessment. Qenee;_flnzeing, 5. 193-200.

Lewis, F. M., Ellison, E. S., & Woods, N. F. (1985L The
impact of breast cancer on the family. §enineze_in

Qnsplesx_nurs1ns. 1(3). 206-213-
Minuchin. s. (1974). Eam1119§_and_familx_therapx-

Cambridge, MA: Harvard University Press.

Mor, V. Laliberte, L., Morris, J. N., & Weinmann, M. (1984).
The karnofsky performance status scale, an examination of
its reliability and validity in a research setting.

Ca , 5;, 2002-2007.

Northouse, L. L. (1988). Social support in patient's and
husbands' adjustment to breast cancer. szsing Reeearch,
;1(2), 91-95.

Northouse, L. L. (1984). The impact of cancer on the _

family: An overview. Intsrnat19nal.129rnal_pf_£sxshiatrx
and_nedisine. 14(3). 215-242.

Northouse, P. G. & Northouse, L. L. (1987). Communication
and cancer: Issues confronting patients, health
professionals, and family members. Qpp;nei_gfi

£_xshesps1al_gnsplpsx. 2(3). 17- 46.

Northouse, L. L., & Swain, M. A. (1987). Adjustment of
patients and husbands to the initial impact of breast

cancer. Eurs1ns_89searsh. 25(4). 221-225.

Oberst, M. T., & James, R. H. (1985L Going home: Patient
and spouse adjustment following cancer surgery. Tepics

in_§lin1_al_nursing. 1(1). 47- 57-

Oberst, M. T., & Scott, D. W. (1988). Post discharge in
distress in surgically treated cancer patients and their

spouses. Bssearch_1n_nurs1n9_ans_nealth. 11. 223- 233.

Oberst, M. T., Thomas, S. E., Gass, K. A., & Ward, S. E.
(1989). Caregiving demands and appraisal of stress among

family caregivers. Qenceg Nuzeing, i2, 209-215.

Olsen, H. (1970). The impact of serious illness on the

family system. Epstsraduate_uedisine 11. 169-174-

81

Parkes, C. M. (1975). The emotional impact of cancer on

patients and their families. I9urnal_2f_Larxnsplpsx_and
9121291. 22. 1271-1279.

Plumb, M. M., & Holland, J. (1977). Comparative studies of
psychological function in patients with advanced cancer:
1. self-reported depressive symptoms. Eeyeneeenerie
Medicine. 22. 264- -276-

Poulshock, S. W., & Deimling, G. T. (1984). Families
caring for elders in residence: Issues in the measurement

of burden. lpurnal_ef_§erpntelegx..12(2). 230-239-

Radloff, L. S. (1977). The CES-D scale: a self-report
depression scale for research in the general population.

Appl1ed_2sxshelpgical_ueasurement. 1. 385-401-

Radloff, L. S., & Locke, B. Z. (1986). The community mental
health survey and the CES-D scale. In M. M. Weissman, J.

K. Meyers. & C- E. Ross (Eds.). memunitx.§urxs¥s_pf
P ° t 'c 's d . (pp.177-189). New Brunswick, NJ:

Rutgers University Press.

Quinn, W. H., & Herndon, A. (1986). The family ecology of
cancer. 12urnal_pf_£sxch2spsial_9ncplegx. 4(1/2). 45-59-

Skorupka, P., & Bohnet, N. (1982). Primary caregivers'
perceptions of nursing behaviors that best meet their
needs in a home care hospice setting. Ceneer Nursing, p,
371-374.

Snaith, R. P. (1987). The concepts and assessment of

depression in oncology. Jenrnei of Eeyenosocieigneoiogy,
5(3), 133-139.

Spiegel, D. Bloom, J., & Gottheil, E. (1983). Family
environment as a predictor of adjustment to metastatic

breast carcinoma. 22urnal_ef_£sxshesesial_9nsplesx. 1.
33-44.

Stetz, K. M. (1987). Caregiving demands during advanced
cancer, the spouse's needs. Qeneer_flnreing, 19, 260-268.

Stommel, M., Given, C. W., 8 Given, B. (1990L Depression
as an overriding variable explaining caregiver burdens.

I__rnal__f_A91n9_and_Esal_h. 2(1). 31 102-

Vess, J. D., Moreland, J. R., & Schwebel, A. I. (1985). An
empirical assessment of the effects of cancer on role

functioning. goprnei of Esyenesoeiel Qneeiogy, 1(1), 1-
16.

82

Vess, J. D., Moreland, J. R., & Schwebel, A. I. (1988).
Psychosocial needs of cancer patients: learning from

patients and their spouses. u o s
92.91291. 2(1/2). 31- ‘50-

Weisman, A. (1979). ering_yipn_geneer. New York: McGraw-
Hill.

Welch, D. (1980). Needs of families of adult cancer
patients. In G. Hongladorn & R. McCorkle (Eds.), Caneer:

A Eamiiy Qieee e. (pp. 23-35). Seattle, WA: Fred
Hutchinson Cancer Research Center, Second Annual Nursing
Symposium.

Welch, D. (1981). Planning nursing interventions for family
members of adult cancer patients. ganeer Nnrsing, A.
365-370.

Welch-McCaffery, D. (1988). Family issues in cancer care:
current dilemmas and future directions. Jo o

Esyengsoeiei Oncoiggy, §(1/2), 199-211.

Wellisch, D. K., Jamison, K. R., & Pasnau, R. O. (1978).
Psychosocial aspects of mastectomy II: the man's

perspective. W. 1.2. 543-
546.

Wellisch, D. K., Fawzy, F. I., Landsverk, J., Pasnau, R. 0.,
& Wollcott, D. L. (1983). Evaluation of psychosocial
problems of the home-bound cancer patient: the
relationship of disease and the sociodemographic
variables of patients to family problems. geprnei_g;

W291. 1(3). 1-15

Woods, N. F., Lewis, P. M., & Ellison, E. S. (1989).
Living with cancer: Family experiences. Qeneer_nnreing,
;;(1), 28-33.

Wright, K., & Dyck, S. (1984). Expressed concerns of adult
cancer patients' family members. Can e Nu s'n , 1, 371-
374 I