Florida State University Libraries
Electronic Theses, Treatises and Dissertations
The Graduate School
2005
Predicting the Risk of Compassion Fatigue:
An Empirical Study of Hospice Nurses
Maryann Abendroth
Follow this and additional works at the FSU Digital Library. For more information, please contact [email protected]
THE FLORIDA STATE UNIVERSITY
SCHOOL OF NURSING
PREDICTING THE RISK OF COMPASSION FATIGUE:
AN EMPIRICAL STUDY OF HOSPICE NURSES
By
MARYANN ABENDROTH
A Thesis submitted to the
School of Nursing
in partial fulfillment of the
requirements for the degree of
Master of Science in Nursing
Degree Awarded:
Spring Semester, 2005
Copyright © 2005
Maryann Abendroth
All Rights Reserved
The members of the Committee approve the thesis of Maryann Abendroth defended on
March 21, 2005.
______________________________
Jeanne Flannery
Professor Directing Thesis
______________________________
Denise Tucker
Committee Member
______________________________
Sandra Faria
Committee Member
Approved:
________________________________________________________________________
Linda Sullivan, Director, School of Nursing Graduate Program
________________________________________________________________________
Katherine P. Mason, Dean, School of Nursing
The Office of Graduate Studies has verified and approved the above named committee
members.
ii
This manuscript is dedicated to my uncle, Dr. John Bucsela. He was a gifted teacher and
scholar who seized life’s experiences with enthusiasm. He instilled in me a desire to
learn, a passion to teach, and the perseverance to fulfill my dreams.
iii
ACKNOWLEDGMENTS
To the special individuals whose invaluable support provided me with the
confidence and perseverance to fulfill my dream.
To Dr. Jeanne Flannery for your guidance, wisdom, and patience throughout this
process, and your unwavering belief in my talents even during the most challenging
times. I will value always your dedication to your students, and your commitment to the
highest of personal and academic standards.
To Dr. Donald Workman for your guidance in study design and statistical analysis
that has not only helped me accomplish this challenging project, but has taught me
analytical skills that will remain with me forever. Thank you for your insistence on
clarity, and belief in my abilities as a teacher and researcher.
To Dr. Denise Tucker for your tremendous support throughout this process. You
helped me see things differently during the most stressful times, which gave me strength
and perseverance to fulfill my dream.
To Dr. Sandra Faria, for your positive attitude and flexibility. Your
encouragement and belief in me instilled confidence that was much appreciated.
To Mr. Paul Ledford, for your organization’s support. Not only did you and your
staff at Florida Hospices and Palliative Care, Inc., encourage me, but you gave me a
voice in this initiative to help Florida’s hospice nurses.
To Mr. Jamie Marsh, for your valuable technical assistance. Your humor and
patience will always be remembered.
Finally, a very special acknowledgment to my husband John, daughter Jennifer
and son John. Thank you for your unconditional support, love, unwavering patience, and
for the sacrifices you made to help me complete this journey.
iv
TABLE OF CONTENTS
List of Tables ………………………………….……………………………………
List of Figures …….…..……………………….……………………………………
Abstract ………………………….………………………………………………….
Page
viii
x
xi
1. INTRODUCTION ………………………………………………………………
1
Statement of the Problem ……………………………………………………...
Significance of the Problem …………………………………………………...
Statement of Purpose …………………………………………………………..
Conceptual Framework ………………………………………………………..
Neuman Systems Model ……………………………………………….
Stress Theory …………………………………………………………..
Compassion Stress and Fatigue Model ………………………………..
Combined Model ……………………………………………………...
Research Questions ……………………………………………………………
Definition of Terms …………………………………………………………....
Assumptions …………………………………………………………………...
Limitations …………………………………………………………………….
Summary ………………………………………………………………………
2
3
4
4
4
5
7
9
11
11
12
12
13
2. REVIEW OF LITERATURE …………………………………………………...
15
Theory …………………………………………………………………………
Selye’s Stress Theory ………………………………………………….
Neuman Systems Model ……………………………………………….
Compassion Stress and Fatigue Model ………………………………..
Empirical Studies ……………………………………………………………...
Selye’s Stress Theory ………………………………………………….
Neuman Systems Model ……………………………………………….
Figley’s Compassion Fatigue Constructs ……………………………...
Stressors in Hospice Nursing ………………………………………….
Compassion Fatigue in Nursing and Other Disciplines ……………….
Summary……………………………………………………………….
15
15
18
20
22
22
23
24
25
31
33
3. METHODOLOGY ……………………………………………………………..
36
Design …………………………………………………………………………
36
v
Setting …………………………………………………………………………
Population and Sampling Plan ………………………………………………...
Protection of Human Subjects ………………………………………………...
Instrumentation ………………………………………………………………..
Compassion Satisfaction and Fatigue Subscales ……………………...
Demographic Questionnaire …………………………………………...
Procedure ……………………………………………………………………...
Data Analysis ………………………………………………………………….
Research Question 1 …………………………………………………..
Research Question 2 …………………………………………………..
Research Question 3 …………………………………………………..
Research Question 4 …………………………………………………..
Summary ………………………………………………………………………
4. RESULTS ………………………………………………………………………
36
40
41
43
43
43
43
46
46
46
46
47
48
50
Description of the Sample …………………………………………………….
Study Population and Sample …………………………………………
Supplemental Data Venue …………………………………………….
Research Question 1 …………………………………………………………..
Demographic Description of the Sample ……………………………...
Work-Related Description of the Sample ……………………………..
Research Question 2 …………………………………………………………...
Prevalence of Compassion Fatigue Risk ……………………………....
Research Question 3 …………………………………………………………...
Associations Between Demographic and ProQOL Factors …………...
Work-Related Factors …………………………………………………
Personal Health Description of the Sample …………………………...
Research Question 4 …………………………………………………………...
Theoretical Contributions to the Prediction of Compassion Fatigue Risk …….
Neuman’s Contribution ………………………………………………..
Selye’s Contribution …………………………………………………..
Figley’s Model ………………………………………………………...
Contribution of the Combined Theorists ……………………………...
Conclusions ……………………………………………………………………
Prevalence of Risk for Compassion Fatigue and Burnout …………….
Predicting the Risk of Compassion Fatigue …………………………...
Summary ………………………………………………………………………
50
50
51
55
55
56
60
61
63
64
65
67
68
70
70
73
75
78
81
81
82
83
5. DISCUSSION …………………………………………………………………...
84
Discussion of the Findings …………………………………………………….
Response Rate …………………………………………………………
External Validity ………………………………………………………
Relationship to Literature………………………………………………………
Conceptual Framework ………………………………………………………..
84
86
87
88
90
vi
Neuman’s Model ……………………………………………………....
Selye’s Stress Theory ………………………………………………….
Figley’s Model ………………………………………………………...
Combined Model ……………………………………………………....
Limitations of the Study ……………………………………………………….
Strengths of the Study …………………………………………………………
Implications for Nursing ………………………………………………………
Nursing Practice ……………………………………………………….
Advanced Nursing Practice …………………………………………....
Nursing Administration ………………………………………………..
Nursing Education ……………………………………………………..
Higher Education ……………………………………………………....
Recommendations for Future Research ……………………………………….
Comparison Studies …………………………………………………....
Policy Analyses ………………………………………………………..
Giving Nurses a Voice ………………………………………………...
Summary ………………………………………………………………………
91
92
93
93
95
96
97
97
98
99
100
100
101
101
101
102
102
APPENDICES ……………………………………………………………………...
104
REFERENCES ……………………………………………………………………...
139
BIOGRAPHICAL SKETCH…...…………………………………………………...
144
vii
LIST OF TABLES
Page
3.1
Florida’s Health Planning Districts …………...................………………….
4.1
Independent Samples t-tests of Venue Ranked by Size of p Value ………… 52
4.2
4.4
Chi-square and Fisher Exact Tests of Venue for Selected Discrete Variables
53
Ranked by Size of p Value ……………………….…………………………
Group Statistics on Dependent Variables of PRN Employee and Full/Part
54
Time................................................................................................................
Univariate Descriptions of Demographic Factors ………………………….. 56
4.5
Univariate Descriptions of Experience, Education and Licensure ………….
57
4.6
Description of Work Setting Factors ………………………………………..
59
4.7
Hospice Nurses with Other Concurrent Employment ………………………
60
4.8
Levels of Risk for ProQOL Subscales ……………………………………… 60
4.9
Compassion Fatigue Risk by Gender, Ethnicity, and Marital Status ……….
4.3
38
62
4.10 Descriptives for Compassion Fatigue Risk by Selected Work-related
63
Factors ……………………………………………………………………….
4.11 Correlation Matrix of Intercorrelations Between ProQOL Subscales ……… 63
4.12 Correlation Values: Demographics and ProQOL Subscale Variables ……...
64
4.13 Correlation Values: Work-related Factors and Dependent Variables ………
66
4.14 PPM Correlations between ProQOL Subscale Variables and Selected
ProQOL Items ……………………………………………………………….
4.15 Correlation Between Personal Health Factors and ProQOL Subscale
Variables …………………………………………………………………….
4.16 Factors Matched to Neuman’s Model and Selected as Potential
Contributors to Prediction of CF Risk ………………………………………
4.17 Neuman’s Regression Model Summary ……………………………………
66
68
71
72
4.18 Factors Matched to Selye’s Model and Selected as Potential Contributors to
74
Prediction of CF Risk ……………………………………………………….
4.19 Selye’s Regression Model Summary ………………………………………. 75
4.20 Factors Matched to Figley’s Model and Selected as Potential Contributors
to CF Risk …………………………………………………………………...
4.21 Correlation Matrix of Variables Selected to Operationalize Figley’s
Variables …………………………………………………………………….
viii
76
77
4.22 Figley’s Regression Model Summary ………………………………………
78
4.23 Independent Variables Selected as IVs in a Composite Model for Predicting
Compassion Fatigue Risk …………………………………………………... 79
4.24 Correlation Matrix of Variables Selected to Operationalize Composite
80
Theorists’ Variables …………………………………………………………
4.25 Composite Theorists’ Regression Model …………………………………… 81
4.26 Burnout and Compassion Fatigue Low and Moderate to High Risk
Frequencies for Florida Hospice Nurses ……………………………………. 82
ix
LIST OF FIGURES
Page
1.1
Compassion Fatigue Conceptual Framework ……………………………….
10
3.1
Hospice Providers in Florida ………………………………………………..
37
3.2
Certificate of Need Hospice Service Areas …………………………………
39
4.1
Sample Regression Model (k = 5) Considering Operationalized Factors
from Neuman’s Systems Theory for the Prediction of Compassion Fatigue
Risk .................................................................................................................
Compassion Fatigue Conceptual Framework (Evolved) ……………………
72
5.1
x
95
ABSTRACT
Health care literature and mainstream media sources have produced volumes of
information regarding the current and projected effects of the nursing shortage. Little,
however, has been written regarding the deleterious effects of this phenomenon, and
other work-related factors, on the nurses in the trenches. Compassion Fatigue (CF)
among care providers has emerged in the literature as a concept worthy of study;
however, the population of hospice nurses has been virtually ignored.
The present inquiry utilized descriptive and inferential statistics to accomplish a
two-fold purpose. Initially, it investigated the prevalence of CF risk among Florida’s
hospice nurses and analyzed relationships among demographic, work-related, and
personal health factors. Secondly, the study employed the use of multiple independent
variables in a regression equation for the prediction of compassion fatigue risk.
Findings revealed that 78% of the sampled hospice nurses were at moderate to
high risk for compassion fatigue with approximately 26% in the “high risk” category.
Additionally, participants from the entire sample were experiencing the overt effects of
stress, manifested in hypertension (30%), depression/PTSD (22%), and headaches (28%).
These effects were, no doubt, exacerbated by the fact that more than half (53%) reported
stress from finances, slightly less than half were encountering five or more patient deaths
per month, and almost 65% were sacrificing their own personal needs for the needs of
their patients.
Major factors such as trauma, anxiety, life demands, and excessive empathy
(leading to blurred professional boundaries) were key determinants of CF risk in a
multiple regression model that accounted for 91% of the variance in this dependent
variable. With knowledge of these few variables, hospice organizations may identify
nurses at risk and take measures not only to provide needed support for these individuals,
but also, to seek to eliminate or reduce the contributing factors.
xi
This inquiry provided a first glimpse into the stressful world of Hospice nurses,
provided a means for the identification of those at risk of compassion fatigue and an
estimate of prevalence in a state where the demands on these caregivers are expected to
increase exponentially. While these results appear to have prescriptive value, replication
is warranted to validate the model's, as well as the study’s, other descriptive findings.
xii
CHAPTER 1
INTRODUCTION
The National Hospice and Palliative Care Organization (NHPCO) estimated that
3,300 operational hospice programs served roughly 950,000 patients in the United States
in 2003 (NHPCO, 2004). This is a 286% increase from 246,000 patients served in 1992
(NHPCO, 2004). In 2002, more than one in four patients who died in this country were
cared for by hospice, and the median length of service in 2003 was 22 days (NHPCO,
2003, 2004).
Caring for dying patients induces considerable stress which includes the challenge
of providing comfort care to patients with complex disease processes as well as being
empathic to families in psychosocial and spiritual crisis (Kulbe, 2001; Sontag, 1996).
Stress from continuously being faced with difficult family dynamics, and multiple
patients dying within a short period of time has an accumulative negative effect on
nurses’ coping abilities (Adams, Hershatter, & Moritz, 1991; Dean, 1998; Keidel, 2002).
Joinson (1992) investigated the nature of burnout in nurses and first coined the
term, compassion fatigue. Figley (1995) defined compassion fatigue as a secondary
traumatic stress reaction resulting from helping or desiring to help a person suffering
from traumatic events. Caregivers experiencing compassion fatigue may develop a
preoccupation with their patients by re-experiencing their trauma, and may exhibit
symptoms of avoidance of reminders, numbing in response to reminders, anxiety, and
persistent arousal. Symptoms of burnout and compassion fatigue are similar; however,
the latter has a more sudden and acute onset than the former which manifests by a gradual
wearing down of caregivers who are overwhelmed and unable to effect positive change.
Additionally, compassion fatigue is associated with a sense of confusion, helplessness,
and a greater sense of isolation from supporters than is seen with burnout. (Figley, 1995,
2002).
1
Statement of the Problem
Between 2000-2020, the nation’s population is expected to grow 18%, and the
subgroup of 65 year olds and older is projected to grow 54% (National Center for Health
Workforce Analysis, 2002b). This growth will result in an increased demand for nurses
especially in geriatrics and areas such as hospice care. In 2002, 81% of hospice
admissions were 65 years old and older (NHPCO, 2004). As the hospice census is
expected to increase, it is important to attract and retain nurses in this specialty area.
They play a vital role as members of an interdisciplinary team providing direct patient
care in the home, inpatient hospice setting, or other facilities.
Despite their vulnerability to the effects of the numerous stressors in caring for
dying patients, hospice nurses must maintain their own physical and psychological health
in order to fulfill their responsibilities to patients, families, and their organization.
Compassion fatigue can challenge a caregiver’s ability to provide services and maintain
personal and professional relationships (Collins & Long, 2003). Numerous studies have
cited the effects of compassion fatigue in other disciplines such as psychology, religious
ministry, emergency management, and veterinary medicine (Davis, 2003; McCann &
Pearlman, 1990; Roberts, Flannelly, Weaver, & Figley, 2003; Wastell, 2002). Also,
research has been conducted on stressors, burnout, and coping measures within the
discipline of nursing (Bene & Foxal, 1991; Hall, 2004; Payne, 2001; Schwam, 1998).
Finally, studies of end of life care revealed stress scores of palliative health care workers
as almost twice as high as those of newly widowed women and higher than patients with
newly diagnosed breast cancer (Vachon, 1987). However, there is very limited
information on the prevalence of, and variables associated with, the risk of compassion
fatigue within the population of hospice nurses. Given that hospice nurses may
experience compassion fatigue, a study of the prevalence of compassion fatigue risk, and
an analysis of demographic, as well as work-related, variables that may predict the risk of
this phenomenon among this population is essential to the profession. Ignoring this type
of study may have deleterious effects on patient welfare because without preventive
knowledge more nurses may fall victim to compassion fatigue which could lead to
increased illness, absenteeism, loss of productivity, increased risk of errors, and financial
burden in a society that is already experiencing a severe nursing shortage. Understanding
2
compassion fatigue can empower nurses to utilize preventive measures that promote self
care, improve patient outcomes, and optimize therapeutic relationships.
Significance of the Problem
It is projected that the nursing population will fall 29% below demand in the year
2020 (HPNA Position Statement, 2003). Findings from the 2000 registered nurse
population survey revealed an estimated 19,175 registered nurses in the hospice
workforce setting (National Center for Health Workforce Analysis, 2002a). There were
16,716 full time Registered Nurses and 2,368 Licensed Practical Nurses working in
Medicare-certified Hospices in 2001 (The Hospice Association of America, 2002) . More
specific demographic data of this nursing specialty are not accessible; however, in 2000
the national average age of all registered nurses was 45.2 years, and has steadily climbed
in recent years (National Center for Health Workforce Analysis, 2002a).
This aging of the nursing workforce, general population growth, increased life
expectancy, and job dissatisfaction due to increased workloads and stress related illness
may be major factors in future nurse shortages in all specialty areas (United States
General Accounting Office, 2001). Hospice nursing requires highly trained skills that
encompass in-depth knowledge of complex symptoms of multiple diseases that affect the
body, mind, and spirit. Comprehensive symptom management, expert assessment skills,
and holistic care that are consistent with the goals of the patient and family are essential
attributes of the hospice nurse. Nurses in this specialty will not escape this shortage
which will result in a direct impact on end-of-life care in the United States (HPNA
Position Paper, 2004).
Hospice caregivers are especially vulnerable to compassion fatigue because they
can become over-involved, over-invested and are frequently exposed to experiences of
loss (Keidel, 2002). The magnitude of this secondary traumatic stress reaction is
unknown in hospice nurses; however, burnout which may be a precursor or a risk factor
for compassion fatigue is revealed in numerous studies (Beaton & Murphy, 1995;
Cerney, 1995; Chrestman, 1995; Dutton & Rubinstein, 1995; Pearlman & Saakvine,
1995; Stamm, 1995; Rudolph et al., 1997; Weiner, 1989, as cited in Collins & Long,
2003). Lack of attention to this secondary traumatic stress reaction in hospice care will
impact the integrity of nursing and will compromise patient welfare. Therefore, it is
3
essential for the Advanced Practice Nurse (APN) to support evidence based research, and
promote new hiring and staffing policies to address compassion fatigue risk in the
hospice nurse population. The APN educator could play an integral role in creating
workshops, learning modules, and presenting in-service opportunities for preventive
efforts. Additionally, the educator could become a liaison to administrators, and assist
them in utilizing intervention strategies when faced with compassion fatigue risk among
the staff. These joint efforts may increase nurse satisfaction and productivity as well as
promote patient safety and enhance end-of-life care.
Purpose of the Study
The purpose of this study is to describe the prevalence of the risk of compassion
fatigue in hospice nurses in Florida, to explore the relationship between various nurse
characteristics and the risk of compassion fatigue, and finally, to provide a model for
predicting the risk of compassion fatigue from knowledge of demographic and workrelated factors.
Conceptual Framework
Three theoretical concepts will be combined to provide the conceptual framework
that will direct this study. Neuman’s Systems Model (1995) and Selye’s (1956, 1976)
stress theory will furnish the basis for assessing the risk of compassion fatigue among
nurses working with terminally ill patients. Integrating Neuman’s Systems Model (1995)
and Selye’s (1956, 1976) work in the area of stress will provide a better understanding of
the relationship between various nurse characteristics and compassion fatigue from a
systems view. Figley’s (1995) Compassion Stress and Fatigue Model will provide a
social science framework to guide the researcher in understanding emotional and
situational factors that could predict the risk of this phenomenon, and thus, aid in
preventive efforts.
Neuman Systems Model
The Neuman Systems Model (1995) is a dynamic systems approach to client
care. The client may be seen as a person, family, group, community or issue, which has a
relationship with stress and its consequences. This model provides a comprehensive,
holistic approach to the client/client system. According to Neuman, the client seeks
harmony and balance within an open system which occurs when there is information and
4
energy exchange within the environment. The client system contains five variables:
physiological, psychological, sociological, developmental, and spiritual which form a
core structure of survival mechanisms.
The client system is composed of the basic structure that serves as the core, lines
of resistance, normal line of defense, and flexible line of defense. The core structure is
composed of the basic survival factors or energy resources of the client. The core is
surrounded by the lines of resistance which are resource factors such as counseling that
help the client fight against stressors which could be variables that lead to the risk of
compassion fatigue. The lines of resistance are surrounded by the normal line of defense
which represents a state of stability for the individual and for the system (Neuman, 1995).
This line can be an example of the client’s usual coping patterns and lifestyle, as well as
developmental stage (Freese, 2002; Neuman, 1995). An example of the normal line of
defense can be the appropriate use of distancing in client care. The model’s outermost
broken ring is the flexible line of defense which is a protective buffer that prevents
stressors from breaking through to the core. This flexible line of defense is affected by
how clients utilize or interact with physiological, psychological, sociological,
developmental, and spiritual variables (Neuman, 1995). An example of this flexible line
of defense could be the hospice nurses’ appropriate use of boundaries and positive
outlook in responding to these five variables in order to prevent the risk of compassion
fatigue. Neuman’s systems approach toward the relationship of stress and its
consequences is linked to Selye’s stress theory which posits that environmental and
occupational stress produces distress among individuals, leading to adverse physiological
and psychological symptoms (Selye, 1976).
Stress Theory
According to Selye (1956, 1976) stress is a nonspecific response of the body to
any demand made upon it. The nonspecific response can be a nervous, immunological or
hormonal response, and the demand on the body can be caused by unpleasant or pleasant
conditions (Selye, 1976). Stress may be caused by pleasant or unpleasant conditions.
Unpleasant or harmful stress is called distress whereas the pleasurable form is called
eustress which causes much less physiological and psychological damage. Also, the
intensity of damage is the result of how one adapts to the stressor (Selye, 1976).
5
Stress is manifested always by a syndrome which encompasses a myriad of
nonspecific changes as they develop while the individual is exposed to a stressor. These
changes, which can alter the structure and/or the chemical composition of the body, can
be signs of damage or manifestations of the body’s adaptive reactions. During both
eustress and distress the body undergoes the same nonspecific responses resulting from
the positive and negative stimuli acting upon it; however, the degree of damage is
affected by the adaptability and coping mechanisms to change (Selye, 1976).
Selye’s General Adaptation Syndrome (GAS) shows the changes and the body’s
adaptation to stress in three stages: the alarm reaction, the stage of resistance, and the
stage of exhaustion (Selye, 1976). All human beings experience the first two stages many
times during their lives because this is a mechanism they use to adapt to various life
events.
During the alarm reaction there is increased sympathetic nervous system activity
as evidenced by the secretion of norepinephrine and epinephrine. Epinephrine causes
vasodilation in the heart, brain and skeletal muscles resulting in increased blood flow to
supply these organs. Norepinephrine causes vasoconstriction of the viscera and skin
which has the effect of redirecting blood flow to the vessels dilated by epinephrine
(McCance & Shelby, 1996; Selye, 1976). This catecholamine activity in this “fight or
flight” response results in increased heart rate, and increased blood pressure.
During the stage of resistance there are continued effects of sympathetic nervous
system activity along with increased vasoconstriction in the skin and viscera. These
physiological processes may lead to conditions such as hypertension. During this stage
of resistance, the effects on the adrenocorticotrophic hormone (ACTH) show increased
corticosteroid secretion, decreased inflammatory response, increased blood glucose and
altered protein and fat metabolism (Selye, 1976). Finally, the onset of the exhaustion
stage shows continued effects of sympathetic nervous system activity that may result in
cardiac failure, and renal failure. There are continued effects of ACTH in the exhaustion
stage as evidenced by a decreased immune response and a decreased resistance to
stressors (Selye, 1976).
This stress response is initiated when a stressor is present or perceived by the
mind. According to Selye, emotional triggers, especially the ones causing distress, are the
6
most important stressors for humans. Physical injury or physical illness is a major
stressor also; however, it is not so common as the emotional stimuli that individuals are
faced with daily. Anxiety causes distress and interferes with performance; however, the
way one perceives or interprets stressors influences the magnitude of the response and
impacts how one copes with its effects (Page & Lindsey, 2003; Selye, 1976).
Air traffic controllers, physicians, dentists, nurses, and lawyers are some of the
professions that are considered stressful. Working long hours predisposes different
occupational groups to cardiac conditions, and rotating shift work produces severe
disturbances in corticoid and adrenaline production (Selye, 1976).Hospice nurses are not
immune to these stressors. They often work long hours, and are exposed to the
cumulative effects of family crisis and multiple patients dying within a short period of
time. Additionally, they may have personal stressors that may overwhelm their adaptive
capabilities. Nurses experience an increase in sympathetic nervous system activity, as
evidenced by the release of catecholamines, in the alarm stage, which then becomes
normalized in the stage of resistance. However, if stressors become too numerous,
initiating the alarm repeatedly, or overwhelming, the stage of resistance is not successful
in normalizing body functions and illness, such as hypertension, may result.
Consequently, chronic or acute illness may be a result of the difficulty the nurses have in
adapting to the repeated emotional distress caused by personal or occupational factors.
Compassion Stress and Fatigue Model
Figley’s etiological model of Compassion Stress and Fatigue (1995, 1997) is
based on the assumption that empathy and emotional energy are necessary to establish an
effective therapeutic relationship. However, those traits can make the caregiver
vulnerable to the costs of caring. Figley noted that there are several factors which either
prevent or predict compassion fatigue. Initially, exposure to the client, empathic concern,
and empathic ability lead to an empathic response. This empathic response along with the
factors of disengagement and a sense of satisfaction lead to residual compassion stress.
Finally, residual compassion stress, prolonged exposure to suffering, traumatic memories,
and the degree of life disruptions result in compassion fatigue. These many components
together form a causal model which predicts this clinical phenomenon (Figley, 1995,
1997).
7
Initially the caregiver is exposed to the client and experiences the energy derived
from the client’s suffering and pain. The caregiver responds with empathic concern
followed by empathic ability which is the aptitude of noticing the pain of others. This
ability to empathize is a hallmark to helping others, yet it also puts one at risk for
compassion fatigue. Following empathic ability is the caregiver’s empathic response
which can cause the caregiver to experience strong emotions as a result of the facets of
the therapeutic role. These four variables are important attributes for caregivers yet, they
are risk factors of compassion fatigue (Figley, 1995, 1997).
The variables of a sense of satisfaction and disengagement are used as measures
to prevent compassion fatigue in this model. When caregivers are satisfied with their
ability to help the client, there is a feeling of satisfaction because there is an
understanding as to where the caregivers’ responsibilities end and the client’s
responsibilities begin. Disengagement is a healthy way in which caregivers can promote
self care by distancing themselves from the client/family (Figley, 1995).
Eventually caregivers may experience compassion stress if the aforementioned
coping actions are unsuccessful. Compassion stress is the residue of emotional energy
from the empathic response to the client (Figley, 1995, 1997). If this stress is unrelieved,
the consequences of a prolonged stress response become apparent as evidenced by
adverse physiological changes in the General Adaptation Syndrome (Selye, 1956).
The final three variables in this model play a role in increasing the likelihood of
developing compassion fatigue. They are prolonged exposure to clients, traumatic
recollections and life disruption. Prolonged exposure indicates an ongoing sense of
responsibility for the client over a protracted period of time. Traumatic recollections are
memories that can trigger symptoms of anxiety or other mental distress. Finally, the
caregiver may experience a life disruption which could be an illness, change in
professional or personal responsibilities, or a change in lifestyle. This added stress
combined with other existing variables in the model can increase the probability of
compassion fatigue, especially if the coping actions of achieving satisfaction and
practicing disengagement are unsuccessful (Figley, 1995).
8
Combined Model
This researcher synthesized Selye, Neuman, and Figley’s theories to create a
compassion fatigue conceptual framework for this study. In this model (Figure 1.1), the
human self is a holistic being who lives in a multidimensional system that is constantly
inundated with distress and eustress. The blue double circular line on the periphery
represents the individual’s environment which encompasses different variables that have
positive or negative effects on stress.
The self is able to maintain a balance of wellness through its protective
boundaries and coping patterns which are the blue dotted and green solid lines
representing the flexible and normal lines of defense. Even though stressors penetrate,
resource factors continue to fight to maintain the protective barriers which are the dotted
lines surrounding the core self, represented as the lines of resistance. When an
accumulation of stressors occur, there is a strike to the self, and there is a breakaway from
protective barriers and equilibrium (i.e., a balanced existence). The strike is depicted by
the lightening bolt which penetrates the core. The result is compassion stress followed by
compassion fatigue, which usually has an acute onset triggered by multiple accumulated
events including demographics, occupational stressors, lack of adaptability, an inability
to cope, and intra-, inter-, and extrapersonal factors.
9
Distress
Eustress
Traumatic
memories
prolonged
exposure to
suffering,
life demands
Empathic
Ability
Disassociation
Self
Compassion
Satisfaction
Residual
Compassion
Stress
Adaptability
Coping
Compassion Fatigue
Predictors:
Demographics, Occupational Stressors,
Lack of adaptability, Inability to cope,
Inter, Intra, Extrapersonal Factors
Figure 1.1. Compassion Fatigue Conceptual Framework combines three theoretical
concepts. The Neuman System’s Model, which represents self and its protective
boundaries, is blended with Selye’s stress theory which depicts the concepts of eustress /
distress, and the variables that represent coping and adaptability. Figley’s Compassion
Stress and Fatigue Model is threaded into this combined model as illustrated by the
effects of positive and negative variables (lavender and pink ellipses) that play a role in
residual compassion stress.
10
Research Questions
In this study, the following questions will be addressed:
1. What are the demographic and work-related characteristics of hospice nurses
sampled for this study?
2. What is the prevalence of the risk of compassion fatigue among the hospice
nurses in the state of Florida?
3. What is the nature of the relationship between demographic, hospice work-related
factors and the risk of compassion fatigue?
4. What are the demographic and hospice work related factors which predict the risk
of compassion fatigue?
Definition of Terms
For the purpose of this study, the following definitions will be utilized:
Compassion fatigue risk: The likelihood of developing compassion fatigue which is a
state of secondary traumatic stress resulting from continuous exposure to highly stressful
care-giving as measured by items 2, 5, 7, 9, 11, 13, 14, 23, 25, and 28 on the Professional
Quality of Life: Compassion Satisfaction and Fatigue Subscales - Revision III (ProQOLCSF-R-III). The level of risk of developing compassion fatigue will be estimated based
on theoretically derived cut-points: a score of 7 or less indicating a low risk; a score of 817 indicating a moderate risk; and a score of 18 or more indicating a high risk.
Compassion satisfaction: The pleasure that individuals derive from doing their work well
as measured by items 3, 6, 12, 16, 18, 20, 22, 24, 27, 30 on the ProQOL-CSF-R-III. A
score of 42 or higher indicates a high level of professional job satisfaction; a score of 3241 indicates a moderate amount of professional job satisfaction. A score below 32
indicates possible low job satisfaction or a focus toward deriving compassion satisfaction
outside of the working environment.
Burnout risk: The likelihood of developing burnout which is feelings associated with
hopelessness and difficulties in dealing with work or doing the work effectively. These
feelings usually have a gradual onset and may be the result of a high work load or a nonsupportive environment. Burnout risk is measured by items 1, 4, 8, 10, 15, 17, 19, 21, 26,
and 29 on the ProQOL-CSF-R-III. A score of less than 19 indicates positive feelings of
11
work ability; a score of 19- 28 indicates a moderate level of stress. A persistent score of
greater than 28 may reflect a high risk of burnout.
Demographic variables: Selected demographic variables regarding age, gender, ethnicity,
marital status, children in the home, responsibilities for elderly or disabled dependents,
health status, and level of education are operationalized as the choices indicated on the
Demographic Questionnaire.
Work related factors: Selected work-related factors regarding burnout risk and
compassion satisfaction as previously defined, length of nursing experience, area of
practice (i.e. inpatient, field, etc.), case load, shift work, volume of trauma/death
exposure, and concurrent employment, are operationalized as the choices indicated on the
Demographic Questionnaire.
Prevalence of the risk of compassion fatigue: The number of participants assessed at
high, medium and low risk for compassion fatigue among the nurses sampled in this
study. The intensity of risk will be measured using the ProQOL-CSF-R-III.
Assumptions
For the purpose of this study, the following assumptions will be made:
1. The participants in this study will answer all survey questions truthfully.
2. The participants in this survey have not received any information regarding
definition of terms, nor were they briefed on the concept of compassion
fatigue.
3. The participants of this study may be at risk for compassion fatigue.
4. Florida hospice organizations and hospice nurses have an interest in knowing
the prevalence of hospice nurses at risk for compassion fatigue.
5. Hospice contact persons will employ appropriately the random selection
procedure to obtain a sample of hospice nurses for this study.
6. Analytical assumptions will be addressed in Chapter 3.
Limitations
There are some limitations noted in this research. This study measures the risk for
compassion fatigue at one point in time. There is a possibility that perceptions may
change over time, due to individual circumstances. Additionally, analytical limitations
will be addressed in Chapter 3.
12
Summary
The dramatic increase of patients under Hospice care from 1992-2003 has been
illustrated (NHPCO, 2004). The use of the Neuman Systems Model (1995), and Selye’s
(1956, 1976) stress theory will provide the framework necessary to explore the
relationship between nurse characteristics and compassion fatigue. Figley’s (1995)
Compassion Stress and Fatigue Model will provide a social science perspective for
understanding the preventive and causative factors that play an integral role in predicting
compassion fatigue. Finally, the statistical model that will be used to predict the risk of
compassion fatigue from knowledge of demographic and hospice work-related factors is
built with consideration of this conceptual framework.
There is a growing interest in understanding the clinical phenomenon of
compassion fatigue and its impact on health care providers (Huggard, 2003). Numerous
studies have cited the effects of compassion fatigue in other disciplines such as
psychology, religious ministry, emergency management, and veterinary medicine (Davis,
2003; McCann & Pearlman, 1990; Roberts, Flannelly, Weaver, & Figley, 2003; Wastell,
2002). Also, research has been conducted on stressors, burnout, and coping measures
within the discipline of nursing (Bene & Foxal, 1991; Hall, 2004; Payne, 2001; Schwam,
1998). However, there is very limited information on studies connecting hospice nursing
to compassion fatigue.
Accumulated stress and loss from continuously being faced with difficult family
dynamics and multiple patients’ dying within a short period of time can have negative
effects on nurses’ coping abilities (Adams, Hershatter, & Moritz, 1991; Dean, 1998;
Keidel, 2002). A study of compassion fatigue risk among hospice nurses is vital because
understanding those risk factors may play a role in retaining nurses in this specialty
whose patient population is expected to increase amidst a severe nursing shortage.
Additionally, the intensity of accumulated loss, as seen in compassion fatigue, may
increase as the hospice client population continues to increase. Hence, nurses will
experience added stress as their caseloads increase, due to the effects of the nursing
shortage.
The purpose of this study is to describe the prevalence of compassion fatigue in
hospice nurses; to explore the relationship between various nurse characteristics and
13
compassion fatigue; and finally, to provide a model for the prediction of compassion
fatigue from knowledge of demographic and work-related factors. This research will
contribute to an understanding of the effects stressors have on hospice nurses, and it will
reinforce the belief that a prolonged stress response can have a negative impact on health
and quality of life (Figley, 1995; Neuman, 1995; Selye, 1976). Lack of attention to the
effects of compassion fatigue in hospice care may impact the integrity of nursing care,
and may compromise patient welfare.
A comprehensive review of literature to follow in Chapter 2 will emphasize
Selye’s Stress Theory, The Neuman Systems Model, and Figley’s Compassion Stress and
Fatigue Model, from a theoretical and empirical perspective that will include stressors in
hospice nursing, and the effects of compassion fatigue on all disciplines. This literature
review will provide the rationale supporting the argument to conduct research in this area.
14
CHAPTER 2
LITERATURE REVIEW
This chapter contains a review of the literature, which is organized under two
headings: theory, and empirical studies. Theoretical literature reviewed includes: Stress
Theory, Secondary Traumatic Stress Responses relating to Stress Theory, Neuman
Systems Model, and The Compassion Stress and Fatigue Model. Literature review in the
area of empirical studies includes, Stress Theory, Neuman Systems Model, Figley’s
Compassion Fatigue Model, stressors in hospice nursing, and compassion fatigue in
nursing and other disciplines
Theory
Selye’s Stress Theory
Stress is considered to be a nonspecific response of the body to any demand. This
demand can be caused by unpleasant or pleasant conditions (Selye, 1976). Stress is
manifested always by a syndrome, and it is a sum of nonspecific changes affecting the
body at any one time. Selye’s General Adaptation Syndrome (GAS), encompasses the
myriad of nonspecific changes as they develop while the individual is exposed to a
stressor. The GAS consists of three stages: the alarm reaction, the stage of resistance, and
the stage of exhaustion (Selye, 1976). All human beings experience the first two stages
many times during their lives because this is a mechanism they use to adapt to various
life events.
The alarm stage is acute and no living organism can remain continuously in this
state because death would occur within hours or days. Survival occurs when adaptive
mechanisms penetrate this state of resistance. Exhaustion ensues when the body can no
longer adapt, and the stress reappears. The symptoms in the exhaustion stage are similar
to the ones in the alarm stage. Immunological defenses are low; however, exhaustion can
be reversible as long as it affects only parts of the body (Selye, 1976). Additionally, the
15
GAS is closely coordinated to the local adaptation syndrome (LAS) which focuses on
tissues more directly affected by stress. For example, inflammation is the result of
microbial invasion in the body (Selye, 1976). There are many physiological processes
that participate in the stress response in the body’s attempt to maintain homeostasis.
Initially, the input from a psychological or physical stressor is processed within the
central nervous system (CNS). The CNS sends signals to the hypothalamus which
coordinates homeostatic adjustments. The three major responses that occur after
hypothalamic integration are: sympathetic nervous system (SNS) discharge through the
autonomic nervous system, the release of certain anterior pituitary hormones, and finally
the release of vasopressin in the posterior pituitary. Tissues, glands, and organs such as
the liver, pancreas, spleen and kidney are affected by responses to SNS discharge. For
example an exocrine gland response is exhibited by increased palmar sweat (Page &
Lindsey, 2003).
Peripheral tissues enervated with SNS nerve endings lead to the release of
norepinephrine which results in glycogenolysis in the liver, a decrease in insulin, and in
increase in glucagon secretion. Other physiological changes from norepinephrine release
include increased vascular smooth muscle contraction and increased heart rate and
contractility. Hence norepinephrine release causes increased blood glucose levels,
increased cardiac output, and the peripheral catabolism of protein and fat. This can lead
to changes in heart rate, respiratory rate, and if stress is of a sufficient magnitude it can
lead to increased blood pressure, and hyperglycemia (Page & Lindsey, 2003).
During the stress response, epinephrine is released from the adrenal medulla. This
release can increase blood glucose levels due to the inhibition of glucose uptake by
peripheral tissues. Epinephrine has greater influence on the cardiac system, and is the
major catecholamine in metabolic regulation (McCance & Shelby, 1996).
Increased secretion of ACTH from the anterior pituitary affects the adrenal cortex
causing aldosterone and cortisol release. Aldosterone affects the renal system because its
release causes retention of sodium and excretion of potassium in the renal tubule cells.
Increased cortisol secretion alters the immune and inflammatory responses, as evidenced
by a decrease in both T cell and B cell lymphocytes (Page & Lindsey, 2003). The release
of vasopressin from the posterior pituitary also has an effect on the renal system resulting
16
in reabsorption of water from the renal tubules. Therefore, the net result of defense of
fluid volume is evidenced by sodium retention due to aldosterone release, water retention
due to antidiuretic hormone (ADH, vasopressin), and an increase in renin in response to
sympathetic discharge (Page & Lindsey, 2003).
A reparative phase usually follows the stress response. This phase is defined by a
decrease in urinary nitrogen secretion, an increase in insulin and thyroid hormones, and a
decrease in glucagon and growth hormone. The goal of the reparative phase is to achieve
protein synthesis, to promote cell proliferation, and restore fat deposits. Although these
are normal physiological responses to stress, the magnitude and duration of stressors may
be so extensive that the homeostatic adaptive mechanisms fail, leading to debilitation and
death (Page & Lindsey, 2003).
McCance and Shelby (1996) noted that psychosocial distress may be predictive of
psychological and physical health outcomes. Depression, insomnia, and fatigue may be
associated with adverse life events. These conditions could lead to physical illness
resulting in transient immune deficiency. Additionally, stressful life events have
exacerbated symptoms of chronic disease such as diabetes and multiple sclerosis
(McCance & Shelby, 1996).
Selye (1976) believed that diseases were due to an inability to cope with stress.
Additionally, he believed that the magnitude of stress depended on how one reacts to the
stressor. The many observable signs of stress include general irritability, impulsive
behavior, inability to concentrate, headaches, muscle pain, and psychoses. In some
occupations stress has reached epidemic levels. Lambert, Lambert, and Yamase (2003)
revealed that health care providers, military officers, technicians, managers, executives,
sports coaches and clergy are most prone to workplace stress in the United States. A
stressful work environment can lead to various problems including unhealthy behaviors,
lowered job motivation, lost work time, lowered decision making ability, and turnover
(Lambert, Lambert, & Yamase, 2003).
McVicar (2003) noted that the emotional cost of caring has been one of the main
sources of distress for nurses for many years; however, the magnitude of the impact of
stress is debated. The eustress of increased level of arousal and mental acuity can be also
manifested as distress as evidenced by apprehension, pessimism, and increased
17
smoking/alcohol consumption. Severe distress may result in burnout as seen with
emotional exhaustion, depersonalization, and disengagement (McVicar, 2003).
Clark and Gioro (1998) illustrated that nurses experience many forms of stress
through indirect trauma. Secondary traumatic stress disorder, otherwise known as
compassion fatigue, has serious and complex implications for nursing. Selye’s research
demonstrated several ways in which organisms respond to stress. These include a
syntoxic response which is to ignore the stressor, a catatoxic, or fight, response which is
to face the stressor; and a flight response which is to flee from the stressor (Kees &
Lashwood, 1996; Selye, 1976). Secondary traumatic stress research indicates that people
indirectly experiencing stress have Selye’s same reactive patterns as those directly
experiencing the stress (Kees & Lashwood, 1996).
The Neuman Systems Model
According to Neuman (1995), the client is viewed from a holistic
multidimensional perspective, and has a relationship with stress and its consequences.
The client seeks harmony and balance within an open system which occurs when there is
information and energy exchange within the environment. The client system contains
five variables: physiological, psychological, sociological, developmental, and spiritual
which form a core structure of survival mechanisms.
The conceptual diagram of Neuman’s model consists of a core structure which is
composed of the basic survival factors or energy resources of the client. A series of
broken rings called the lines of resistance surround the core. These lines of resistance are
resource factors that help the client fight against a stressor. Outside of these initial rings is
a solid circle called the normal line of defense which defines a state of stability for the
individual and for the system. This line can be an example of the client’s usual coping
patterns and lifestyle as well as developmental stage (Freese, 2002; Neuman, 1995). The
model’s outermost broken ring is the flexible line of defense which is a protective buffer
that prevents stressors from breaking through to the normal line of defense. This flexible
line of defense is affected by how clients utilize or interact with physiological,
psychological, sociological, developmental, and spiritual variables (Freese, 2002;
Neuman, 1995).
18
One of the assumptions of this model includes the view that wellness occurs when
there is harmony and when system needs are met; whereas illness occurs when there is a
disjunction among the parts of the system due to unmet needs. Stressors are major factors
that can lead to disequilibrium within the system. They are considered tension-producing
stimuli that occur within the client system. There are three types of environmental
stressors that are present within internal, external, and created environments. They are
intrapersonal, interpersonal, and extrapersonal stressors. Intrapersonal stressors, or forces,
arise from within individuals; interpersonal ones originate between persons, as evidenced
by role expectations; and extrapersonal forces occur outside the person, as seen with
unexpected life events (Freese, 2002; Neuman, 1995). The various effects of compassion
fatigue may fall into several of these categories of stressors. An example of an
intrapersonal stressor is seen when nurses are unable to disengage themselves from their
patients and families. This stressor occurs when nurses dwell on their patient’s condition
to the point of preoccupation outside the work environment. This type of behavior can
lead to symptoms of survivor guilt, anxiety and powerlessness (Figley, 2002).
Interpersonal stressors occur when nurses see themselves as health providers who need to
please their patients and superiors even if they neglect their own health and well being.
This stressor appears when these caregivers perceive that their nursing role supercedes all
their other roles and personal needs. For example, a hospice nurse may experience an
interpersonal stressor when she has difficulty saying no to working extra shifts due to low
staffing. Finally an example of an extrapersonal stressor may be due to the financial
constraints of having to work many hours, perhaps having to work more than one job as a
single mother in order to maintain the home.
Other components of the model emphasize that stability, or homeostasis,
preserves the character of the system. In order for this to occur an adjustment in one
direction is countered by movements in the opposite direction. This results in a feedback
regulatory system in order to promote stability. Input and output are matter, energy, or
information that are exchanged between the client system, and feedback that circles
around input and output makes the system self regulatory in regard to a desired health
state, or a desired goal outcome (Neuman, 1995). In order for hospice nurses to maintain
stability and homeostasis, they must continuously self evaluate their mental and physical
19
health in order to maintain optimal wellness. It is easy to become over-involved with
patients and their families, and therefore, lose sight of maintaining a balance in one’s own
life, as evidenced by the ability to step back and maintain one’s professional boundaries.
If energy balance is not monitored, symptoms of stress, burnout, and compassion fatigue
may ensue.
In order to maintain a desired health state Neuman focused on prevention as
intervention. Interventions are actions to help the client maintain system stability, and
Neuman supported the initiation of an intervention when a stressor is suspected or
identified. The three levels of intervention are primary, secondary, and tertiary
prevention. Primary prevention occurs when a stressor is suspected or identified.
Secondary prevention occurs after symptoms from stress have happened, and tertiary
prevention results after secondary because it focuses on rehabilitation toward optimal
stability (Freese, 2002; Neuman, 1995). Nursing is a very stressful occupation, and
hospice nursing adds additional emotional demands of providing end-of-life care to
clients, and their families who are often in a state of emotional distress. Nurses in this
specialty area become stressed and may not be aware of the intensity of their stressors
which can lead to burnout and compassion fatigue. Caregivers and administrators who
become aware of the risks for compassion fatigue in hospice nursing should be more
inclined to institute primary and secondary prevention modalities to maintain the health
and well being of this population.
The Compassion Stress and Fatigue Model
Figley (2002) noted that compassion fatigue is a state of tension and a feeling of
preoccupation with traumatized patients. This preoccupation can be evidenced by reexperiencing the traumatic events, by avoiding reminders of the event, and persistent
anxiety associated with the patient. Compassion fatigue or secondary traumatic stress
reduces one’s capacity or interest in bearing the suffering of others.
The Compassion Stress and Fatigue model developed from social science research
contains several variables to predict compassion fatigue and hence to prevent and treat it
(Figley, 1995, 2002). The model is diagramed using the eleven variables that prevent or
enhance the risk of developing compassion fatigue. Initially the caregiver is exposed to
the client and directly experiences the emotional energy of suffering and pain. The
20
caregiver, thus, provides empathic concern which is the motivation to respond
professionally to those in need. Following the concern is empathic ability which is the
aptitude of noticing the pain of others. This ability to empathize is hallmark to helping
others, yet it also puts one at risk for compassion fatigue. Following empathic ability is
the empathic response which is based on the caregiver’s insight into the feelings and
thoughts of the client. In order to accomplish this, caregivers often project themselves
into the clients’ perspective. This may cause caregivers to have strong emotions similar to
those that are experienced by clients (Figley, 1995, 2002).
The concepts of satisfaction or a sense of achievement and disengagement are
used as measures to prevent compassion fatigue in this model. When caregivers are
satisfied with their ability to help the client, there is an understanding as to where the
caregivers’ responsibilities end and the client’s responsibilities begin. Disengagement is a
healthy way in which caregivers can promote self care by distancing themselves from the
client/family. They do this by letting go of the thoughts, and feelings associated with their
interaction with the client in order to focus on their own lives (Figley, 1995, 2002).
Eventually caregivers may experience compassion stress if the aforementioned
coping actions are unsuccessful. Compassion stress is the residue of emotional energy
from the empathic response to the client (Figley, 1995). If this stress is unrelieved, the
consequences of prolonged stress response become apparent, as evidenced by adverse
physiological responses as seen in Selye’s General Adaptation Syndrome (1956, 1976). A
prolonged stress response can have a negative impact on health and quality of life
(Figley, 1995; Neuman, 1995; Selye, 1976).
According to Figley’s model (1995), there are three additional factors that play a
role in increasing the probability of compassion fatigue. They are prolonged exposure,
traumatic recollections, and life disruption. Hospice nurses caring for the terminally ill
are continuously exposed to the trauma of patient death and family crisis. Prolonged
exposure may be alleviated by breaks between care-giving or by vacations when
attainable. Traumatic recollections are memories that can trigger symptoms of anxiety,
depression, or even Post Traumatic Stress Disorder (PTSD). It is not uncommon to care
for patients who remind nurses of their loved ones who have died. Additionally, these
memories may be from former traumatic experiences which could be triggered by certain
21
clients. These recollections can link caregivers to previous trauma or bad experience and
cause them to have strong emotional reactions. Finally, the caregiver may experience a
life disruption which could be an illness, change in professional or personal
responsibilities or a change in lifestyle. This added stressor, which is similar to Neuman’s
extrapersonal stressor (1995), coupled with other existing factors, such as occupational
stressors, may increase the probability of compassion fatigue (Figley, 1995).
Empirical Studies
Selye’s Stress Theory
Olofsson, Bengtsson and Brink (2003) conducted a study to assess nurses’
experience of stress in the workplace. Four nurses were interviewed in this qualitative
study that used a grounded theory approach. The interviews were tape recorded and
transcribed; and the analysis utilized open and axial coding to organize text into themes.
The results outlined themes of frustration, hopelessness, inadequacy, powerlessness, and
not feeling acknowledged. The core category in this study was considered the absence of
response. Nurses felt that there was no or little or no social responsiveness. There was a
lack of support and emotional feedback from supervisors in that work environment. It
was evident that environmental and psychological stress were affecting these nurses
based on the dialogue from the interviews. They showed signs of emotional
powerlessness, lack of energy, and depression resulting from their work perceptions.
Their physical symptoms correlated to the result of emotional and psychosocial stress
seen in Selye’s theory of stress. This study appeared thorough as it outlined themes and
summarized findings; however, sample saturation was not mentioned; therefore, it was
difficult to ascertain if sample size was adequate for this study.
AbuAlrub (2004) conducted a correlational descriptive survey using a
convenience sample of 303 hospital nurses to assess job stress, job performance and
social support. The theoretical framework of this study centered on Selye’s stress theory,
and the view that nursing was one of the most stressful professions (Selye, 1976). The
results indicated that there was a negative correlation between job stress and social
support from co workers (r = -10, p <.01). The researcher omitted the death and dying
subscale because many of the participants did not work in intensive care units. This did
not appear as a viable reason to omit this scale. The study’s limitations noted that
22
generalizations were not possible since this research utilized a sample of convenience.
However, a hierarchical regression analysis was utilized to test the hypothesis that stated,
as perceived job stress increases, nurses with high social support from co-workers will
perform better than will nurses with less support. Job stress was added to the second step
of the regression equation. The third step of the regression equation noted that the
squared term of the job stress variable was implemented to explore the presence of a
nonlinear relationship between job stress and job performance. This third step noted that
there was a 2% net change added to the cumulative R2 when the squared term of the job
stress variable was added to the regression equation. Therefore, this indicated that job
stress was negatively associated with job performance at p <.05. Additionally, the
squared term of job stress was positively associated with job performance at p < .05
which outlined a U-shaped relationship between job stress and job performance. There
were additional steps associated with this hierarchical regression analysis; however, the
final model showed that 20% of the variation in job performance was explained by the
following: a) the background variable, b) job stress, c) the squared term of job stress, d)
social support from coworkers, and e) the interaction between job stress and social
support from co-workers. Overall, the results indicated that there was a positive effect of
social support on job performance; however, the author recommended that further studies
should be conducted based on inconsistencies from previous literature findings
(AbuAlRub, 2004).
Neuman’s Systems Model
Gigliotti (1999) directed a study of women’s multiple role stress utilizing
Neuman’s Systems model as the theoretical framework. The literature reviewed indicated
that there was scarce research concerning the effects of stress on women who occupy
both the maternal and student role. In this study a convenience sample of 191 women
attending college for the first time were surveyed using a demographic questionnaire and
a 10 item scale that measures role stress. The mean age was 36.8 years, and the arithmetic
mean number of children was 2.5. Initially, an Analysis of Variance (ANOVA) was used
to support this study. The results indicated a positive correlation between participant ages
and perceived social support (r =.15, p <.04). The women in the older age group (> 35)
23
had the same mean perceived multiple role stress as did the younger women (t =.31).
After the ANOVA was performed there was a statistically significant four-way
interaction (F = 6.22, p <.01) between age, maternal role involvement, student role
involvement, and perceived social support. Additionally, a hierarchical multiple
regression analysis was performed because an ANOVA design loses information by
reducing continuous variables to dichotomous ones (Gigliotti, 1999). Four research
questions were presented with this analysis and the full model explained that there was
24% of the variance in perceived multiple role stress for women in the older age group
and nothing for the younger women. Specifically, for older women, their age potentiated
the effects of the maternal and student role involvement which exerted pressure on
Neuman’s flexible line of defense. These age variables also potentiated the effect of the
social support variable which resulted in a negative impact on the flexible line of defense.
Interestingly, perceived multiple role stress was unrelated to number of enrolled credits,
income, and number and ages of children. The results of this study emphasized the
importance of having social support as a buffer for stress, and noted that Neuman’s five
variables interacted in various ways to explain normal line of defense invasion. This
study was thorough; however, it was very difficult to understand due to the multiple
designs interfacing among different layers of variables.
Figley’s Compassion Fatigue Constructs
Meyers and Cornille (2002) conducted a study to assess the prevalence of
secondary traumatic stress symptoms in child protective service (CPS) workers, and to
identify factors that were linked to secondary trauma. A sample size of 205 male (17%)
and female (83%) CPS workers were surveyed to examine relationships of a) family of
origin functioning, b) personal traumatic history, c) exposure to child abuse victims’
trauma and, d) gender to secondary traumatic stress (STS) symptoms of the professional.
An Analysis of Variance (ANOVA) analytical design was utilized in this study. The
findings revealed that CPS workers who were employed five or more years experienced a
higher degree of STS symptoms than those working less than 5 years. There was a
significant difference between the severity of obsessive compulsive symptoms and
longevity of work. The longer the professional was employed the greater the symptoms
he or she experienced (F = 3.82, p <.05). Veteran workers, defined as those employed
24
more than one year, reported more symptoms of nervous tension and panic attacks than
those working less than 1 year. Additionally, professionals who worked > 40 hours per
week had greater symptoms of anxiety than those working less (M =.61 = < 40 hours vs.
M =.81 = > 40 hours). Additionally, a comparison was made of the intensity of STS
symptoms reported by gender. Findings revealed that female workers had more
symptoms of anger, irritability, exaggerated startle response compared to males in the
study. Females also reported to have more muscle pain, cardiovascular, gastrointestinal
and respiratory problems Meyers and Cornille (2002) noted that the CPSs who had
enmeshed family-of-origin relationships had more nightmares, intrusive thoughts, and
images than CPSs who were raised in a less enmeshed environment. In addition to these
findings, Figley’s (1995) theories of compassion fatigue/secondary trauma emphasized
that persons exposed to traumatized children are especially vulnerable to the side effects
of secondary traumatic stress. These professionals tend to endure this pain on a daily
basis throughout their careers.
Stressors in Hospice Nursing
Dean (1998) conducted research using 33 nurses from three hospice services to
investigate responses to difficult or demanding work related situations. The key elements
of this study sought to identify causes of stress and to identify areas where greater staff
support was needed. This study was performed in a large Midwestern state using the Self
Inventory of Situational Responses-TC (SISR-TC) and the Spielberger State Anxiety
Inventory (SSAI) questionnaires. Demographic variables were not noted on the
participants of this study. The participants completed questions and were asked to rank
likely causes of difficult or demanding situations. The categories were: a) intractable
symptoms, b) communication issues, c) death/loss, d) administrative issues, and e) other.
The data were ranked, and the general distribution properties were described with
frequencies and percentages. Results indicated that 42% of the respondents ranked
management of intractable symptoms the highest, and communication issues were ranked
second highest by 50% of the respondents. Finally, almost 48% ranked issues related to
death and dying in third place or lower. Additionally, the participants noted also that
difficult situations including dealing with various family dynamics and communicating
with physicians unfamiliar with hospice philosophy were quite challenging. The SISR-
25
TC findings indicated that a demanding situation was looked upon more as a challenge
than as a threat. The results of the SSAI revealed high anxiety levels among the hospice
nurses compared to the norm for working females. The researcher noted that this result
warranted further investigation. Finally it was revealed that greater staff support was
needed in ongoing in-service sessions, in addition to more open communication within
multidisciplinary teams, and more education of palliative care symptom management
(Dean, 1998). This article was presented well; however, it did not address adequately the
specific research design, and did not address any measures of central location which are
valuable in data analysis. Values of p < 0.001 were noted in discussing the differences
among the variables.
A study by Gray-Toft and Anderson (1986-87) utilized a qualitative study to
examine the sources of stress experienced by eight registered nurses and five licensed
practical nurses in an 11 bed hospice unit at a large, Midwestern community hospital. The
data from this qualitative study included observations, interviews with nurses, written
stressful incident reports completed by nurses, and support group discussions. During this
6-month study, 102 patients were admitted to the unit who ranged in age from 25-84
years. Their predominant diagnoses were malignancies, and the average length of stay
was 13.7 days. The following themes emerged as sources of stress for hospice nurses: a)
a lack of understanding of the hospice concept by hospital administrators and staff from
other units and, b) the process of supporting the patient and family in the dying process,
leading to greater involvement for the nurse which results in increased emotional
demands. The researcher noted that nurses experienced stress when they were exclusively
involved with their terminal patients and experienced a true sense of loss when their
patients died.
Another source of stress in the Gray-Toft and Anderson study (1986-87) noted
dealing with patients to whom the physician neglected to reveal a terminal diagnosis and
prognosis. Other themes included the difficulty of dealing with depressed or hostile
patients and their families who were needy, overly critical, or angry. Some of the nurses
felt threatened or overwhelmed by the ways families coped with their situation. Also, the
nurses’ stress gave rise to feelings of anger, helplessness, guilt, and frustration because
distraught families sometimes relied heavily upon the nurses, and at times developed
26
unrealistic expectations of them. Nurses also experienced stress when they could not
grieve after a patient’s death because they needed to attend to other nursing duties. These
feelings were compounded by those nurses who had recently experienced the death of
someone close to them, outside the work setting. Finally, it was noted that despite their
numerous stressors in hospice nursing, most nurses in this study did not want to return to
their previous nursing positions. This was an interesting qualitative study that provided a
level of deeper understanding toward the causes of stress and burnout within this nursing
population.
Mallett, Price, Jurs and Slenker (1991), conducted a study to compare the
occupational stress, death anxiety, burnout levels and social support of a sample of 209
hospice nurses and 167 critical care nurses. The mean age of hospice nurses was 40.2
years, and 34.8 years for the critical care nurses. Other demographic variables included
marital status, number of years in nursing, hours worked weekly, education, and staff
support group meeting attendance. A random sample of 150 Medicare certified hospice
organizations were selected. The directors were instructed to select randomly two female
nurses to complete the questionnaires. Critical care units were located in the same cities
as the selected hospice organizations.
Occupational stressors were analyzed and compared to critical care and hospice
nurses using the Stressful Situation Scale. Burnout was measured using a modified
version of the Maslach Burnout Inventory, and death anxiety was measured with the
Collett-Lester Fear of Death Scale. Social support was analyzed with the Social Support
Questionnaire developed by Sarason, Levine, Basham. Additional components of the
general questionnaire included demographic information, social support, stressful
situations, burnout and death anxiety (Mallett, Price, Jurs & Slenker, 1991)
Chi squared values and t-tests were used to analyze demographic characteristics
of the two samples. Hospice nurses were older, had more children, had more nursing
experience, worked fewer hours and attended support meetings more regularly than
critical care nurses. Results of the burnout inventory noted that critical care nurses
utilized depersonalization more frequently and had more emotional exhaustion than
hospice nurses. The t-test noted differences between the burnout scores of the two
groups. The mean burnout score for critical care nurses was 48.6 (SD =19.9) where as the
27
mean score of hospice nurses was 34.2 (SD =16.0). The extremely stressful situations that
hospice nurses ranked were inadequate staffing (35%) and unqualified staff (32%)
whereas the three extremely stressful situations critical care staff ranked were inadequate
staffing (49%), incompetent care (46%) and unqualified staff (44%). The results of this
study indicated that hospice nurses experienced less occupational stress and lower
burnout than critical care nurses. Pearson coefficients indicated a positive relationship
between burnout and occupational stress (r = .31); as occupational stress increased,
burnout increased. A low correlation was found between burnout and support (r = -.12);
as support increased, burnout decreased. This study compared means of burnout scores to
other previous studies; however, there was never any discussion of median values.
Limitations of this study revealed that the results should not be generalized to all critical
care and to all hospice nurses. Although the hospice organizations were randomly
selected there was no randomization in sample selection (Mallett, Price, Jurs & Slenker,
1991).
A study by Masterson-Allen, Mor, Laliberte, and Monteiro (1985) examined staff
burnout in a hospice setting using a multivariate regression analysis to identify
demographic, occupational and organizational predictors of burnout. The data were
collected from 26 hospices by the Health Care Financing Administration which were data
that was also part of the 1983 National Hospice Study (Greer, Mor, Sherwood, Morris &
Birnbaum as cited in Masterson-Allen et al., 1985). An additional 14 hospices were
selected by independent evaluators to serve as comparison settings. Eighty-eight percent
of the 1065 sample was female with a mean age of 43 years. Thirty-three percent were
nurses, 31.5 % were secretarial/administrative personnel, 17.3% were social workers,
10.9% were aides and 7.4% were supervisory personnel. Results of the regression
analysis indicated that age had the largest impact of all the variables. As age increased
burnout scores dropped. Additionally, high education levels were related to high burnout
scores. Overall, demographics accounted for 15% of the variance in the burnout variable
(adjusted R2 = 0.148). Occupational characteristics accounted for 6% of the variance (R2
change = 0.059), and all organizational characteristics combined were R2 change = 0.005.
Finally partial F tests noted a level of significance less than 0.001 for demographic and
occupational characteristics. Some of the overall findings were a concern for the
28
researchers because there had been numerous studies done that found the younger, better
educated staff member as a target for occupational stress and burnout. Also, full time
employment and longer employment showed high effects of burnout which suggest that
prevention measures should intensify as tenure increases. Finally, other factors that were
not in this study could have attributed to burnout. They included personal attributes,
motivation, temperament, and psychological stability (Masterson-Allen, Mor, Laliberte,
& Monteiro, 1985).
Sontag (1996) conducted a cross-sectional survey of all hospices in one state
regarding the concept of hospice total care, and whether it was implemented uniformly in
all hospices in a Midwestern state. The total care concept is defined by a provision of
services addressing physical, emotional, social, and spiritual needs. In the first phase of
this study 34 hospice program directors were interviewed. During the second phase of the
study social workers, chaplains and nurses were surveyed using questionnaires with open
ended items regarding subjective views of the various facets of hospice care. Ninetyeight nurses, 29 social workers, and 21 chaplains responded to the survey. Nurses were
identified as the ones most involved in the physical care of patients.
Open ended questions were used to question nurses about challenges they
experienced working in the hospice field. The results were divided into themes, and
ranked by frequency. No other statistical measures were utilized. The two most frequent
challenges were related to psychosocial concerns and the family dynamics of hospice
clients. Thirty-seven percent of the nurses identified challenges of dealing with complex
psychosocial needs of these patients and their families. These difficulties included
emotional problems and spiritual issues that, if unresolved, could cause very high anxiety
and stress at the end of life. The second most frequent challenge these nurses addressed
included the learned ability to balance meeting patients’ emotional needs while not
getting too involved with them and their families. An additional concern included their
struggle to keep balance and objectivity in their lives as they dealt with their own
emotional responses. One nurse noted how easy it was to become very close to patients at
the end of life. It is unlike other types of relationships. The third most frequent challenge
was adequately managing patients’ pain and symptom control along with the frustration
29
of working with physicians who are not current with their own knowledge of pain
management and hospice philosophy (Sontag, 1996).
In this study social workers reported role blurring with the nursing staff as a
difficulty in defining their role as social workers. It was noted that nurses often assume
responsibility for some social work responsibilities. High case loads for this discipline
was also a challenge or stressor. The challenges that chaplains face were mainly not
having enough time to meet the needs of the patients. Some felt that the role of chaplain
was not considered equal to the other roles of the multidisciplinary team. In conclusion,
the results of this study indicated that nurses are the primary care providers in that state’s
hospices. However, this survey noted that fewer than half the nurses had bachelor
degrees; therefore many did not feel equipped to provide psychosocial and spiritual
support outside their scope of practice. (Sontag,1996). Other results of the likert
questionnaire from the second phase of the study specifically pertaining to the
relationship between staff perception of hospice care and discipline, indicated that nurses
were less likely than the other disciplines to agree that patients’ medical needs were the
primary focus of their hospice program ( χ 2 (2, N =146) = 10.34, p = .006).
Consequently, there was a minimum amount of statistical measurement and analysis in
this study, and a limitation noted that the survey instruments had unknown reliability and
validity; however, the interviews that were conducted were straightforward which
allowed for clarification opportunities.
Kulbe (2001) conducted a study on stressors and coping measures of hospice
nurses. It was modeled after the Duffy and Jackson 1996 study. Kulbe’s research
consisted of a convenience sample of 221 hospice nurses from 25 hospices in New
Jersey. The Hospice Nursing Census Questionnaire and a Demographic questionnaire
were utilized. There was reference made to the validity but not to the reliability of this
tool in this study. Forty-four percent of the 221 questionnaires were returned. All the
participants were female and the mean age was 45.5 years. Survey respondents reported a
divorce rate of 20.6% compared to the 10.9% from the Duffy and Jackson (1996) study.
The major stressors reported in Kulbe’s survey were ranked from one to nine in order of
severity. The highest frequency of stressors were the amount of paperwork, followed by
the number of patients seen in one day, then to too many patients dying in a short period
30
of time, and working with physicians who misunderstand hospice philosophy. Other
stressors included poor relationships between nursing staff, the hospice facility
environment, and poor communication among members of the interdisciplinary team.
One of the limitations noted was an inadequate amount of statistical analysis that was
limited to rank and frequency distribution which did not provide a rich discussion of
research results.
Compassion Fatigue in Nursing and Other Disciplines
Adams, Hershatter, and Moritz (1991) conducted a study on accumulated loss
phenomenon among hospice care givers. The term accumulated loss fit the definition of
compassion fatigue which was later coined by Joinson (1992). Adams et al.(1991), noted
that feelings of accumulated loss, due to volumes of death witnessed in this profession,
causes unresolved feelings and lack of closure for hospice nurses. The researchers
indicated that accumulated loss reactions appeared different from reactions of loss and
burnout. They hypothesized that a phenomenon exists among hospice caregivers which
can be identified. One hundred out of 157 hospice caregivers responded to a 92-item
accumulated loss questionnaire and four open-ended questions. The survey was created
by the researchers; however, reliability and validity data were absent. A Verimax
Rotation with a five factor analysis was used and it revealed the following five
characteristics of accumulated loss: lack of closure, dying and death concerns, ideals
versus reality incongruity, identification-distancing, and diminished boundaries. After the
data were received, the authors made a general statement regarding the accumulated loss
phenomenon which results from being overexposed to the real, or idealized, death and
dying process on a daily basis. One of the limitations noted was an absence of statistical
values in the five characteristics of accumulated loss discussion even though the results
were summarized in narrative. Additionally a Chi Square test was used to compare the
professional and non professional care caregivers. Again narrative generalizations were
clearly outlined; however statistical values were not given.
The open-ended questions centered on the least and most rewarding aspects of
work, work frustrations, and what losses at work (not death-related ones) affected one the
most. The least rewarding aspects included rapid turnover of patients. It was considered
stressful to admit patients, and then they die a few hours later. The most rewarding aspect
31
occurred when ideals are reality, for example, being part of a peaceful death. The losses
included loss of staff and ideals. Finally, the main frustrations were a perceived lack of
support from the interdisciplinary team (Adams, Hershatter, & Moritz, 1991). This was
an intriguing study that appeared to bring nurses’ secondary traumatic stress into the
limelight. Even though the words compassion fatigue were not utilized, they could have
been easily substituted for the authors’ term of accumulated loss.
A descriptive qualitative pilot project by Maytum, Heiman, and Garwick (2004)
studied 20 nurses who work with children with chronic conditions in order to identify the
triggers and coping strategies that these nurses use to manage or prevent compassion
fatigue and burnout. The 20 informants were identified through purposive sampling and
all agreed to participate. Their pediatric nursing experience ranged from four to 36 years,
and their education level ranged from LPNs to ARNPs. Eleven open-ended questions and
probe questions were utilized based on a review of literature. In order to prepare for their
interviews, the participants were sent a patient scenario and broad definitions of
compassion fatigue and burnout a week before the interviews.
Findings reported a host of symptoms and triggers associated with burnout and
compassion fatigue. The majority of the 32 triggers were work related which were
divided into work overload, system issues, professional roles, and caring for children
with chronic conditions and their families. Other triggers included dealing with very
emotional families with unreasonable expectations, and seeing repeated painful
procedures and treatments done to children. The most frequent personal trigger was
becoming overly involved with patients or families, and crossing professional barriers.
Numerous coping strategies were also cited (Maytum, Heiman, & Garwick, 2004).
This was an interesting study; however, it appeared that the participants did not
fully understand the difference between compassion fatigue and burnout. They were
given basic definitions and then were asked to apply those labels to themselves. It is
uncertain if the participants were completely accurate in their assessments because some
of the interviewees seemed to blend or interchange the two concepts in their
commentaries (Maytum, Heiman, & Garwick, 2004).
32
Roberts, Flannelly, Weaver, and Figley (2003) studied compassion fatigue among
chaplains, clergy and other respondents after the September 11th, 2001 terrorist attacks.
The American Red Cross conducted a one day conference in which a survey was
distributed to 650 conference attendees. The instrument consisted of demographic
information which included clinical training type, and the distance from the participants’
home and workplace from Ground Zero. The second part of the survey consisted of the
Compassion Satisfaction and Fatigue Test. Four hundred three questionnaires were
completed and returned. Three hundred seventeen clergy represented 78.5% of the
survey participants whereas the rest of the sample consisted of seminary students, mental
health practitioners, and executives of mental health and disaster relief agencies.
An Analysis of Covariance (ANCOVA) was used, and the covariate was work zip
code. Factors such as workplace proximity to Ground Zero, religion, and length of time
volunteering for a relief agency had no effect on the dependent variables. The results
indicated that a large group of participants were at high risk for compassion fatigue. An
Analysis of Covariance revealed that American Red Cross volunteers had lower
compassion fatigue scores (M = 30.2; p < 0.05) than those who volunteered from other
agencies (M = 26.4). There were no other statistical measures noted in this study specific
to an analysis of covariance which included calculations of differences between group
means. The findings indicate that a substantial number of clergy and others living in the
tri-state New York area are currently at a significant risk for compassion fatigue. This
result was based on a hypothesis that clergy and disaster relief workers were experiencing
compassion fatigue because they were directly exposed to the trauma of Ground Zero,
and were still caring for families of loved ones who died in the terrorist attacks, (Roberts,
Flannelly, Weaver & Figley, 2003).
Summary
The purpose of this study is to describe the prevalence of compassion fatigue risk
in hospice nurses in Florida, to explore the relationship between various nurse
characteristics and the risk of compassion fatigue, and finally, to provide a model for
predicting the risk of compassion fatigue from knowledge of demographic and workrelated factors.
33
Selye’s Stress Theory (1976) has been discussed in order to outline the significant
role the stress response plays in all aspects of life, and that it has been recognized as a
factor in the health professions. In the Neuman Systems Model the client is seen from a
holistic multidimensional perspective, and has a relationship with stress and its
consequences (Neuman, 1995). The use of these two models provided the framework
necessary to explore the relationship between nurse characteristics and the risk of
compassion fatigue. Figley’s (1995) Compassion Stress and Fatigue Model provided a
social science perspective for understanding the preventative and causative factors that
play an integral role in predicting this phenomenon.
A review of the literature revealed that compassion fatigue, otherwise known as
vicarious traumatization, has received considerable attention within numerous health care
professions other than nursing (Figley, 2002; Roberts, Flannelly, Weaver, & Figley,
2003; Wastell, 2002). Nurses who work in hospice care, emergency room settings and
psychiatric units are engaged in trauma work, and little attention has been given to the
effects of compassion fatigue on this profession (Clark & Gioro,1998; Schwam, 1998).
Studies have been conducted on stress and burnout, which is a precursor to compassion
fatigue, in many professions including hospice nursing (Dean, 1998; Bene & Foxall,
1991; Payne, 2001). From a qualitative perspective, nurses experienced stress when they
were exclusively involved with their terminal patients and experienced a true sense of
loss when their patients died (Gray-Toft & Anderson, 1986-87; Maytum, Heiman, &
Garwick, 2004). From a quantitative perspective, results revealed that high education
levels were related to high burnout scores, and stress from continuously being faced with
difficult family dynamics, and multiple patients dying within a short period of time has a
cumulative effect on nurses’ coping abilities (Adams, Hershatter, & Moritz, 1991; Dean,
1998; Gray-Toft, & Anderson 1986-87; Keidel, 2002; Masterson-Allen, Mor, Laliberte,
& Monteiro, 1985).
Continuous stress may heighten the risk of compassion fatigue which can have an
impact on the caregiver’s ability to provide services, and maintain therapeutic
relationships (Figley, 1995; McCann & Pearlman, 1990). This potential increased risk
within hospice nursing will not be beneficial to the viability of the profession, and to a
34
future of optimal end-of-life care for an aging U.S. populace which is already being
affected by the nursing shortage (HPNA Position Paper, 2004).
It is hoped that data gained by this inferential study will contribute to a greater
awareness of compassion fatigue risk in Florida’s hospice nurse population, and that it
will empower administrators, advanced practice nurse educators, and caregivers to utilize
preventive measures that promote self-care among this population. The benefits of using
these preventive measures will increase productivity and safe practice; promote nurse
retention; improve patient outcomes; and optimize therapeutic relationships. A discussion
will follow in Chapter 3 that will provide a comprehensive explanation of the research
methodology selected for the present inquiry. The discussion will include the study
design, population/sampling plan, instrumentation, procedure and data analysis.
35
CHAPTER 3
METHODOLOGY
This chapter describes the methodology applied to the study. The design, setting,
population/sampling plan, and protection of human subjects are addressed. Additionally,
instrumentation, procedure, data analysis, and analytical assumptions are discussed.
Design
The purpose of this study was to describe the prevalence of compassion fatigue
risk within a population of Florida hospice nurses; to explore the relationship between
various nurse characteristics and the risk of compassion fatigue; and finally, to provide a
model for predicting the risk of compassion fatigue from knowledge of demographic and
work-related factors among hospice nurses.
In order to accomplish these objectives a non-experimental, correlational
(descriptive) design was chosen utilizing cross-sectional data. Also, the design is
predictive, as those demographic and work-related factors found to be strongly associated
with the risk of compassion fatigue will be utilized as independent variables in a linear
regression analysis.
Setting
This study took place at hospice organizations throughout Florida, which has been
recognized as the first state in the nation to utilize hospice care as a medical/legal
alternative for the terminally ill. According to the Hospice Foundation of America
(HFA), in 2001, there were 37 hospice programs in Florida which, along with their
satellite sites (Figure 3.1) represented 85 locations across the state. As of 2003, there
were 41 hospice programs in the state according to the Florida Hospice and Palliative
Care Incorporated directory listing.
36
Figure 3.1. Hospice Providers in Florida. The highest concentration of programs is in
three main areas of the state. They are a) Boca Raton/Pompano/Fort Lauderdale, b)
Greater Orlando, and c) New Port Richey/St Petersburg/Tampa Bay (HFA, 2002).
The state of Florida is statutorily divided into 11 health service planning districts
which provide a framework for projections of need for beds or health services (Table
3.1). These 11 planning areas are further divided into sub-districts for allocation of
hospice programs (Figure 3.2).
37
Table 3.1
Florida’s Health Planning Districts
District Counties
1.
Escambia, Okaloosa, Santa Rosa, and Walton Counties
2.
Bay, Gulf, Leon, Wakulla, Calhoun, Holmes, Liberty, Washington, Franklin,
Jackson, Madison, Gadsden, Jefferson, and Taylor Counties
3.
Alachua, Dixie, Lafayette, Putnam, Bradford, Gilchrist, Lake, Sumter, Citrus,
Hamilton, Levy, Suwannee, Columbia, Hernando, Marion, and Union
Counties
4.
Baker, Nassau, clay, St. Johns, Duval, Volusia, and Flagler Counties
5.
Pasco and Pinellas Counties
6.
Hardee, Polk, Highlands, Hillsborough, and Manatee Counties
7.
Brevard, Orange, Osceola, and Seminole Counties
8.
Charlotte, Hendry, Collier, Lee, Desoto, Sarasota, and Glades Counties
9.
Indian River, St. Lucie, Martin, Okeechobee, and Palm Beach Counties
10.
Broward County
11.
Miami-Dade and Monroe Counties
Between 1997 and 2000 the largest increase in hospice admissions occurred in the
Pensacola area (District 1) at 63.7% followed by 57.3% for the Orlando region (District
7) and 53.7% for the Palm Beach area (District 9). The entire state experienced a 33%
increase in admissions from 46,608 patients in 1997 to more than 62,000 in 2000. The
under age 65 hospice patient population has been on the rise since 1997. From 1999 to
2000, admissions doubled from 5.8% to 10.6% in that age category. Even though in 2000
the Pensacola area had the largest percentage increase in the number of admissions, the
actual 1,852 census was comparatively small when compared with the 8,128 census in
the Palm Beach area, which experienced the largest increase in the actual number of
admissions. The four geographic areas with the largest census of new hospice patients
were: Palm Beach (8,128) and Broward (8,083), Pasco/Pinellas (7,126) and Miami-Dade
(6,228). These regions represent approximately 36% of the 62,214 total state admissions
in calendar year 2000 (HFA, 2002).
38
Figure 3.2. Certificate of Need Hospice Service Areas in Florida
In 2000, Broward County (District 10) and the Palm Beach area (District 9)
represented the highest rate of hospice patients per 1,000 resident deaths. In District 10
there were 519 hospice patients per 1,000 resident deaths, and in District 9 the rate was
475. The lowest rates of hospice patients per 1,000 resident deaths occurred in the
Panhandle (District 2) and Miami-Dade (District 11) where the rates were 329 and 334
per 1,000 patient deaths, respectively (HFA, 2002).
There were 162,804 resident deaths in 2000, and 125,628 of these deaths were
residents 65 years of age and older. This age category represents 83% of hospice patients
(HFCA, 2002). In 2001, there were 56,941 hospice admissions age 65 or over, and there
were 11,323 admissions under age 65. Of the total 68,264 admissions, a cancer diagnosis
was more prevalent among the under age 65 category (AHCA, 2002). The other
39
diagnoses of the younger age group were divided into Chronic Obstructive Pulmonary
Disease (COPD), Congestive Heart failure (CHF) and neurological illness. Younger
patients with a cancer diagnosis were six times more likely to enter hospice care than
ones with other types of illness (HFA, 2002).
Hospice care has a big impact in Florida. In 2001, there were more than 68,000
hospice patients statewide. Each hospice death directly affects approximately 10-15
survivors from immediate family to friends (HFA, 2002).
Population and Sampling Plan
The target population for the present inquiry, was registered nurses (RN),
advanced registered nurse practitioners (ARNP), and licensed practical nurses (LPN) who
met the following inclusion criteria: a) 18 years of age or older, b) employed by a hospice
organization in Florida, and c) interact directly with patients and their families. This
population may work in any area of the organization such as: a) freestanding inpatient
hospice facility care, b) home care, and c) hospice admissions.
There is no available information on the characteristics of Florida hospice nurses;
however, findings from the 2000 registered nurse population survey revealed an
estimated 19,175 registered nurses in the hospice workforce setting in the United States
(National Center for Health Workforce analysis, 2002a). The demographic profile of
Florida nurses reveals that there were 147,320 RNs, 56,428 LPNs and 8,255 ARNPs with
active license status in December of 2002. The median ages of these three nurse
categories were 46, 45, and 47 years old, respectively. Of the 212,003 nurses, 39,896
were between 57-65 years of age, implying that 13.5% will retire by 2010 (Florida Center
for Nursing [FCN], 2003; Gregg & Brunell, 2003). Female was the dominant gender in
all groups; however, 30% of the licensee gender data was not accessible (FCN, 2003). In
1996-1997, there were 75.2% of RNs in Florida who were non-Hispanic white, whereas
13% were African American which was a close reflection of the state’s general
population at the time. Additionally, Hispanic /Latinos comprised 9% of the RN
population. In 2002 there was a 55.1% staff RN and a 51.8 % LPN turnover rate in
Florida, compared to the 48.9% national rate for both groups (FCN, 2003; Gregg &
Brunell, 2003).
40
The sampling frame for the study consisted of hospice nurses employed by
Florida hospice organizations, during the study period, who agreed to participate in this
study. These nurses in the accessible population met the same criteria for inclusion as
stated for the target population.
The sampling plan required that all hospice organizations be selected from a
directory provided by Florida Hospices and Palliative Care Incorporated, the organization
representing hospice providers in all 67 counties throughout the state. Additionally the
plan requested that designated hospice contact persons utilize a finite population random
selection process to select participants for this study from each of their organizations.
They were asked to assign a number to each nurse employee, and then select the study
participants from that list utilizing the finite population random selection procedure
provided by the researcher.
Upon completion of this selection process, the potential participants were given a
packet containing a cover letter, a demographic questionnaire, and the Professional
Quality of Life: Compassion Satisfaction and Fatigue Subscales - Revision III (ProQOLCSF-R-III). The cover letter indicated that completion of these instruments implied that
the potential participants had given consent to be involved in this study. The participants
sent the completed instruments to the researcher in self addressed, stamped envelopes
provided in the packet. The researcher made follow-up phone calls or visits with the
designated contact persons from the organizations if responses were not received within 3
weeks of the mailings. A total of 600 surveys were planned to be mailed to provide for a
minimum sample size of 178 responses, based on an anticipated return rate of
approximately 30%.
Protection of Human Subjects
This study was conducted following approval by the Florida State University
Institutional Review Board (IRB). A copy of the application approval is in Appendix A.
Participants were given an explanation both verbally and in writing that participation in
the study was strictly voluntary.
This researcher contacted Florida Hospices and Palliative Care Incorporated, and
asked them for a letter supporting this study, and allowing access to Florida hospice
organizations once IRB approval was obtained. Copies of Florida Hospices and Palliative
41
Care Incorporated support letters are in Appendix B. The directory of these hospice
organizations is open to the public via internet access. Subsequently, hospice nurses were
asked to participate through a finite population random selection process instituted by
hospice contact persons under the guidance of the researcher.
The participants of this study received a research packet that was distributed to
them by designated contact persons from the organization. The packet included a cover
letter, demographic questionnaire, the Professional Quality of Life: Compassion
Satisfaction and Fatigue Subscales - Revision III (ProQOL- CSF-R-III). The cover letter
outlined some minimal risks with this study. The risks in this research may have involved
mild anxiety while the participants were thinking about unpleasant experiences associated
with patient/family situations within hospice nursing. The letter indicated that submission
of the completed surveys constituted informed consent. Each organization’s Family
Support Counselors were asked to be available for research participants as needed. The
participants were informed that their names would never appear on the survey
instruments, and that they could terminate their participation at any point without any
prejudice or penalty. The time frame to review the materials, and to complete these
documents, should not have exceeded 30 minutes.
Hospice administrators would not have had access to the completed surveys
because the participants returned the completed documents in pre-addressed, stamped
envelopes that were provided by the researcher in the packet. Each facility was identified
by a coding process on the survey forms; however, precautions were taken to maintain
the confidentiality of these mailings, as any identifying information was deleted during
data entry. Outcomes were analyzed and reported as region specific and aggregate group
results. Individual facilities were not identified in outcome measures, reports, or
publications. The raw data are stored in a secured cabinet for a period of 5 years in the
researcher’s home. The researcher, committee chair, and the statistical consultant have
access to the raw data. After the designated period, the records will be destroyed by
December 30, 2009. The participants may have been given time during work to complete
the instruments; however, this was at the discretion of each facility’s administration, and
was not required by the researcher. The participants received no financial payment for
participating in this research from the researcher; however, the knowledge gained from
42
this study will be valuable to hospice nurses and their administrators in helping to
identify and prevent compassion fatigue risk in this population. Finally, each hospice
organization will receive an executive summary upon completion of this study.
Instrumentation
Compassion Satisfaction and Fatigue Subscales
Stamm (2002) developed the Professional Quality of Life: Compassion
Satisfaction and Fatigue Subscales - Revision III (ProQOL-CSF-R-III), a 30-item selfadministered behavioral assessment for professionals working with survivors of traumatic
stress (Appendix C). The 2003 version is the current revision of the original Compassion
Fatigue and Satisfaction 66 (CSF 66) item scale (Stamm & Figley, 1996) which has not
demonstrated adequately the psychometric separation of the concepts of burnout and
vicarious traumatization (Larsen, Stamm, & Davis, 2002).
Stamm established reliability of the ProQOL-CSF-R-III tool with scores of .87 for
Compassion Satisfaction alpha, .72 for Burnout alpha, and .80 for Compassion Fatigue
alpha (Stamm, 2002). Despite the fact that the ProQOL-CSF-R-III has fewer items than
the CSF-66 item instrument, the authors report more reliable scores. The major advantage
of the ProQOL-CSF-R-III tool is that it provides for psychometric separation of
compassion satisfaction, burnout, and compassion fatigue/secondary trauma. The
questions on this 30-item instrument are equally divided among these three categories,
and receive separate sub scores.
Demographic Questionnaire
A basic demographic questionnaire was developed by this researcher to include
demographic and work-related information (Appendix D). This instrument was based on
items used in Zimmerman’s (2000) study. Also, it utilized concepts from Figley’s
Compassion Stress and Fatigue Model (1995), Neuman Systems Model, and Selye’s
stress theory.
Procedure
Upon completion of a successful defense of the Prospectus and approval from
Florida State University’s Institutional Review Board (IRB), the researcher contacted the
administrators of all Florida Hospice organizations. The hospices were selected from a
directory obtained from Florida Hospices and Palliative Care Incorporated which is the
43
state organization of Hospice service providers. The initial contact was made by e-mail,
at which time the researcher introduced herself, briefly described the study, invited the
organization to participate and indicated that a phone call follow-up would take place in
10 days (Appendix E). This e-mail had three attachments. The first attachment was an
outline of the study (Appendix F), and the second attachment was a copy of a letter from
Florida Hospice and Palliative Care Incorporated supporting the inquiry and its objectives
(Appendix B). The third attachment was a survey requesting information about the
organization, and requesting feedback regarding interest and participation in the study
(Appendix G). The researcher requested that the administrators return this survey
electronically. After 10 days, the researcher contacted those hospice administrators by
phone who had not responded to the e-mail, and explained the project, outlined the
benefits and risks, and invited the organization to participate in this study. The context of
this phone conversation was identical to the survey instrument attachment in the initial email (Appendix G). After this initial phase of the study was completed, the researcher
grouped the administrators’ responses into the following categories: a) immediate
approval, b) refusal to participate, and c) conditional approval. Organizations giving
immediate approval were selected for the study, whereas those refusing were not part of
this study. Organizations with conditional approval received a follow up contact to obtain
their final decision, and no other information was sought at that time. The follow-up date
was chosen and approved by the administrators; however, this contact was not to exceed
6 weeks after their initial response that indicated they wished to participate.
Designated contact persons from the organizations agreeing to participate
received a cover letter with an explanation of how to execute the finite population
random selection procedure. Initially they obtained a list of nurses (LPN, RN, ARNP)
employed at their organization responsible for direct care of patients. They were then
instructed to begin the selection process by consecutively numbering the list beginning
with number 1. Then the contact persons flipped a coin. A result of heads directed them
to begin the selection with the first person on the list. A result of tails directed them to
begin the selection with the second person on the list. As they went down the list they
were asked to select every kth person. This k value was unique to each hospice
organization, based on a proportional sampling procedure dependent upon both the
44
number of hospice facilities agreeing to participate in the study, and the number of
eligible nurses employed by each of these facilities.
These contact persons were encouraged to follow this procedure and, if for any
reason a selected nurse could not participate in the study, the contact persons were to
continue down their list selecting every kth person. If the number of nurses the researcher
asked them to select exceeded the number that could participate, then the number that
could participate through this random selection process were to take the survey. They
were encouraged to stay with this selection process, and not select nurses on their own to
meet the numbers the researcher requested. This letter requested also that the
administrators have Family Support Counselors (FSCs) available, as needed, for
participating nurses (Appendix H). Along with this cover letter they received research
packets for the participants.
Each participant was given a packet at the facility by the designated contact
person. The number of packets each facility received was based on the results of a
proportional sampling procedure. Each participating hospice provided the principal
investigator with the total number of eligible nurses in their organization. From these
reports, each hospice organization’s proportion of the total Florida eligible population
was calculated, adjusted for response rates, and constituted each organization’s
contribution to the study sample.
The participants received an explanatory cover letter indicating that submission of
the completed surveys would constitute informed consent (Appendix I). They were asked
to complete a demographic instrument (Appendix D), and the ProQOL-CSF-R-III tool
(Appendix C). Also, they had the researcher’s phone number on hand, and had access to
family support counselors as needed. In order to maintain confidentiality and provide
comfort, the participants were responsible for keeping their own documents secure, and
mailed their responses to the researcher in the provided addressed, stamped envelope.
As the researcher began to receive responses, she entered them into a data base. If
responses were not received within 3 weeks of the mailings, then the researcher made a
follow up phone call or visit (Appendix J) with the designated contact person from the
organization. When at least 20 completed responses were received from the demographic
45
questionnaire and the ProQOL-CSF-R-III, pilot analyses were completed for data entry
errors and analytical validation.
Data were de-identified and compiled using the Statistical Package for the Social
Sciences (SPSS) software program version 12. The data were analyzed and interpreted
with guidance from a statistician. Confidentiality was maintained, to the extent allowed
by law, during this entire process, and when not in use, the data were stored separately in
a secured cabinet at the researcher’s home.
Data Analysis
The following research questions were formulated for this study. A discussion of
the analytical design to address each question is also included.
Research Question 1
Research question 1 inquires of the demographic and work-related characteristics
of hospice nurses sampled for this study. For this research question, descriptive statistics
were used to summarize these demographic and work-related factors. For those
continuous variables which are interval in scale, central location was described with the
arithmetic mean and median. Score dispersion was described with ranges, interquartile
ranges, and standard deviations. The general distributional properties of these data were
expressed with frequencies, percentages, cumulative percentages, and skew. Central
location was illustrated, utilizing only the median for those variables for which the scale
of measurement can be argued to be ordinal, at best. Score dispersion was described with
ranges and interquartile ranges, and distributional properties were described with
frequencies and percentages. Finally, those variables which are discrete in nature and
nominal in scale, were illustrated with frequencies and percentages.
Research Question 2
The prevalence of compassion fatigue risk among the hospice nurses in Florida is
the focus of the second research question. This variable was treated as continuous and
interval in scale. For this research question, descriptive statistics including frequencies,
relative frequencies, and traditional summary descriptors were utilized.
Research Question 3
Research question 3 inquires about the nature of the relationship that hospice
nurses, demographic, and work-related variables have with the risk of compassion
46
fatigue. Pearson Product Moment (PPM) Correlation Coefficients were used for
continuous, interval scaled variables for which linearity can be reasonably argued.
Contingency Coefficients were generated for discrete variables, and Point Biserial
Coefficients were used for associations between continuous and discrete dichotomous
variables.
Research Question 4
The final research question focuses on the demographic and hospice work-related
factors which predict the risk of compassion fatigue. A multiple regression analysis was
used to predict the risk of compassion fatigue from demographic and hospice workrelated independent variables. These variables were selected subsequent to scrutiny of
correlation matrices containing the dependent variable and the prospective independent
variables.
A Global Test was conducted for the combined contributions of all independent
variables in the full model.
1.
The Null Hypothesis is H0: The independent variables will provide nothing
toward the perfect prediction of compassion fatigue risk. (H0: popR2 = 0)
2.
The Alternate (Researcher’s) Hypothesis is Ha: One or more of the independent
variables will provide a significant contribution to the prediction of compassion
fatigue risk. (Ha: popR2 ≠ 0)
Since hypotheses would be tested for this question, a minimally adequate sample size of
178 was determined after consideration of: alpha (0.01), power (0.90) and a moderate
effect size of 9% of the variance accounted for as referenced in Cohen’s table of sample
sizes (Cohen, 1990). In this study an alpha level of .01 states that 1% of all appropriate
statistical tests will lead the researcher to make a Type 1 error which is the probability of
rejecting H0 when H0 is true. In this study, a power level of .90 represents the relative
proportion of times in which the proper statistical test will result in a rejection of H0 when
H0 is false. Therefore, the conclusion associated with power leads to a definite claim that
the alternate hypothesis (Ha ) is true. Effect size (ES) is the value which indicates the
degree, or magnitude, of falsity of H0 when H0 is false (Brewer & Workman, 2003). It
specifies the size of the difference between H0 and Ha; therefore, an ES of R2 equal to
0.09 states that the independent variables, or the selected independent variable, must
47
account for at least 9% of the variance of the dependent variable (compassion fatigue
risk) in order for the researcher’s regression equation to be of practical importance. The
ES chosen for this study was based on Cohen’s definition for a moderate effect in a
regression analysis (Cohen, 1990).
Pending statistical significance of the Global test, Sub Globals were conducted to
test the unique contributions of each of the independent variables. For the Global and Sub
Global tests of hypotheses, the following analytical assumptions are required:
1. The dependent variable is normally distributed.
2. Linear relationships exist between dependent and independent variables.
The researcher provided descriptive statistics and distributional descriptions to give the
reader comfort with the degree to which these first two assumptions are tenable.
3. The scores on the independent variables are known without error. This
assumption is required for a regression analysis. The selected instruments were as valid
as possible.
4. The observations are randomly selected. This assumption was met by this
study.
5. Observations are independent. There is no statistical test to determine whether
or not the observations, once selected, are independent of each other; however, the
random selection of the participants maximized the probability that the observations were
independent. A possibility for failure existed by virtue of the fact that the sample
consisted of workmates.
Other analyses were performed as deemed appropriate and necessary subsequent
to scrutiny of the data. Results of all analyses will be discussed in Chapter 4.
Summary
The purpose of this study was to describe the prevalence of compassion fatigue
risk in hospice nurses, to explore the relationship between various nurse characteristics
and the risk of compassion fatigue, and finally, to provide a model for predicting the risk
of compassion fatigue from knowledge of demographic and work-related factors. A nonexperimental descriptive design utilizing cross-sectional data, as well as descriptive and
inferential statistics, was used to accomplish these objectives. Research questions 1, 2,
and 3 are all descriptive and were answered utilizing analytical tools appropriate for the
48
nature of the variables and their scale of measurement. Research question 4 is inferential
and was addressed with the construction of a linear regression model for the prediction of
the dependent variable (risk of compassion fatigue). Demographic and work-related
factors were selected as independent variables for the model, subsequent to scrutiny of
linear associations between these factors and the dependent variable. It is hoped that the
findings in this study will be useful in providing a means of identifying and predicting the
risk of compassion fatigue in the Florida hospice nurse population. A comprehensive
discussion will follow in Chapter 4 that will present the results of the data analyses.
49
CHAPTER 4
RESULTS
This inquiry investigated the prevalence of compassion fatigue risk within a
population of Florida Hospice nurses. Demographic and work-related factors were also
considered for their associations with risk of compassion fatigue (CF). Subsequent to
consideration of the strength of such associations, a multiple regression model was
constructed to determine the combined and individual contributions of these and other
variables toward the prediction of CF Risk for the population under investigation. The
study was guided by Neuman’s Systems Model (1995), and Selye’s Stress Theory (1956).
Figley’s Compassion Stress and Fatigue Model (1995) was likewise incorporated to
provide further conceptual support for the framework undergirding this study.
This chapter addresses the statistical findings of the inquiry. Both descriptive and
inferential techniques were chosen on the basis of the nature of the data provided by the
instruments, as well as the degree to which each analytical tool was deemed appropriate
for the research questions and study context.
Description of the Sample
Study Population and Sample
A sample of 433 hospice nurses was selected at random from a target population
of 1744 nurses employed by 17 of 40 (43%) Florida Hospice organizations agreeing to
participate in the study. The sample of nurses from each hospice was selected, utilizing a
stratified random sampling technique, in order to provide a statewide representative
sample of hospice nurses. These registered nurses (RNs), advanced registered nurse
practitioners (ARNPs), and licensed practical nurses (LPNs) met the following inclusion
criteria: a) 18 years of age or older, b) employed by a hospice organization in Florida, and
c) interact directly with patients and their families. They worked in any area of the
organization such as: a) freestanding inpatient hospice facility care, b) home care, and c)
hospice admissions.
50
Potential respondents were instructed to complete two instruments: The
Professional Quality of Life Compassion Satisfaction and Fatigue Subscales: R-III
(ProQOL; Appendix C) and a demographic instrument (Appendix D). Designated contact
persons from each hospice organization distributed packets containing the instruments, a
cover letter and a postage-paid return envelope to the randomly selected nurses in their
respective agencies. One hundred eighty-seven (43%) nurses returned the instruments
within an 8-week time frame. One hundred sixty-six of those returned, provided usable
data.
Supplemental Data Venue
Subsequent to the initial data collection period, approximately 93% of the
minimally adequate observations (166/178) had been obtained. In order to demonstrate
support for the study as well as increase suitable responses, Florida Hospices and
Palliative Care, Inc., the state hospice organization, encouraged nurses to participate in
the study during the 20th Annual Florida Hospices and Palliative Care Symposium in
Palm Beach, Florida on December 13 – 15, 2004. Permission to utilize these additional
data was obtained from the investigator’s supervisory committee, as well as the
Institutional Review Board for the host university (Appendix A). A cover letter, and the
two instruments were distributed to 150 nurses attending the symposium (see Appendices
K, L, and M for the conference materials). Attendees were asked to complete the
instruments and place them in a designated lock box at the conference site. Seventy-six
(51%) participants returned the instruments during the 3-day conference, and 50 of the
returned instruments provided usable data. Data obtained from the two venues (mail
versus conference) were analyzed to determine the extent to which the data collection
venue was differentially associated with all relevant variables and, thus, potentially
confounding the final results.
Venue was tested for continuous variables with independent samples t-tests (see
Table 4.1). Venue (k = 2) was likewise tested for discrete variables in contingency tables
utilizing Chi-square and Fisher Exact (2 X 2 contingency table of frequency data) tests.
The data were scrutinized for the degree to which the assumptions required for these tests
were tenable.
51
Table 4.1.
Independent Samples t-tests of Venue Ranked by Size of p Value
Continuous Variables ranked
Mean Std Error of
by p value
Difference
Difference
Average nurse patient ratio per day
-1.283
0.369
Average hours worked per week
-5.264
1.680
Compassion fatigue risk
-1.859
1.059
Years professional nursing
2.018
1.837
experience
Burnout risk
-0.875
0.958
Average cases per week
9.140
10.123
Years hospice experience
-0.309
0.814
Compassion satisfaction
-0.363
0.974
Patient deaths in past 30 days
-0.136
1.472
*Statistically significant at the .01 level of alpha
tc
df
p
-3.480 57 0.001*
-3.133 213 0.002*
-1.756 214 0.081
1.098 212 0.273
-0.914
0.903
-0.379
-0.372
-0.092
214
145
214
214
206
0.362
0.368
0.705
0.710
0.927
The homogeneity of variance assumption, required for this statistical test was
deemed tenable for all variables except burnout risk (Fc = 4.116, p = .044), and the reader
will note that two of the above variables tested statistically significant for venue. The
first, average nurse/patient ratio, can be explained in terms of sample size differences for
the two groups (nmail = 45, nconference = 14) as well as differences between the groups with
respect to how this item on the instrument was answered. In-patient hospice facility
nurses were instructed to provide responses to this question whereas field nurses were
instructed to answer a similar question which asked for average caseload of patients per
week. Despite the fact that this variable tested statistically significant, the result was not
considered detrimental to the objectives of the study due to the fact that the hospice work
setting variable (see table 4.2) did not test significant for the two venue groups (χc =
8.841, df = 8, p = .356).
The other statistically significant variable in this analysis (average hours worked
per week), can be similarly explained. Sample sizes for the two groups (nmail = 165,
nconference = 50) were sufficiently different to have a deleterious effect on the test statistic
and subsequent p value even though the mean difference on this variable was relatively
small. Similarly, this variable correlated only minimally with Compassion Satisfaction,
Burnout and Compassion Fatigue Risk (r = .121, r = .175 and r = .130 respectively). For
these reasons, both average caseload of patients per week and average hours worked per
52
week were not considered sufficiently different for the two data collection venues to
warrant their exclusion in further analyses nor the consideration of venue as confounding
for the study context.
More than 20 discrete variables were likewise tested for differences with respect
to venue (see table 4.2). These tests were conducted utilizing Chi-Square analyses of data
in contingency tables. Fisher Exact Tests were conducted for 2X2 contingency tables.
Table 4.2
Chi-square and Fisher Exact tests of Venue for Selected Discrete Variables
Ranked by Size of p Value
df
p (non-directional)
Discrete Variables
χ c2
PRN employee (Yes/No)
Full or Part Time
Involved in shift work
Caring for loved one
Self sacrifice
Stress from finances
Staff works as team
Marital Status
Work rotating shifts
Work setting
Highest Level of Nursing Ed
Level of care
Experience death loved one
Specialization certification
Discrete Variables
8.51
8.51
9.72
2.56
1.88
1.80
4.61
5.80
1.61
8.84
4.30
11.22
0.50
3.09
χ c2
Ethnicity
Experienced pt traumatic death
Nursing Professional licensure
Gender
Depression/PTSD
Contract employee
Smoking behavior
Experience headaches
Hypertension
* Fisher Exact p values
3.750
1.010
0.659
0.300
0.186
0.303
0.007
0.002
0.000
1.00
1.00
6.00
1.00
1.00
1.00
3.00
4.00
1.00
8.00
4.00
12.00
1.00
4.00
df
5
2
2
1
1
1
1
1
1
0.002*
0.002*
0.14
0.157*
0.183*
0.198*
0.20
0.21
0.29
0.36
0.37
0.51
0.515*
0.54
p (non-directional)
0.586
0.601
0.719
0.737*
0.846*
1.00*
1.00*
1.00*
1.00*
It was noted that two discrete variables on which the groups differed significantly
for venue, was whether or not the nurses were a PRN employee and employment status
(full vs. part-time). The size of the test statistics for these variables and hence, the p
53
values, can be explained in terms of severe differences in sample sizes (n1 = 25, n2 =
191). These variables were also analyzed to determine the extent to which they may have
had an undue influence on the three dependent variables: compassion satisfaction,
burnout risk, and compassion fatigue risk, and differences were determined to be minimal
(see Table 4.3).
Table 4.3
Group Statistics on Dependent Variables for PRN Employee and Full/Part-time
PRN
Std. Std. Error
Employee
n Mean Deviation
Mean
191 40.95
5.90
.43
Compassion Satisfaction No
Yes
25 40.20
7.00
1.40
191
25
20.65
18.00
6.12
3.50
.44
.70
191
25
14.03
10.96
6.64
5.62
.48
1.12
191
25
41.13
38.84
5.93
6.49
.43
1.30
Full-time
Part-time
191
25
20.70
17.64
6.08
3.73
.44
.75
Compassion Fatigue Risk Full-time
Part-time
191
25
14.12
10.28
6.67
4.83
.48
.97
Burnout Risk
No
Yes
Compassion Fatigue Risk No
Yes
Full/Part Time
Compassion Satisfaction Full-time
Part-time
Burnout Risk
The median compassion fatigue risk values for full-time and part-time employees
were 14 and 11, respectively. Both values were within the moderate risk category (range
= 8-17) defined by the ProQOL. The difference between the groups with respect to
employment status was, therefore, deemed acceptable for the given research context.
Given the foregoing results, data from the two venues were combined into a
single sample of 216 (37%) usable instruments which exceeded the study’s minimally
adequate sample size of 178 by approximately 20%. For purposes of clarity the following
results are reported in the order and terms of each research question specified in Chapter
3
54
Research Question One
Research question one inquired of the demographic and work-related
characteristics of the hospice nurses sampled for the study. Each of these variable
subgroups will be described in separate sections.
Demographic Description of the Sample
Demographic data were obtained utilizing an instrument developed by the
investigator and deemed face and content valid by measurement and content experts.
Internal consistency reliability for the ProQOL (All items = .662) as well as for its 3
subscales (Compassion Satisfaction = .862; Burnout = .685; Compassion Fatigue = .805),
was determined utilizing Crombach’s Alpha provided by SPSS version 12.
Participants (n = 216) were predominantly female (n = 204, 94%), and Caucasian
(n = 177, 81.9%). Their ages ranged from 23 to 76 years, with approximately 25% of the
sample, younger than, or equal to, age 45, and the median age being 50.5 years (mean =
53.9, SD = 9.05). The cutoff age for the upper quartile was 56 years and 30 participants
(11%) were 60 years of age or older. The interquartile range was 11 years. The males (n
= 12) in the sample were slightly older (meanmales = 53.7, meanfemales = 50.2) than the
females. Similarly, the ages for both African Americans (mean = 43.65) and Asians
(mean = 46.00) was substantially less than that of the white non Hispanic participants
(mean = 51.24). With respect to marital status, the age distributions were as one might
expect with single nurses (mean = 45.33) being the youngest, widowed nurses (mean =
61.75) the oldest, and those in the other three categories approximating the mean age for
the total sample. The majority of participants in the sample were married (59.7%, n =
129) and approximately a quarter (23.1%, n = 50) were divorced. Most (57.9%, n = 125)
had no children living at home. Almost all (n = 9, 75%) of the 12 males in the sample
were married with only one reported being single, whereas approximately 59% of the
females were married. The reader can refer to Table 4.4 for a more detailed univariate
description of the demographics for the study participants.
55
Table 4.4
Univariate Descriptions of Demographic Factors
Variable
Age
Factor
Gender
Male
Female
Ethnicity
While/Non Hispanic
African American
Hispanic
Native American
Asian
Other
Marital Status
Single
Married
Separated
Divorced
Widowed
Children at Home
0
1
2
3
4
Mean/Median
Std. Dev. Min Max Skew
50.39/50.50
9.06
23
76 -.149
Frequency Percentage
(n)
(%)
12
204
5.6%
94.4%
177
20
6
2
5
6
81.9%
9.3%
2.8%
.9%
2.3%
2.8%
24
129
5
50
8
11.1%
59.7%
2.3%
23.1%
3.7%
125
38
37
11
3
58.4%
17.8%
17.3%
5.1%
1.4%
Work-related Description of the Sample
One-hundred one (46.8%) of the nurses had Associate Degrees in Nursing,
whereas, only nine (4.2%) were Masters prepared. Two of the nine nurses with Masters
degrees were ARNPs. There were two additional ARNPs who were Baccalaureate
prepared rather than having Masters or higher degrees. The RNs (n = 183, 84.7%) were
the largest group represented in the sample, which was not unusual because this group
encompassed nurses with the Associate, Diploma, Baccalaureate, and Masters degrees.
The nurses with backgrounds from vocational/technical schools were LPNs (n = 29,
13.4%). However, three of those LPNs noted coming from diploma programs, and one
indicated having an Associate degree in nursing. Sixty-one (28.2%) of the individuals in
this study had one or more specialty certifications. Forty-four (24.1%) of the RNs were
56
certified hospice and palliative nurses (CHPNs), and six (20.7%) of the LPNs were
certified hospice and palliative licensed nurses (CHPLNs). The majority of the
participants who held these credentials had more experience as hospice nurses, and had
been in the nursing profession longer than those without hospice credentials. Table 4.5
provides a more detailed univariate description of the work-related characteristics of the
participants.
Table 4.5
Univariate Description of Experience, Education and Licensure
Standard
Variable
Mean/Median Deviation Min Max
20.19 / 20.00
11.37 1.00 55.00
Years Professional Nursing
Experience
5.65 / 04.00
5.03
.08 24.00
Years Hospice Nursing
Experience
Frequency Percentage
Factor
(n)
(%)
Nursing Educational Background
Master’s
9
4.2%
Baccalaureate
45
20.8%
Diploma
36
16.7%
Associate
101
46.8%
Practical / Vocational
25
11.6%
Licensure
ARNP
4
1.9%
RN
183
84.7%
LPN
29
13.4%
Certification
CHPN (RN)
44
20.3%
CHPLN (LPN)
6
20.7%
Skew
.51
1.43
More than half (n = 130, 60.2%) of the nurses were field nurses who provided
hospice care in private homes, and made visits to hospice patients in nursing homes. The
field nurses in this category also worked in continuous care environments where they
provided one-on-one care to a patient in the home for an 8 – 12 hour period. One quarter
of the nurses in the study (n = 50, 23.1%) worked in in-patient hospice units, whereas,
other nurses worked as new patient admissions nurses, which required them to provide
services in hospitals, nursing facilities, and private homes to patients and their families.
57
Table 4.6 categorizes the participants in their hospice setting. There was no difference in
the number of years of professional nursing experience between the field and inpatient
unit nurses, and there was only a slight difference between inpatient unit nurses (n = 50,
M = 6.12, md = 4.75) and field nurses (n = 130, M =5.61, md = 3.75) with respect to
years of hospice experience. Additionally, both-full time (M = 19.76, md = 20.0), and
part-time (M = 23.44, md = 20.0) nurses had the same number of years of professional
nursing experience (M = 19.76, md = 20.00).
The nurses’ average hours worked per week was 40.5 (md = 40, SD = 10.6) for
the study participants; however, the weekly time worked ranged from 8 – 90 hours. Field
nurses tended to work slightly longer hours (M = 42.07, md = 40.0, SD = 10.4) than the
nurses in the inpatient units (M =36.4, md = 37.5, SD = 8.93); however, the nurses who
provided care in multiple settings worked the longest hours. For example, Field/Inpatient
unit nurses worked an average of 48.8 hours/week (M =48.8, md = 50, SD = 8.07). See
Table 4.6 for a univariate description of these variables.
The nurses who did not work in an inpatient unit reported having an average
weekly caseload of 19.8 (md = 13, SD = 54) patients. One of the triage nurses reported a
caseload of 500 which probably skewed the results of the mean value for this specific
average weekly caseload analysis. The participants who provided care in an inpatient unit
noted their average care ratio was one nurse to approximately 6 patients (M = 5.59, md =
6, SD = 1.31). The nurses were specifically asked about shift work. Approximately 41%
(n = 86) of the participants said they worked shifts, and the 12 hour shift (7 am–7 p.m.)
was the most frequently reported (n = 28, 32.2%) among the 87 participants responding
to this question. Additionally, 36 (41.4%) members of this group of nurses who worked
shifts stated they worked rotating shifts.
The majority of the nurses in the study (n = 137, 63.4%) felt that they worked as a
team most of the time. A larger percentage of the 50 inpatient hospice nurses noted they
worked well as a team only some of the time (22.0%, n = 11) compared to the 130 field
nurses who thought they worked as a team only some of the time ( 9.2%, n = 12). A
larger percentage of the field nurses (22.3%, n = 29) indicated that they worked as a team
all of the time. Table 4.6 provides a synopsis of the various team work categories.
58
Table 4.6
Description of Work Setting Factors
Variable
Mean/Median
Standard
Deviation
40.56 / 40.00
10.61
8.00
19.84 / 13.00
54.15
1.00
5.59 / 06.00
1.31
2.00
Average hours worked
per week (n =215)
Average Caseload per
week (n =147)
Average nurse/patient
ratio per day (n = 59)
Min Max Skew
90
-.340
500 8.310
10
.521
Frequency Percentage
(n)
(%)
Variable
Work Setting
Field
Inpatient hospice unit
Admissions (hospital, home, nursing home)
Triage (telephone contact) with patients in
all settings
Combination: Field/Inpatient
Combination: Field/Admissions
Other combinations
Shift Work (n = 86)
7am-3pm
3pm – 11pm
11p -7am
7am- 7pm
7pm -7a.
Teamwork (n = 216)
All of the time
Most of the time
Some of the time
Rarely
130
50
16
5
60.2%
23.1%
7.4%
2.3%
5
5
5
2.3%
2.3%
2.4%
22
14
3
28
19
25.3%
16.1%
3.4%
32.2%
21.8%
45
137
33
1
20.8%
63.4%
15.3%
.5%
Twenty-seven (12.6%) of all the nurses in this study held other jobs in addition to
hospice employment. The predominant type of other employment was direct patient care
nursing in other settings. Fourteen (7.4%) of the 189 full-time hospice nurses had other
concurrent employment, compared to thirteen (52%) of the 25 part-time nurses. The total
number of hours worked per week, regardless of employment status (full vs. part-time),
was essentially equivalent. Table 4.7 provides a more detailed description of hospice
nurses with concurrent employment.
59
Table 4.7
Hospice Nurses with Other Concurrent Employment
Variable
Full -Time Hospice Nurses
(n = 14)
Hours worked per week
Other hours per week
Part-Time Hospice Nurses
(n = 12)
Hours worked per week
Other hours per week
Mean/Median
Standard
Deviation
Min
Max
Skew
41.64/42.00
13.69/12.00
6.00
6.27
32.00
5.00
50.00
24.00
-.213
.760
15.37/16.00
31.83/40.00
5.04
11.66
8.00
8.00
24.00
40.00
.031
-1.181
Research Question Two
Research question two inquired of the prevalence of the risk of compassion
fatigue among the hospice nurses in the state of Florida. The purpose of this section is to
describe not only prevalence of compassion fatigue for the entire sample, but also to
identify specific subgroups for which such risk exists. The reader is reminded that
compassion fatigue risk was operationalized with the ProQOL, and the authors of this
instrument also provided a scoring process by which score ranges could be categorized
into minimum, moderate, and high levels for compassion fatigue risk, and the other two
variables (burnout risk and compassion satisfaction). Score ranges for the various levels
of all three subscales are defined in Table 4.8
Table 4.8
Levels of Risk for ProQOL Subscales
Level
low
moderate
high
Score Ranges
Compassion Fatigue Risk Burnout Risk
< 18
<7
8-17
19-28
> 18
> 29
Compassion Satisfaction
< 31
32-41
> 42
The reader will note in Table 4.8, that there are extreme scores for all three
subscales designated “low”, “moderate”, and “high” risk. This is the case despite the fact
that Compassion Satisfaction is negatively correlated (as one would expect) with both
Compassion Fatigue and Burnout. The authors of ProQOL constructed the items on this
60
instrument so that it could be easily self-scored and the results identifying the participant
as being in either a “low,” “moderate” or “high” risk/satisfaction category. This scoring
plan is independent of the negative association, as the latter resulted (see results for
Research Question Three) simply from a predominance of scores on Compassion Fatigue
and Burnout that were above the mean, being paired with scores on Compassion
Satisfaction that were below the mean, and vice versa.
Prevalence of Compassion Fatigue Risk
The nurses in this study exhibited scores mainly in the moderate category of
Compassion Fatigue Risk. This finding is not surprising inasmuch as the instrument’s
authors used quartile splits with their norming group to define the three risk categories.
For the entire sample, the median score for compassion fatigue risk was 14 (mean = 13.6,
SD = 6.59). The minimum score was one, and the maximum score was 39. Also, 21.3%
of the sample (n = 46) produced scores that placed them in the low category. One
hundred-thirteen participants (52.3%) were in the moderate category, and 57 (26.4%)
were in the high risk category. The reader should note that this sample had fewer
participants in the lower risk category and more in the higher risk category than was the
case for the norming group described by the authors of the ProQOL. Table 4.9 provides
further descriptions of the sample with respect to their demographic traits and risk
category.
The reader will note that a greater proportion of males (n = 4, 33.3%) were in the
high risk category for compassion fatigue risk, than were the females (n = 53, 26%).
Also, despite the fact that the sample sizes are low for the non-white ethnicities, African
Americans were disproportionately categorized in the moderate to low risk categories,
while the Hispanics and Asians were disproportionately classified in the moderate to high
risk categories.
61
Table 4.9
Compassion Fatigue Risk by Gender, Ethnicity, and Marital Status
Variable
Compassion Fatigue Risk Levels
Low
Moderate
High
Freq. Percentage Freq. Percentage Freq. Percentage
(n)
(%)
(n)
(%)
(n)
(%)
Gender
Male
Female
Total
Ethnicity
White/Non Hispanic
African American
Hispanic
Native American
Asian
Other
Marital Status
Single
Married
Separated
Divorced
Widowed
3
43
46
25.0%
21.1%
21.3%
5
108
113
41.7%
52.9%
52.3%
4
53
57
33.3%
26.0%
26.4%
37
7
0
0
0
2
20.9%
35.0%
0%
0%
0%
33.3%
95
9
4
1
1
3
53.7%
45.0%
66.7%
50.0%
20.0%
50.0%
45
4
2
1
4
1
25.4%
20.0%
33.3%
50.0%
80.0%
16.7%
6
27
2
9
2
25.0%
20.9%
40.0%
18.0%
25.0%
16
67
2
24
4
66.7%
51.9%
40.0%
48.0%
50.0%
2
35
1
17
2
8.2%
27.1%
20.0%
34.0%
25.0%
Finally, the age of participants in each of the three risk categories showed little
variability (Medianlow = 52.0, Medianmoderate = 51.0, Medianhigh = 50.0). There was also
very little difference between the groups with respect to nurses’ years of professional
nursing experience (Medianlow = 16.0, Medianmoderate = 20.0, Medianhigh = 20.0) and years
hospice experience (Medianlow = 4.0, Medianmoderate = 4.5, Medianhigh = 3.0). Table 4.10
further describes demographic and work-related characteristics in relation to compassion
fatigue risk categories.
62
Table 4.10
Descriptives for Compassion Fatigue Risk by Selected Work-related Factors
n
Mean
Median
Standard
Deviation
LPN
RN
ARNP
29
183
4
14.28
13.60
12.50
14.00
14.28
12.50
6.74
6.53
10.08
Nursing Education
Vocational Training
Associate Degree
Diploma
Baccalaureate
Master’s
25
101
36
45
9
14.24
13.21
13.33
14.82
12.77
14.00
13.00
13.50
15.00
16.00
7.23
6.56
5.40
7.07
7.59
Specialization / Certification
CHPLN
CHPN
6
6
13.83
13.84
14.00
14.00
5.74
5.74
Variable
Licensure
Research Question Three
Research Question three inquired about the relationship between demographic,
hospice work-related factors, and the risk of compassion fatigue. It was decided to
include the nurses’ personal health factors in this section due to the nature of these
questions on the demographic instrument, and their expected relationship to stress, and
the stress response. Associations between these factors and the other two subscales of the
ProQOL are also included where informative.
While the signs of the intercorrelations between the subscales of the ProQOL
were as expected, the sizes of those intercorrelations were less than anticipated (see Table
4.11).
Table 4.11
Correlation Matrix of Intercorrelations Between ProQOL Subscales
Compassion
Burnout
Compassion
Satisfaction
Risk
Fatigue Risk
Compassion Satisfaction
1.000
Burnout Risk
-.532
1.000
Compassion Fatigue Risk
-.362
.659
63
1.000
The negative correlations between compassion satisfaction and the other two
factors indicates that in general, scores that are above the mean on compassion
satisfaction are paired with scores that are below the mean on the other two factors and
vice versa. Likewise the positive correlation between burnout risk and compassion
fatigue risk indicated that study participants were either above the mean on both variables
or below the mean on both. The size of the correlation between burnout risk and
compassion fatigue risk suggests that if this value were squared (r2 = .43), approximately
43% of the information that is needed to predict perfectly compassion fatigue risk is
being provided by knowledge of a participant’s burnout risk score, and thus
approximately 57% of such information can be attributed to other variables (1 – r2 = .57)
Associations Between Demographic and ProQOL Factors
There was virtually no correlation between participants’ age and their scores on
compassion fatigue risk (r = -.04), burnout risk (r = -.097), and compassion satisfaction
(r = .076). Likewise, the correlations between the three subscale variables and the other
demographic factors were low, as noted in Table 4.12
Table 4.12
Correlation Values: Demographics and ProQOL Subscale Variables
Correlations
Subscale Variables
Demographic Variables
CF
BO
CS
Ethnicity
0.203
0.209
0.068
Marital status
0.149
0.150
0.044
Children at home
0.132
0.169
0.209
Highest level of nursing education
0.103
0.149
- 0.136
Certification/Specialization
0.064
0.099
0.082
Licensure
0.043
0.116
0.100
*Gender
0.009
- 0.075
- 0.059
* Point Biserial Correlation Coefficients. All other values are Eta Correlation coefficients
Further, when subsets of the sample were analyzed, these correlations remained
surprisingly low. However, the negative correlation between burnout risk and age
64
supported the literature which noted that younger individuals are at a higher risk for
burnout.
Work-Related Factors
Correlations did exist between compassion fatigue risk and whether or not nurses
had experienced a patient’s traumatic death (reta = 0.244), burnout risk (reta = 0.323) and
compassion satisfaction (reta = 0.187). While these values were less than expected, the
categorical (k = 3) nature of the traumatic death variable could have been partially
responsible for such values. Studies have noted that burnout and stress levels rise when
nurses do not receive psychological/social support after this type of stressor (Adams,
Hershatter, & Motitz, 1991). A subset of the sample (n = 77) was selected on the basis of
the nurses responding “no” (n = 40) or “yes; with no support given” (n = 37) to having
experienced a patient’s traumatic death. This variable was then correlated with the three
ProQOL subscale variables. The results indicated higher Point Bi-serial correlations for
compassion fatigue risk (r = .360), burnout risk (r = .490), and compassion satisfaction (r
= -.257) than the previous correlations. The size of the correlation between burnout risk
and the subgroup of participants who experienced patients’ traumatic death, but did not
receive support, suggests that if this value were squared (r2 = .24), approximately 24% of
the information needed to predict perfectly burnout risk is being provided by knowledge
of whether or not a nurse had experienced a patient’s traumatic death and failed to
receive psychosocial support.
The investigator’s review of the literature (Kulbe, 2001; Masterson-Allen, Mor,
Laliberte, & Monteiro, 1985) revealed that nurses’ work-related variables such as: long
work hours, high patient caseloads, multiple deaths occurring within a short period of
time, and shift work have stressful effects on the individual which can then lead to
burnout and compassion fatigue. While the correlations reported in Table 4.13 support
earlier findings, the size of the correlations was less than expected.
65
Table 4.13
Correlation values: Work-related Factors and Dependent Variables
Correlations
ProQOL Subscale
Variables
CF
BO
CS
0.244
0.323
0.187
0.212
0.199
0.170
0.133
0.331
0.242
0.130
0.175
0.121
0.128
0.165
0.150
0.105
0.015 -0.006
0.103
0.180
0.170
0.024
-0.041 -0.043
0.008
-0.014
0.168
-0.039
-0.006
0.098
Nurse Work-related Variables
**Experienced patient traumatic death
**Level of Care
**Staff works as a team
Average hours work per week
**Work setting
Average nurse/patient ratio
**Shift work
Years professional nursing experience
Years hospice experience
Average caseload
Number of patient deaths exposed to in last 30
days
-0.032
-0.037
*Contract employee
-0.080
-0.062
*PRN employee
-0.149
-0.143
*Full or part time
-0.186
-0.165
* Point Biserial Correlation Coefficients
** Eta Correlation Coefficients
All others are Pearson Product Moment Correlation Coefficients
0.022
0.070
-0.040
-0.122
Five of the work-related variables were selected within the ProQOL instrument
and analyzed for their associations with the three ProQOL subscale variables, shown in
Table 4.14. Being overwhelmed and losing sleep due to nurses’ work environment, and
trauma is supported by theory as trigger points to stress, burnout, and compassion fatigue
(Selye, 1976).
Table 4.14
PPM Correlations between ProQOL Subscale Variables and Selected ProQOL Items
Correlations
Independent Variables
“Infected” by traumatic stress
Being “on edge” due to helping
Losing sleep over patients’ trauma
Feelings of being overwhelmed by work and caseload
Feelings of being “bogged down” by the system
66
ProQOL Subscale
Variables
CF
BO
CS
0.709 0.487
-0.234
0.703 0.596
-0.348
0.645 0.437
-0.140
0.475 0.654
-0.220
0.437 0.670
-0.420
Personal Health Description of the Sample
One hundred-sixteen (53.7%) of the nurses from all ethnic backgrounds who
participated in this study, claimed that personal finances was a source of stress . Almost
18% (n = 38) were cigarette smokers, and more than a quarter of the nurses had frequent
headaches (n = 61, 28.2%). Sixty-five (30.1%) had been diagnosed with hypertension at
the time of the study and approximately 22% (n = 48) were diagnosed with depression or
post traumatic stress disorder (PTSD). Additionally, 21% (n = 44) were taking care of an
elderly or disabled parent or loved one.
There was virtually no difference in compassion fatigue risk categories among
cigarette smokers, and those caring for a loved one. However, 35% (n = 17) of those
diagnosed with depression or PTSD were in the high risk CF category as opposed to 24%
(n = 40) among those who did not have depression or PTSD. A greater proportion of
those who had financial stress (n = 36, 31%), and those who experienced frequent
headaches (n = 22, 36.1%) were also in the high risk category for compassion fatigue
risk.
Nurses who are inclined to put their patients’ needs ahead of their own may be
prone to paying attention to their own needs last in other areas of their lives. This
characteristic can lead to stress. One hundred thirty-six participants (63.8%) stated they
had a tendency to sacrifice their own psychological needs to satisfy the needs of their
patients. In this 136 participant sub-sample, 50% (n = 107) were Caucasian, and 8% (n =
17) were African American. However, there were 20 African Americans within the entire
216-participant sample; therefore, those 17 who tended to self sacrifice represented 85%
of the African American ethnic group. Eighty-two nurses (60.3%) from the self sacrifice
sub-sample had more financial stress than the group who indicated that they did not
sacrifice their own needs for their patients’ needs. Additionally, these 136 nurses had
fewer years of hospice experience (M = 4.9, md = 3.0) than the participants who did not
have that tendency to sacrifice (M = 6.9, md = 5.0); however, there was virtually no
difference in age between the nurses who self-sacrificed and the ones who did not. The
group that claimed to self-sacrifice had a much higher proportion of nurses in the high
67
risk category for CF (nhigh risk = 47, 34%) than those nurses who answered “no” to the self
sacrifice question (nhigh risk = 10, 13%).
Hospice nurses encounter more patient deaths than nurses in many other
specialties. Repeated exposure to suffering can lead to the risk of compassion fatigue.
Within a 30-day period, they were exposed to an average of 7 deaths (md = 5.00, SD
8.85), and within the last year almost 42% (n = 90) experienced the death of someone
close to them.
According to the work of Selye, Neuman, and Figley, the independent variables
listed in table 4.15 are stressors that may lead to, or are, an outcome of compassion
fatigue risk and burnout risk. Researchers have reported that nurses who identify too
closely with dying patients have a tendency toward stress and burnout (Riggio, & Taylor,
2000). The less than expected size of the correlations in Table 4.15 can again, be partially
explained by the categorical nature of the independent variables.
Table 4.15
Correlation between Personal Health Factors and Pro-QOL Subscale Variables
Correlations
Personal Health Variables
ProQOL Subscale Variables
CF
BO
CS
Self sacrifice for patients’ needs
0.313
0.264
-0.178
Financial stress
0.282
0.345
-0.216
Headaches
0.217
0.246
-0.163
Depression / PTSD diagnosis
0.213
0.219
-0.078
Outside employment
0.106
0.093
0.118
Hypertension diagnosis
0.076
0.035
-0.015
Death of loved one in past year
0.058
0.036
0.030
Smoking behavior
0.040
0.053
0.069
Caring for a loved one
0.004
0.022
0.018
*Hours per week outside employment
-0.095
-0.081
-0.122
* Pearson Product Moment Correlation Coefficients
All others are Point Biserial Coefficients
Research Question Four
This question inquired of the demographic and hospice work-related factors
which, when included in a multiple regression model, might be used to predict
Compassion Fatigue Risk. The objectives for this question are two-fold. First, there has
68
been sufficient research on correlates with Compassion Fatigue and its close ally,
Burnout, to warrant investigation of the combined and unique contributions of these
variables in a prediction equation. Second, theorists such as Figley, Selye, and Neuman
have provided sufficient evidence to warrant both a test of their theories as they apply to
the current research context as well as a closer inspection of the previous synthesis of
their theories (see Chapter 1) grounded in the present inquiry’s data.
This question will be answered utilizing Multiple Regression (one dependent
variable; multiple independent variables). The results will be both descriptive and
inferential and for the latter, will employ a significance criterion of .01. Power for the
tests of significance, was set at .90 and effect size was set at 9% (r2 = 0.09). The minimal
sample size (n = 178) determined for the inferences, was obtained from Cohen’s sample
size tables (Cohen, 1990). As it was not known at the time of sample size determination,
how many independent variables (k) would be necessary to explain best the variance in
Compassion Fatigue Risk, an arbitrary estimate of k = 5 was used for the sample size
determination.
The logic employed for the regression analyses included: a) scrutiny of the
correlation matrices for Compassion Fatigue Risk, (CFR, with the Demographic, Workrelated and Personal Health variables as well as individual items on the ProQOL
instrument; b) ranking the correlates with CFR from largest to smallest, c) eliminating
from consideration, potential independent variables (IVs) which demonstrated sufficient
intercorrelations to induce contamination by multicolinearity, d) testing of global
contributions of multiple IVs in the full model and finally, e) testing of the unique
contributions (sub-global) of single IVs while controlling for the effect of the other IVs in
the full model.
The assumptions required for the Multiple Regression analysis are:
1. the distribution of error terms is Normal
2. the dependent variable (DV), Compassion Fatigue Risk, is continuous in nature and
interval in scale
3. independent variable (IV) scores are known without error
4. the sample of scores is randomly selected and
5. the scores obtained from each case (participant) are independent.
69
The fourth assumption was met for the participants included in the mail out. However,
data obtained from the nurses attending the 20th Annual Florida Hospices and Palliative
Care Symposium, were not random, but sufficient analyses were conducted to ensure that
these data would not contaminate the results. A similar explanation can be used for the
assumption of independence. Scatterplots of the standardized residuals were scrutinized
for each regression analysis for evidence of the degree to which the Normality
assumption was tenable for the resulting inferences. No evidence was found that
suggested a deviation from this assumption. The investigator is reasonably comfortable
with the requirements of the second assumption, due to the nature of the DV, as well as
the derivation of the Pro QOL scores. The third assumption, that the IV scores must be
known without error, is a requirement of Multiple Regression since these regression
models have only one error term. The only allowable source of error, therefore, is
measurement error in the DV (Compassion Fatigue Risk). Since neither of the
instruments utilized in this inquiry can be assumed to have provided perfectly valid
responses, the potential for erroneous responses from the participants must be included as
a limitation of the study.
Theoretical Contributions to the Prediction of Compassion Fatigue Risk
The models and theories of Figley, Neuman, and Selye were analyzed for their
potential contribution to a composite explanatory model for predicting Compassion
Fatigue Risk. To accomplish this empirical objective, variables originating in the
Demographic/Work-related/Personal Health instrument as well as the ProQOL were
matched to one or more of the theorists’ models. The central theme that quantity and
intensity of stressors affect a client’s coping abilities resulted in an overlap of some
variables among the three theoretical categories. When these divisions were completed,
each theorist’s variables were correlated and ranked. Based on this list of theorist and
rank correlations, four regression analyses were generated.
Neuman’s Contribution
Betty Neuman’s systems theory is central to the study of Compassion Fatigue
Risk inasmuch as her model is concerned with the intra-, inter- , and extrapersonal factors
that contribute to a nurse’s risk of impairment. She believed that the individual has a
70
relationship with stress and its consequences. The basic survival factors and energy
resources of the individual are represented by a core structure which is surrounded by
various lines of resistance and defense that represent coping patterns and protective
buffers against stress. There are three types of environmental stressors that she
categorizes as: intrapersonal, interpersonal, and extrapersonal. Hospice nurses can be
affected by stressors in numerous ways. They may become too empathic toward patients
and their families, and therefore have a tendency to self sacrifice their own needs for the
needs of their patients. They can become, preoccupied with their patients even outside
the hospice setting. These nurses can become overwhelmed and feel bogged down by the
system, in addition to having other life demands and stressors such as personal financial
stress. These five variables operationalized by the demographic and the ProQOL
instruments, were correlated to compassion fatigue risk. Table 4.16 outlines the variables
and their correlation values.
Table 4.16
Factors Matched to Neuman’s Model and Selected as Potential Contributors to
Prediction of CF Risk
Variable
Preoccupied by those I help
Feeling overwhelmed by my work load
Bogged down by the system
Self sacrifice for others needs
Financial stress
Source
ProQOL
ProQOL
ProQOL
Demographic
Instrument
Demographic
Instrument
Type
Intrapersonal
Extrapersonal
Extrapersonal
Interpersonal
Correlation
With CFR
.562
.475
.437
.313
Extrapersonal,
Other life
demands
A regression model (see Figure 4.1) was constructed utilizing these variables and
2
accounted for approximately 50% ( Radn
= .481) of the variance in the dependent
variable, compassion fatigue risk.
71
.282
Y = β̂ 0 + β̂1 X1 + β̂ 2 X2 + β̂ 3 X3 + β̂ 4 X4 + β̂ 5 X5 + e
where: Y = Compassion Fatigue Risk
βi = Partial Regression Coefficient for ith Independent Variable
X1 = Preoccupied by those I help
X2 = Feeling overwhelmed by my work load
X3 = Bogged down by the system
X4 = Financial stress
X5 = Self sacrifice for others needs
e = Prediction Error
Figure 4.1. Sample Regression Model (k = 5) Considering Operationalized Factors from
Neuman’s Systems Theory for the Prediction of Compassion Fatigue Risk
This Global Model tested statistically significant (p < .001) indicating that at least one of
the five IVs is making a non-zero contribution to the prediction of compassion fatigue
risk in the population of Florida hospice nurses. The prediction error rate for this model
(Standard Error of the Estimate) was 4.78 units. See Table 4.17 for a model summary.
Table 4.17
Neuman’s Regression Model Summary
Full
Model
R
.702
Full
Model
R2
.493
Full
Model
p value
<.001
Adjusted
R2
.481
Standard
Error of the
Estimate
4.78
Partial
Regression
Coefficients
2.103
(Part
Correlations)2
.151
tc
7.797
p value
<.001
Feelings of being
overwhelmed
Bogged down by the system
1.112
.032
3.582
<.001
0.953
.027
3.300
.001
Self-sacrifice for others needs
1.251
.007
1.690
.093
Stress from finances
1.516
.011
2.131
.034
Independent
Variables (predictors)
Preoccupied with those I help
Fc
39.705
The coefficient of determination (R2) for this model also exceeded the researcher’s
72
a priori setting of Effect Size (9%) and was thus not only statistically significant but also
deemed of practical importance. Subglobal tests of each of the independent variables in
this model were conducted to determine the extent of their unique contributions when
controlling for the effect of the other IVs in the model. The null and alternate hypotheses
for testing the independent variable: Preoccupied with those I help, are provided as an
example below:
H0:
pop
R Y2 1⋅2345 = 0
Alternate Hypothesis Ha:
pop
R 2Y1⋅2345 ≠ 0
Null Hypothesis
As the reader will note from scrutiny of the model summary, while three of the five IVs
were statistically significant at the α = .01 level, only one IV (Preoccupied with those I
help) exceeded the researcher’s pre-specified 9% criterion (ES) for “importance.” This
variable accounted for approximately 15% of the variance in Compassion Fatigue Risk
when controlling for the other IVs in the model.
Selye’s Contribution
Selye’s stress theory posited that stress is a nonspecific response of the body to
any demand made upon it. For humans, emotional triggers are huge stressors, and are not
uncommon in a hospice setting. He believed that the nursing profession was stressful
because of long hours and the emotional cost of caring. Anxiety, apprehension,
irritability, loss of concentration can lead to distress which can result in physiological
responses such as headaches and exhaustion. Selye believed also that these stressors
affect the individual whether they are perceived or real, and diseases result due to an
inability to cope with stress.
Variables originating from the Demographic and Pro-QOL instruments were
linked to Selye’s stress theory and then correlated with compassion fatigue risk. The five
variables exhibiting the greatest correlation with the dependent variable are listed in
Table 4.18. Anxiety which occurs during times of stress was evidenced by the variable,
being “on edge” due to helping. This variable correlated the highest with compassion
fatigue risk (r = .703).
73
Table 4.18
Factors Matched to Selye’s Model and Selected as Potential Contributors to Prediction
of CF Risk
Variable
“On edge” due to helping
Frightening thoughts due to work
Memory loss
Experience frequent headaches
Full or part-time work
Source
ProQOL
ProQOL
ProQOL
Demographic
Instrument
Demographic
Instrument
Type
Emotional
Correlation
with CFR
.703
Emotional
Physiological
Physiological
.674
.420
.213
Work-related
-.185
When the aforementioned five variables were included as independent variables in a
2
= .711) of the
multiple regression model, they accounted for approximately 70% ( Radn
variance in compassion fatigue risk. In other words, approximately 70% of the
information needed to predict perfectly compassion fatigue risk was being provided by
knowledge of scores on those five variables. This Global Model tested statistically
significant (p < .001) with those five variables; the prediction error rate was 3.55 units,
and the resulting Global R2 exceeded the researcher’s a-priori effect size of 9%. The
result was, therefore, considered of practical importance. See table 4.19 for Selye’s model
summary.
Subglobal tests of each of the independent variables in this model were conducted
to determine the extent of their unique contributions when controlling for the effect of the
other IVs in the model. As the reader will note from scrutiny of the model summary,
three of the five IVs were statistically significant at the α = .01 level while two variables:
“On edge” due to helping and Frightening thoughts due to work accounted for
approximately 20% and 13% respectively when controlling for the other IVs in the
model. These values exceeded the investigator’s pre-set effect size (9%) and thus were
deemed to be of practical importance.
74
Table 4.19
Selye’s Regression Model Summary
Adjusted
R2
.711
Standard
Error of the
Estimate
3.556
Fc
105.785
Full
Model
p value
<.001
Partial
Regression
Coefficients
2.945
(Part
Correlations)2
0.196
tc
12.027
p value
<.001
Frightening thoughts due to
work
Memory loss
3.030
0.128
9.720
<.001
1.184
0.032
4.917
<.001
Experience frequent
headaches
Full or part-time work
0.133
0.000
0.230
.818
-0.716
0.001
-0.918
.360
Full
Model
R
.847
Full Model
R2
.718
Independent
Variables (predictors)
“On edge” due to helping
Figley’s Model
Figley’s etiological model of compassion stress and fatigue is based on the
assumption that empathy and emotional energy are necessary to establish an effective
therapeutic relationship; however, those same traits may make a caregiver vulnerable to
experiencing the emotions of the client (Figley, 2002). The model identified 10 factors
that contribute to compassion fatigue. Two of the 10 factors are considered coping
actions that represent an inverse relationship to residual compassion stress. When a
caregiver achieves a sense of satisfaction with his/her work, and has the ability to
depersonalize from trauma, then the effects of compassion fatigue may be lowered or
prevented. The tendency to self-sacrifice is the variable in this study that represents a lack
of being able to depersonalize. The survey specifically asked hospice nurses if they had a
tendency to sacrifice their own personal and psychological needs in order to satisfy the
needs of their patients.
According to Figley (2002), if the aforementioned two coping mechanisms are not
successful, the effects of residual compassion stress continue to build, and a negative
impact on health and overall quality of life could occur. Variables such as prolonged
75
exposure to suffering over a protracted period of time, traumatic memories, and other life
demands, along with residual compassion stress result in compassion fatigue.
Three risk factors, traumatic memories, prolonged exposure to suffering, and
other life demands, can be explained further by citing specific variables in this study that
represent those three factors in Figley’s model. These variables were selected by the
researcher based on her understanding of the constructs of Figley’s model. These
variables originated from both the demographic and ProQOL instruments. Table 4.20
provides a list of the variables that were matched to Figley’s Compassion Stress and
Fatigue Model and selected as potential contributors to the prediction of CF risk.
Table 4.20
Factors Matched to Figley’s Model and Selected as Potential Contributors to CF Risk
Variable*
Being “infected” by traumatic stress
“On edge” due to helping
Depressed due to helping
Frightening thoughts due to work
Situational avoidance
Difficulty with personal and
professional separation
Type
Exposure to Suffering, Residual
Compassion Stress
Residual compassion Stress, Exposure
to Suffering
Prolonged Exposure to Suffering over a
protracted period of time, Emotional,
Physiological
Traumatic Memories
Prolonged Exposure to Suffering over a
protracted period of time, Traumatic
Memories
Residual Compassion Stress, Prolonged
Exposure to Stress
Residual Compassion Stress, Traumatic
Memories
* All variables on this list originated from the ProQOL
Preoccupied with those I help
The seven variables listed in Table 4.20 were correlated with compassion fatigue
risk. Those correlations as well as their intercorrelations are provided in the following
correlation matrix (Table 4.21).
76
Table 4.21
Correlation Matrix of Variables Selected to Operationalize Figley’s Variables
Compassion
Difficulty
Pearson
Fatigue
Traumatic On
Frightening
with
Correlation Risk
Stress
Edge Depressed Thoughts
Avoidance Separation Preoccupied
Compassion
1.000
Fatigue
Risk
Infected by
Traumatic
.708
1.000
Stress
On Edge
.703
.550 1.000
due to
Helping
Depressed
.674
.434 .516
1.000
due to
Helping
Frightening
.673
.501 .391
.470
1.000
Thoughts
due to Work
Situational
.632
.442 .351
.361
.597
1.000
Avoidance
Difficulty:
Pers/Prof
.615
.278 .360
.305
.213
.202
1.000
Separation
Preoccupied
with those I
.562
.290 .332
.227
.212
.228
.461
1.000
Help
When the seven variables listed above, were included as IVs in a multiple
regression model for predicting compassion fatigue risk, they accounted for
2
approximately 93% ( Radn
= .930) of the variance in the dependent variable. This Global
Model tested statistically significant (p < .001) and the coefficient of determination far
exceeded the researcher’s a priori effect size. The prediction error for this model was 1.8
units. See table 4.22 for Figley’s model summary.
Subglobal tests of each of the independent variables in this model were conducted
to determine the extent of their unique contributions when controlling for the effect of the
other IVs in the model. As the reader will note from scrutiny of the model summary, that
while all of the seven IVs were statistically significant at the α = .01 level, none of them
met or exceeded the researcher’s pre-specified level of importance (ES = 9%).
77
Table 4.22
Figley’s Regression Model Summary
Fc
401.250
Full
Model
p value
<.001
(Part
Correlations)2
.025
tc
8.793
p value
<.001
1.074
.018
7.533
<.001
Depressed due to helping
1.435
.027
9.114
<.001
Frightening thoughts due to
work
Situational avoidance
1.400
.018
7.531
<.001
1.353
.022
8.201
<.001
Difficulty with personal and
professional separation
Preoccupied with those I help
1.321
.049
12.202
<.001
.937
.026
8.967
<.001
Full
Model
R
.965
Full Model
R2
.932
Variables (predictors)
Being “infected” by
traumatic stress
“On edge” due to helping
Adjusted
R2
.930
Standard
Error of the
Estimate
1.755
Partial
Regression
Coefficients
1.412
Contribution of the Combined Theorists
Basic themes of stress, anxiety, and the effects of trauma were threaded
throughout the variables that often overlapped among the three theories. It was observed
also, that variables expressing nurses’ over involvement with their patients’ were
common indicators of compassion fatigue risk. A total of seven variables representing
each of the theories, were chosen to build a composite multiple regression model to
predict compassion fatigue risk. The lowest correlation to compassion fatigue risk was
being preoccupied with those I help (r = .562), whereas, the highest correlation was being
“infected” by traumatic stress (r = .708). See Table 4.23 for the factors that matched the
combined theorists’ model and were selected as potential prediction of CF risk
contributors.
78
Table 4.23
Independent Variables Selected as IV’s in a Composite Model for Predicting Compassion
Fatigue Risk
Variable
Being “infected” by
traumatic stress
“On edge” due to
helping
Depressed due
to help
Frightening
thoughts due to
work
Difficulty with
separation
Preoccupied with
those I help
Feelings of being
overwhelmed
Type
Exposure to Suffering,
Residual Compassion
Stress
Residual compassion
Stress, Exposure to
Suffering Emotional /
Physiological
Prolonged Exposure to
Suffering over a
protracted period of time,
Emotional, Physiological
Emotional / Physiological
Traumatic Memories
Residual Compassion
Stress, Prolonged
Exposure to Stress
Intrapersonal,
Residual Compassion
Stress, Traumatic
Memories
Extrapersonal
Theorist
(Part-Cor)2 in
Theorist Model
Correlation
with CFR
Figley
.025
.708
Figley,
Selye
.018
.196
.703
Figley
.027
.674
Selye,
Figley
.128
.018
.673
Figley
.049
.615
Neuman,
Figley
.151
.026
.562
Neuman
.032
.479
The reader will note that the IV’s previously considered in the various theorist’s
models ranged, in their unique contributions to the prediction of compassion fatigue risk,
from 19.6% (“On edge” due to helping - Selye) to 1.8% (“On edge” due to helping –
Figley and Frightening thoughts due to work – Figley). These values are considerably
less than their squared correlations with the dependent variable due to the necessary
removal of their joint variability with the other IVs in the model before calculating the
variance they share uniquely with compassion fatigue risk. The correlation matrix shown
in Table 4.24 lists not only the correlations of the selected seven IVs with compassion
fatigue risk, but also, the intercorrelations between these variables. As was the case with
these variables in their respective theorists’ models, it is likewise true for the composite
model, that the greater the intercorrelations between the IVs in the model, the more
79
shared variance that must be removed before determining the unique contribution of each
to the prediction of the DV.
Table 4.24
Correlation Matrix of Variables Selected to Operationalize Composite Theorists’
Variables
Pearson
Correlation
Compassion
Fatigue
Preoccupied
Those I Help
Difficulty:
Pers/Prof
Separation
Infected by
Traumatic
Stress
Depressed
due to
Helping
Feelings of
Being Overwhelmed
On Edge due
to Helping
Frightening
Thoughts
due to Work
Compassion
Fatigue
Preoccupied
Those I
Help
Difficulty:
Pers/Prof
Separation
Infected
by
Traumatic Stress
Depressed
due to
Helping
Feelings
of Being
Overwhelmed
On
Edge
due to
Helping
Frightening
Thoughts
due to
Work
1.000
0.562
1.000
0.615
0.461
1.000
0.708
0.290
0.278
1.000
0.674
0.227
0.305
0.434
1.000
0.479
0.284
0.304
0.358
0.347
1.000
0.703
0.332
0.360
0.550
0.516
0.506
1.000
0.673
0.212
0.213
0.501
0.470
0.258
0.391
1.000
2
The seven IVs listed above, accounted for approximately 91% ( Radn
= .907) of the
variance in compassion fatigue risk and tested statistically significant (p < .001) with a
prediction error of 2.01 units. Subglobal tests of the unique contributions of each of the
seven variables were all statistically significant (p < .001) except for feelings of being
overwhelmed (p = .433). Similarly, only one of the IVs (“On edge” due to helping)
shared enough variance with compassion fatigue risk (16.1%) to be judged as making an
important contribution according to the investigator’s a priori effect size (9%). The part
correlation reflects each variable’s correlation after controlling for the other variables.
For example, the part correlation of frightening thoughts due to work with compassion
80
fatigue risk controlling for the other six variables is ry 1• 2 , 3 , 4 , 5 ,6 ,7 =.228 which indicates
that approximately 5%, (.2282 ), of the information needed to predict compassion fatigue
risk comes from that specific variable after controlling for the six other variables. It is
the percentage of variance in compassion fatigue risk uniquely accounted for by
frightening thoughts due to work, and not by the other six variables. The regression
model summary for the composite model is outlined in Table 4.25
Table 4.25
Composite Theorists’ Regression Model
Full
Model
R
.954
Full
Model
R2
.910
Adjusted
R2
.907
Theorists
Variables
(predictors)
Preoccupied by those
I help
Difficulty with
personal and
professional
separation
Being “infected” by
traumatic stress
Depressed due to
helping
Feelings of being
overwhelmed
“On edge” due to
helping
Frightening thoughts
due to work
Neuman,
Figley
Standard
Error of the
Estimate
2.01
Full Model
p value
Fc
295.874
Partial
Regression
(Part
Coefficients Correlations)2
.986
.0290
<.001
tc
8.185
p value
<.001
1.310
.0480
10.535
<.001
Figley
1.588
.0330
8.662
<.001
Figley
1.471
.0290
8.105
<.001
Neuman
.100
.0002
.785
.433
Figley,
Selye
Selye,
Figley
1.082
.1610
6.279
<.001
2.080
.0519
10.875
<.001
Figley
Conclusions
Prevalence of Risk for Compassion Fatigue and Burnout
Evidence supports the ProQOL as a valuable instrument, not only for its
operationalization of the dependent variable: Compassion Fatigue Risk, but also for its
two other subscales: Burnout and Compassion Satisfaction. When using the “Low,”
“Moderate” and “High” risk classifications specified by the ProQOL’s scoring plan,
81
Florida Hospice Nurses are an at-risk population for burnout and compassion fatigue (see
Table 4.26) with 91% of those in the moderate to high risk category for burnout, being
also at moderate to high risk for compassion fatigue.
Table 4.26
Burnout and Compassion Fatigue Low and Moderate to High Risk Frequencies for
Florida Hospice Nurses
Burnout
Compassion Fatigue
Low Risk
n = 84 38.9%
n = 46 21.3%
Moderate to High Risk
n = 132 61.1%
n = 170 78.7%
High Risk
n = 22 10.2%
n = 57 26.4%
If, in fact, these numbers are representative of the entire population of Hospice
Nurses in Florida, the host organizations could easily enter into a state of crisis resulting
from the documented deleterious outcomes of these conditions. It is estimated that there
are approximately 3500 hospice nurses in the state of Florida based on an accessible
population of 1744 for 43% of the hospices in the state. If the results of this investigation
are generalizable to the entire state, approximately 2500 nurses could be in a moderate to
high risk category for compassion fatigue.
While it was true for the current investigation, that those who should have been
classified as moderate to high risk, were, in general, so classified, the demographic and
work-related data reported by the nurse participants, were of little discriminating value
for describing the prevalence of risk for these conditions. This finding suggests that the
literature on compassion fatigue and burnout risk may have placed a greater emphasis on
demographic and work-related factors as discriminating variables, than should be the
case. It also implies that the means by which hospice nurses cope with their stressful
work environments, while balancing other responsibilities, such as personal finances and
family responsibilities, are most likely, unique, highly structured defense mechanisms
that defy description through quantitative, empirical means.
Predicting the Risk of Compassion Fatigue
Despite the evidence, in the target population, which suggests that the prevalence
of risk for compassion fatigue is alarmingly high, the knowledge provided by the
theorists, foundational to the conduct of this investigation, was more than sufficient for
82
the construction of a composite model for the prediction of this phenomenon with
minimal error. With knowledge of these few variables, hospice organizations may
identify nurses at risk and take measures, not only to provide needed support for these
individuals, but also, to seek to eliminate, or reduce, the contributing factors. These
measures, if implemented in a timely and efficient manner, may also have a positive
impact on loss of productivity, increased risk of errors, turnover, and resulting financial
costs to organizations.
Summary
This chapter provided statistical findings of demographic and work-related data of
the sample in this study. The prevalence of compassion fatigue risk was assessed within
this group of randomly selected Florida hospice nurses providing direct patient care in
various settings. Additionally, variables representing constructs from three theorists were
correlated with compassion fatigue risk, and finally regression analyses generated four
regression models to predict compassion fatigue risk among hospice nurses. All four
models were statistically significant, and the variables of the composite model provided
91% of the information needed to perfectly predict compassion fatigue risk. A discussion
of the factors that could have had an effect on the outcomes of this study will be
discussed in Chapter 5.
83
CHAPTER 5
DISCUSSION
The clinical phenomenon of compassion fatigue has received considerable
attention among many health care professions; however, its prevalence among hospice
nurses has been virtually ignored. These end-of-life caregivers are especially vulnerable
because they can become too empathic due to the degree and duration of contact with
their patients, their emotional investments, and frequent exposure to loss (Keidel, 2002).
Understanding the prevalence of compassion fatigue risk and implementing a model that
predicts this risk, could empower nurses and administrators to utilize preventive measures
that promote self care, improve patient outcomes, and reduce absenteeism and loss of
productivity.
This inquiry investigated the prevalence of compassion fatigue risk among
Florida’s hospice nurses, and it provided a model for predicting this risk from
associations between demographic and work-related variables. This chapter discusses the
findings of the study and possible related trends. It will relate the findings to literature,
and it will incorporate the study’s results in the conceptual framework which was used to
guide this research. Additionally, limitations and strengths will be reviewed, and
recommendations for future research will be presented following the implications to
nursing practice, advanced nursing practice, nursing administration, and nursing
education at all levels.
Discussion of the Findings
It is not uncommon for nurses in this specialty of nursing to experience secondary
traumatic stress; therefore, it was not surprising that Florida Hospice nurses are an at risk
population for compassion fatigue. Approximately 79% (n = 170) of the sample were in
the moderate to high risk category for compassion fatigue. The researcher found that
84
demographics and work-related factors were of little discriminating value for describing
the prevalence of risk for this phenomenon. This implies that these nurses possess
intricate defense mechanisms to cope with caring for the terminally ill and their families,
who are often in a state of crisis. These coping abilities may be inherent, and/or may have
been learned from years of nursing experience. It was interesting to note, however, that
none of the Hispanic and Asian nurses scored in the low risk category. It is unknown if
there was a cultural component behind the discrepancy in scores because of the small
numbers of Asian and Hispanic nurses in the sample. Further research may be beneficial
in determining factors that contribute to these differences.
There were other groups within the sample that scored in the high compassion
fatigue (CF) risk category. These nurses indicated that they were undergoing financial
stress, or were diagnosed with depression/PTSD. Since PTSD is closely linked to
compassion fatigue (Figley, 2002), it was not unusual that those participants exhibited
higher risk scores because their coping mechanisms may have been affected by their own
trauma and health conditions.
Additionally 83% (n = 47) of the sampled participants who were classified in the
high CF risk category responded positively that they self sacrifice for others’ needs. This
was an important finding because those who indicated they self sacrifice also admitted to
other health problems resulting from physiological stress. This behavior of nurses’ caring
more for their patients’ needs than for their own needs is reflective of an unhealthy level
of empathy, which is a risk factor for compassion fatigue. Not only were the nurses who
self sacrificed in a high risk category but they exhibited high risk behaviors.
Approximately 64% (n = 136) of the sampled nurses said they sacrificed their
own personal and psychological needs for the needs of their patients. Among this group
there was a greater percentage of smoking behavior, financial stress, headaches, and
hypertension than those who responded negatively to that question. This result confirms
similar findings in the literature with respect to the personal health behaviors of selfless
caregivers.
The researcher in the current study was able to claim the alternate hypothesis
because the results of this study revealed that one or more of the independent variables
provided a significant contribution to the prediction of compassion fatigue risk. These
85
variables were grouped into demographics (financial difficulties, self sacrificing
tendency), work-related stressors (full vs. part-time work, lack of support after trauma),
and health factors (headaches, anxiety, depression). These variables reflected the impact
that stress has on compassion fatigue risk, which is exacerbated when nurses exhibit a
lack of self care.
The seven variables that predicted CF risk in the composite model were a) being
“infected” by traumatic stress, b) being “on edge” due to helping, c) being depressed
due to helping, d) having frightening thoughts due to work, e) having difficulty between
personal and professional separation, f) being preoccupied with those I help and g)
having feelings of being overwhelmed. These behaviors could have been influenced by
the demographic, work-, and health-related factors the nurses revealed in the study. These
behaviors also revealed a central theme of the negative effects of overly identifying with
patients, resulting in unintentionally, vicariously experiencing their pain and anxiety.
This was important because unhealthy levels of empathy, coupled with life demands and
health factors, are directly related to compassion fatigue risk, and distress. According to
Selye (1970), unremitting stress is a cause of additional physiological and psychological
health concerns. Being able to predict compassion fatigue risk via these risk factors is
paramount to the health and welfare of hospice nurses and their ability to have effective
therapeutic relationships.
Response Rate
The overall response rate of the study was 37% after adjusting for data that were
not usable. Seventeen out of 40 hospice organizations were represented through the
mailed surveys, and nurses from five additional organizations completed surveys during
the 20th Annual Hospice and Palliative Care conference in December 2004. Results of
data analyses (Chi-square, Fisher Exact and independent samples t-tests) between the two
venues revealed that there were four statistically significant variables between the two
groups; however, those results were not considered detrimental enough to the objectives
of the study to warrant their exclusion in further analyses, nor to warrant the
consideration of venue as confounding for the study context.
86
External Validity
This sample of 216 hospice nurse participants from both data collection venues
represented 22 of the 40 hospice organizations across the state located in the Florida
Panhandle (12%), Northeast (29.6%), Central (17.1%), Southwest (11.6%), and Southeast
(29.6%) regions. Surveys were collected from small organizations employing very few
nurses to large organizations employing over 200 nurses. The hospice organizations were
located in metropolitan and in rural areas across the state. The majority of the nurses in
this study were RNs (84.7%, n = 183), followed by 13.4% (n = 29) LPNs, and
approximately 2% (n = 4) ARNPs. Nationally, approximately 86% (n = 16,716) of
hospice nurses are registered nurses whereas, 14% (n = 2,628) are LPNs (Hospice
Association of America, 2002).
This sample had a higher proportion of RNs compared to the licensure
distribution of all nurses in Florida which was RNs (n = 173,000, 62%), LPNs (n =
97,000, 35%) and approximately 9,700 (3.4%) ARNPs (Florida Center for Nursing
[FCN], 2004); Gregg & Brunell, 2003). One would expect more RNs than LPNs in this
field of nursing than in general nursing because of the heavy emphasis on case
management within a multidisciplinary team setting.
The Associate degree was the predominant type nursing education (46.8%, n =
101) in this study followed by 20.8% (n = 45) with Baccalaureate degrees, 16.7% (n =
36) with Diploma backgrounds, and 4.2% (n = 9) with Master’s degrees. This distribution
closely reflected the education levels of Florida’s Registered Nurse population who have
mainly Associate degrees (42%), followed by 26% Baccalaureate, 25% Diploma, and 7%
with Masters / Doctoral degrees (FCN, 2003, 2004).
Seventy percent of the nation’s nurses work full time which is similar to Florida’s
nurse population (National Center for Health Workforce analysis, 2002a), whereas 88%
(n = 191) of the nurses in this study worked full time. Caseloads for this sample of nurses
(md = 13) were higher than the national median hospice nurse case load (md = 10, M =
9.8) as reported by the National Hospice and Palliative Care Organization’s (NHPCO)
National Trend Summary Report for 2000-2003 (NHPCO, 2004). Additionally, the nurse
to-patient ratios of one nurse-to-six patients (md = 6) were higher in this study, than the
one to four ratios recommended by the Center to Advance Palliative Care. The reason for
87
these recommended ratios are considerations of psychosocial, educational, family and
bereavement issues that are not factors in other staffing considerations (CAPC, 2002).
The norming group data for the ProQOL instrument was a composite of separate
cases; however, to date, there are three norming groups composed of general health
workers (clinicians and administrators), child/family workers, and school personnel.
Results indicated that general health workers had fewer compassion fatigue risk
symptoms than the other two groups; child-family workers had more burnout
symptomology; whereas, the group of teachers was most satisfied with their work
(Stamm, 2005).
This sample had fewer participants in the lower compassion fatigue risk category
(n = 46, 21.3%), and more nurses in the higher risk category (n = 57, 26.4%), than was
the case for the norming group (M = 11.37) described by the authors of the ProQOL
instrument (Stamm, 2005). Additionally, it was not surprising that the majority of this
study’s sample was in the moderate range (md = 14) because quartile splits were used to
label the three risk categories (25% = low, 50% = moderate, 25% = high).
Relationship to Literature
The literature reviewed and presented for this study included numerous findings
and recommendations related to the impact of compassion fatigue on health care workers
(Clark, & Gioro, 1998; Figley, 2002; Roberts, Flannelly, Weaver, & Figley, 2003;
Wastell, 2002). Additionally, studies have been conducted on stress and burnout, which
may be precursors to compassion fatigue in many professions, including hospice nursing
(Bene & Foxall, 1991; Dean, 1998; Payne, 2001).
The finding that hospice nurses utilize highly structured defense mechanisms was
supported by Payne (2001) who noted that hospice is a positive environment in which to
work even though the nature of the work is difficult, and may lead to burnout. This
finding is supported further by Wright (2002) who researched the qualities of hospice
nurses. The author noted that these nurses have a true sense of “calling”, and are
compassionate and accepting of the choices made by patients and their families (Wright,
2002). The accumulated factors are there to predict compassion fatigue risk; however,
nurses’ coping abilities, such as healthy professional distancing learned from years of
88
nursing experience, balance out many of those stressors. Healthy distancing may be
achieved by a sense of self-care, as evidenced by consciously taking time off from work,
especially when stress begins to accumulate.
In this study, 17% (n = 37) of the nurses indicated that they received no support
after a patient’s traumatic death, and 83.3% (n = 31) of those participants were in the
moderate to high risk CF category. An inability to debrief after a traumatic event
diminished internal coping mechanisms. This finding confirms the results of earlier
studies (Dean, 1998; Mallet, Price, Jurs, & Slenker, 1991), which demonstrated that lack
of support after experiencing a patient’s traumatic death had an impact on these nurses.
More generally, the literature supports the belief that lack of support during occupational
stress and trauma leads to psychological distress (AbuAlRub, 2004; Olofsson, Bengtsson,
& Brink, 2003). Behaviors emanating from these stressors revealed a central theme of
this study, that nurses who become overly empathic with their patients are those that are
most at risk for compassion fatigue. This central theme was highly supported by Riggio
and Taylor (2000) who identified that empathy is an essential aspect of hospice nursing
when it takes the form of “perspective taking” and “empathic concern”; however, once
empathy becomes unhealthy it leads to “personal distress” which negatively affects
nursing care and leads to stress and burnout. Therefore, unhealthy empathy leading to
blurred professional boundaries was supported in empirical literature as a major stressor
in hospice nursing.
Nurses who worked shifts (n = 86) in this study had lower compassion fatigue
risk scores (md = 12.5) than nurses who did not work shifts (md = 15.00). There was
virtually no difference between compassion fatigue risk scores of nurses who worked
rotating shifts (md = 13.89) versus those who did not rotate (md = 13.62). This finding
contradicted the literature, which indicated that working shifts predisposes different
occupational groups, including nurses, to physiological stress and health conditions
(Selye, 1976). One of the main reasons why the findings in this study may have turned
out differently is the unique highly structured coping mechanisms that hospice nurses
may possess which cannot be explained in a quantitative study. They may have life styles
that lend themselves better to evening shift work, which otherwise, can disrupt families
with children. These nurses may be better organized to balance the stress of long, twelve
89
hour shifts, and perhaps a rotating shift schedule, and still find time to maintain a
household, and sustain their relationships with significant others. Additionally, stress is
multifactorial, so each factor builds upon others differently in different people, so an
impact on compassion fatigue may not be easily observed by one or two variables alone.
Finally, there were gaps in the literature because there were virtually no studies
done on hospice nurses and compassion fatigue; however, the fact that compassion
fatigue exists among this population was verified by literature, which reported that nurses
who work in hospice care, emergency room settings, and psychiatric units are engaged in
trauma work (Clark & Gioro, 1998; Schwam, 1998). Compassion fatigue risk among
chaplains and other respondents after the September 11, 2001, New York Twin Towers
disaster was studied, and it was reported that 55% of the sample (n = 403) were in the
moderate to high compassion fatigue risk category (Roberts, Flannelly, Weaver, &
Figley, 2003). Another finding of the study noted that workplace proximity to Ground
Zero, and length of time volunteering for a relief agency, had no effect on compassion
fatigue risk (Roberts, Flannelly, Weaver, & Figley, 2003). These findings were similar to
this study which reported that demographic and work-related variables did not have a
major discriminating value on compassion fatigue risk; however, the participants in this
study had a higher percentage (79%, n = 170) of being in the moderate to high risk
category than the ones in the aforementioned study. The nurses in this study were
exposed to ongoing career stressors, such as patient death and family crisis, on a
continuous basis which was part of their daily work stressors, whereas, the participants in
the other study had a controlled period of episodic acute stress.
Conceptual Framework
Three theoretical concepts were combined to provide the conceptual framework
that guided this study. Neuman’s Systems Model (1995), and Selye’s stress theory
(1956, 1970), provided the basis for assessing the risk of compassion fatigue in palliative
care nurses, and presented a better understanding of the relationship between various
nurse characteristics and compassion fatigue risk from a systems perspective. Figley’s
etiological model of Compassion Stress and Fatigue contributed strongly to a social
science framework that guided the researcher in understanding emotional and situational
90
factors that were instrumental in predicting the risk of this phenomenon, and thus aid in
preventive efforts.
Neuman’s Model
Neuman’s Systems Model was used as a component in the framework for this
study. The nurses in this sample had a relationship with stress and its consequences, and
their relationship with physiological, psychological, sociological, developmental, and
spiritual variables impacted how well they were able to cope with internal and external
stressors. The intrapersonal, interpersonal, and extrapersonal factors represented by the
variables in this sample, may be supportive, such as the unique ability to cope. Common
patterns in this sample included concepts of grief and loss. For example, becoming too
emotionally involved with patients, having too many patients dying within a short period
of time, experiencing the death of a friend or family member, and an increasing
workload, overwhelmed nurses (represented by the system) and wore down the flexible
line of defense. This flexible line is re-strengthened by intra-, inter-, and extrapersonal
factors when the nurse replenishes her energy and coping abilities by taking time off from
work, or even by getting a good night’s sleep.
If the stressors penetrate the flexible line, then the normal line of defense must
protect the system. This line represents a state of stability, and exhibits major coping
patterns developed over a lifetime. If these stress coping mechanisms fail, the stressors
enter into the lines of resistance which are a series of lines, surrounding the core self, that
represent resource factors that fight against stressors. Examples of these resource factors
are the ability professionally to detach from a nurse-patient relationship, and the
knowledge to seek assistance when such is needed. These lines of resistance are similar
to one’s genetic make up and, therefore, are unique to each individual. If stressors pierce
through these lines and directly affect the core self, the individual or system, experiences
a failure, such as the development of the condition or sustaining the full impact of the
injury.
For purposes of this study, specific stressors reflected by Neuman’s Model
included a tendency to self sacrifice for patients’ needs, being preoccupied by patients,
and behaviors centering on an inability to cope with a stressor such as feeling
overwhelmed by workload, and being bogged down by the system. An example of an
91
extrapersonal factor, which served to stress the system, attributed to Neuman’s theory,
was the financial stress variable which asked the nurses if maintaining a financial budget
was a source of stress. This specific stressor was also reflected as a life demand theme
utilized as a compassion fatigue stressor in the combined conceptual model (Figure 5.1).
Two of Neuman’s variables (mentioned above), played an instrumental role in the
composite regression model for predicting compassion fatigue risk. These variables were
being preoccupied by patients, and feelings of being overwhelmed by workload.
Being preoccupied by patients can be a component of excessive empathy, which
can be an intrapersonal or an interpersonal factor which lead to compassion fatigue if
coping mechanisms are not effective. The feeling of being overwhelmed by workload is
an example of an extrapersonal stressor which can lead to feelings of frustration, anxiety,
and play a role in burnout as well as compassion fatigue. Since only 26% of the nurses in
this study were in the high CF risk category, there were aspects of strengthening factors
that probably had an impact on nurses’ coping abilities. Strong spiritual beliefs
(intrapersonal factor) may have provided strength to the participants in order to process
their stressful work environment, and to cope when life demand stressors, such as the
death of a loved one, impacted their lives. Possessing spiritual strength is referenced in
literature as a quality of a hospice nurse (Gaydos, 2004).
Selye’s Stress Theory
Selye, like Neuman, posited that environmental and occupational stress produces
distress among individuals leading to adverse physiological and psychological symptoms.
Stress may be in the form of distress, or eustress, to a lesser degree, which causes an
increased level of arousal and may lead to anxiety, apprehension, destruction of health,
and unhealthy behaviors, such as tobacco use. The nursing profession is particularly
stressful, according to Selye (1976), due to the physiological effects of long hours, shift
work, and rotating shifts. The emotional cost of caring leads to anxiety, chronic illness,
and exhaustion, if coping mechanisms are insufficient, or do not exist. A state of anxiety
and other stressors were represented by this theory with the following variables: a) being
“on edge” due to helping, b) frightening thoughts due to work, c) memory loss d)
frequent headaches and e) full or part-time work. These factors may lead to chronic
conditions such as hypertension, and were incorporated into the prediction model
92
representing Selye’s stress theory. The key variables that produced physiological stress
(being on edge, and having frightening thoughts) were incorporated into the regression
composite model utilized to predict compassion fatigue risk.
Figley’s Model
Figley’s conceptual model provided a belief that empathy and emotional energy
were essential to establish an effective therapeutic relationship; however, prolonged
exposure to suffering without respite increased compassion fatigue risk. In addition to
work-related variables, such as lack of support after traumatic events, nurses exhibited
behaviors resulting from exposure to traumatic events, as seen by their responses on the
ProQOL. Variables such as being “infected” by traumatic stress, and being on edge due
to helping could have been applied to the beliefs of all three theorists because these
variables were the reactions nurses may have to stress which is a central theme to
Neuman and Selye. These variables, as many others, were threaded through all three
theorists’ beliefs, and produced a theme of excessive empathy towards patients which
was a factor that led to compassion fatigue risk. Other variables attributed to Figley’s
model, that reflected this theme and reactions to trauma, were difficulty separating
professional life from personal life, frightening thoughts due to work, and being
preoccupied with those I help.
Combined Model
The researcher synthesized Selye’s, Neuman’s, and Figley’s theories to create a
compassion fatigue conceptual framework for this study. This model (Figure 5.1) was
altered from the initial model (Figure 1.1) after the findings of the study were analyzed.
The main constructs still remained, which were that the human self is a holistic being
who lives in a multidimensional system that is constantly inundated by various stressors
such as trauma, life demands, and factors that cause a blurring of professional boundaries,
such as excessive empathy, which is an unhealthy form of empathy toward patients. It is
important to note that physiological stress (anxiety) is the result of trauma and it may be
associated with feelings resulting from excessive empathy. Trauma, excessive empathy,
and life demands are key stressors which lead to compassion fatigue which is evidenced
by the combined theorists’ conceptual model.
93
The self is able to maintain a balance of wellness through its protective
boundaries and coping patterns which are represented by the various dotted and solid
circles around the core self. Even though stressors penetrate, resource factors continue to
fight to maintain the protective barriers which are the dotted lines surrounding the core
self, represented as lines of resistance. A healthy sense of empathy and strong coping
mechanisms of compassion satisfaction and healthy detachment play a role in protecting
the self. The solid blue line which surrounds the lines of resistance is called the normal
line of defense. It represents a state of stability for the individual; therefore, strong
protective factors such as coping, support, adaptability, and self care are embedded
within that line to protect the individual from outside stressors. The outermost circle is
the flexible line of defense which is a protective buffer that prevents stressors from
breaking through to the normal line of defense. An example of this flexible line of
defense is the hospice nurses’ appropriate use of patient / professional boundaries when
psychological stressors that break those healthy boundaries threaten to penetrate. For
example, if a nurse begins vicariously to experience his or her patient’s pain, and
continuously ignores established professional boundaries, the flexible line of defense
becomes threatened and pierced, if the nurse does not reverse his/her behavior.
The resulting excessive empathy, which is either an intrapersonal or interpersonal
factor that serves to stress the system, then threatens the normal line of defense. If the
behavior continues or worsen, the nurse will lose perspective on therapeutic relationships
despite a life time of coping mechanisms, such as putting one’s personal and family’s
needs above patients’ needs. If the nurse continues to exhibit excessive empathy despite a
lifetime of coping behaviors, the normal line of defense is surpassed, and the remaining
lines of resistance are all that is left between the stressor and the self. At this point, nurses
have individualized coping abilities such as healthy emotional detachment linked with
empathy that is limited to a genuine concern toward their patients. If nurses do not
implement these coping mechanisms, they may develop more anxiety related
symptomology, and act in inappropriate ways towards patients and staff that can
jeopardize their professional integrity. If all the lines of resistance are crossed, the self is
directly impacted by compassion fatigue which results in physiological and psychological
harm to the individual.
94
Trauma, Life Demands,
Excessive Empathy
Blurred Professional
Boundaries
SELF
Compassion
Fatigue
Figure 5.1. Compassion Fatigue Conceptual Framework combines three theoretical concepts.
The Neuman System’s Model, which depicts self and its protective boundaries, is blended with
Selye’s stress theory which depicts the concepts of anxiety associated with trauma, life demands
and the variables that represent coping and adaptability. Figley’s Compassion Stress and Fatigue
Model is threaded into this combined model, as illustrated by the effects of trauma, life demands,
and excessive empathy leading to blurred professional boundaries. These stressors are depicted
by the wavy red line which illustrates the stressors to be unpredictable, coming in waves with
various intensities. All these concepts may be applied to each of these theorists, and have an
impact on compassion fatigue.
_____________________________________________________________________________
Limitations of the Study
This study measured the risk for compassion fatigue at one point in time. There is
a possibility that perceptions may have changed over time, due to individual
circumstances. Communication barriers between the researcher, facility administrators,
and contact persons contributed to the mail survey yield of 166/433 usable surveys (38
%). Because of administrative delays, the survey packets often did not reach the
designated contact persons in a timely manner. An effort was made to achieve a higher
95
mail response rate by follow-up calls to the non-responding facilities. An additional 50
usable responses were obtained from 150 surveys distributed to nurses attending a state
hospice symposium in December, 2004. The conference symposium return rate was 33%.
This collection method was utilized after FSU institutional review board approval was
attained (Appendix A). Therefore, the combined return rate between the two venues was
37% (216/583).
Other reasons for the low response rate may have been a general disinterest in the
study due to the focus on the aftermath of Hurricanes Charley, Frances, Ivan, and Jeanne,
which caused deleterious effects to both Florida coasts in August and September, 2004.
All Florida hospice facilities were invited to participate in this study; however, no effort
was made to follow up with hospices in those areas which may have been most adversely
affected by these storms. One hospice organization in Florida had the roofs destroyed in
five of their administration buildings, and several staff had lost their homes. That
organization elected, understandably, not to participate in the study. Even though data
collection was delayed, there were existing stressors related to these storms that may have
impacted follow through, as recounted by the contact persons from several of the hospice
organizations. Although the overall response rate was 37%, it was not a true limitation
because the average return rate for mailed surveys is 20% (Norwood, 2000).
Additionally, when homogeneous groups are surveyed, significant response-rate bias is
probably unlikely (Ma, Samuels, & Alexander, 2003).
Though survey instruments provide confidentiality and the freedom to respond
candidly, the responses from the sample did not allow for a more qualitative forum, due
to the nature of the study design. Had the questionnaires been formatted to allow for
comments from the participants, a more comprehensive analysis may have provided the
researcher a clearer understanding of other factors that may have contributed to
compassion fatigue risk. There were no other limitations in this study except for the one
statistical limitation pertaining to the Multiple Regression that was outlined in chapter 4.
Strengths of the Study
The method of random selection of the study sample allowed the findings to be
generalized and applied to the population of Florida hospice nurses. A stratified random
96
sampling technique was utilized, in order to provide a representative sample of hospice
nurses from each organization. Furthermore, venue comparison analyses were conducted
to determine if the mailed surveys and symposium surveys were independent of each
other. Efforts were made to avoid duplication between the mailed surveys and the
conference surveys. Initially, contact persons from the hospices participating in the mail
surveys were alerted to remind nurses not to participate in the conference survey if they
completed a mail survey. Additionally, the researcher added specific notations on the
conference surveys to alert nurses not to participate if they had already done so through
the other venue. Therefore, both venues were utilized, and it was determined that there
were no significant differences between and among the groups. The overall sample of
216 (37%) was reflective of the national hospice nurse population, and the Florida
general nurse population. Supporting comparison of data was provided in the Discussion
of Findings section.
Implications for Nursing
Nursing Practice
Nurses considering a career in hospice could reflect on their own qualities, such
as their in-depth knowledge of complex diseases that affect the body, mind, and spirit.
They must consider also, whether they would be able to alter their philosophy of care.
Nurses entering hospice from another nursing specialty often have an existing emphasis
on aggressive measures to treat a disease/chronic condition followed by interventions to
rehabilitate and educate. This philosophy is different from a palliative end-of-life care
approach which centers on pain management and other comfort measure interventions,
such as ease of anxiety and shortness of breath. As they reflect on their qualities and
training, they need to evaluate whether or not they have certain tendencies that may put
them at risk for CF. These tendencies that the nurses have may include, consistently
working too many hours, even when they are told to take time off, and excessively
thinking about their patients, even when they are not their caregiver. Novice, as well as
experienced, hospice nurses could utilize the findings as a guide, or as a barometer, in
reducing compassion fatigue risk behaviors. Nurses can learn about the risks of
compassion fatigue through questionnaires developed by researchers, and through
97
community presentations, and magazine articles. Knowing risk behaviors, such as
developing an unhealthy level of empathy, would alert nurses to reassess their behaviors
during patient interactions. Additionally, knowing these risk factors would provide the
means for informed choices thus potentially minimizing error as nurses evaluate their self
care practices and coping skills, not only in the work setting, but in other environments.
In-service learning opportunities, workshops, and follow-up meetings coordinated by
nurse educators could result in an accessible resource network for all hospice
professionals. Additionally, the multidisciplinary team could be involved in teaching
nurses and other hospice health care professionals about this phenomenon. The
knowledge gained from this research could promote peer-to-peer support among these
hospice caregivers who may not realize that they are experiencing symptoms of
compassion fatigue.
Advanced Nursing Practice
Advanced Practice Nurses (APNs), whether they are advanced palliative nurse
practitioners, nurse managers, supervisors, or clinical nurse specialists, could play a key
role in assessing, evaluating, and intervening against this preventable clinical
phenomenon. APNs who have a solid understanding of compassion fatigue risk
behaviors, or attributes, could play a huge role in a) communicating risk assessments with
staff, b) promoting immediate interventions, and c) developing staff training on
personality profiling and coping behaviors for preventive efforts. APNs could also
provide important linkages with members of the other disciplinary teams within hospice,
in order to create multidisciplinary training modules to prevent compassion fatigue risk,
enhance coping behaviors, and promote healthy emotional detachment. Since the
educator is one of the major roles of APNs, they would be responsible to educate nursing
administration regarding the most current research practices to prevent the risk of
compassion fatigue. They would also play a crucial role in educating, and helping to
create policies on appropriate nurse-patient ratios and manageable caseload levels in the
hospice setting, based on national recommendations.
98
Nursing Administration
It is paramount for nursing administrators to understand the risks and predictors of
compassion fatigue. Their understanding would lead to policies that would promote nurse
retention, improve patient care, and enhance the integrity of the organization. Alert
administrators must be observant of nurses who exhibit signs of compassion fatigue risk.
Nurses who often define themselves as “being their job” may be at particular risk because
they may easily begin to blur the professional/personal boundaries with their patients.
The blurring may become so powerful that the nurse might feel that professional
disappointment is directly linked to his or her own self worth. Often these health care
workers are viewed as highly effective nurses because they care so much for their
patients; however, eventually, these same nurses jeopardize the integrity of the
organization by exhibiting inappropriate behaviors such as lending patients money,
frequently visiting patients when they are not on call, and contacting family members
regarding non-hospice care issues. These observations must be communicated to nurses
during their evaluations, and interventions must be in place to prevent worsening
symptomology. It is truly unfortunate when this is not communicated to nurses because
they are often not cognizant that they are behaving in a non-therapeutic fashion.
The outcome of this research could provide nurse administrators with important
tools when hiring hospice nurses, especially by assessing their risk for compassion
fatigue. The variables utilized in the regression models were high predictors of
compassion fatigue risk; therefore, prospective employees could initially, review and
change interviewing policies and methods, as well as implementing CF and burnout
topics in orientation programs for new nurses. Yearly employee evaluations could also
serve as monitoring tools for CF risk. Eventually organizations could administer a test
utilizing CF risk variables, and the results of the test would produce a compassion fatigue
risk score which, then, would be leveled into a low-, moderate-, or high-risk category.
These results would help employers provide prospective hospice nurses, who were at
higher risk for this phenomenon, with more intense training during orientation.
Additionally, all nurses could be monitored and followed throughout their tenure with
hospice in order to prevent potential CF symptoms from progressing.
99
Nursing Education
The nurse educator is a key player in educating the nursing and the general
community. Nursing education that lays a foundation in research provides a stronger
emphasis in evidence-based learning. The outcomes of this study were based on research
and it is the responsibility of those involved in nursing education to communicate these
results, continue research on variables that predict this phenomenon, and develop
interventions that reduce compassion fatigue in a population that is already in a serious
nursing shortage.
By utilizing evidence-based research, nurse educators could have an impact on
health care organizations by helping to promote policy change, providing awareness of
the risks of compassion fatigue, and showing the benefits of appropriate coping skills.
From a financial perspective, the knowledge gained from the ability to predict
compassion fatigue could save organizations thousands of dollars by appropriately
training prospective employees who may be at risk for this phenomenon, or by
redirecting prospective applicants before they enter this nursing specialty.
Nursing education also plays a key role in providing current hospice nurses with
opportunities to learn more about this phenomenon. Nurses and educators should share
their knowledge with others including social workers, hospice chaplains, nursing
assistants, and nurses in other areas. Workshops, community presentations, and other
opportunities for continuing education regarding compassion fatigue risk, are vital
implications for nursing that have evolved from this research.
Higher Education
All undergraduate curricula should provide information about the risks and the
effects of compassion fatigue, as well as stress and burnout, in nursing. Although
compassion fatigue is an acute condition, the nurse is often unaware of the
symptomology that he or she is exhibiting, and is unaware of his or her behaviors and life
events that have produced this condition. Nursing students are not immune to this
phenomenon, and therefore, need to be cognizant of conditions that may occur in certain
areas of nursing. Graduate nursing students are studying to be leaders in their field, and it
is their responsibility to be knowledgeable of, and to promote compassion fatigue
research. Graduate nursing students are educators in each of their roles, and should
100
promote compassion fatigue self awareness, and model appropriate self care behaviors to
the nursing community.
Recommendations for Future Research
Comparison Studies
It would be beneficial to replicate this study with a larger national random sample
of nurses to learn if the results are comparable to other parts of the country. Other areas
may have different patient populations and organizational policies, which may, or may
not, impact CF risk. Additionally, comparison studies of hospice nurses, social workers,
and chaplains of the hospice multidisciplinary team may provide information on the
prevalence and predictive factors of the risk of compassion fatigue among other
disciplines within the hospice setting.
Policy Analyses
Studies analyzing the effectiveness of current hospice policies/practices regarding
hiring practices, orientation procedures, staff training, and employee yearly reviews
would be useful to caregivers and organizations. Assessing whether organizations a)
monitor CF / burnout risk among hospice nurses, b) enhance nurses’ understanding of
this phenomenon, c) properly place new hires in their work setting with trained
preceptors, and d) offer training on the importance of personal health issues, could
provide valuable information leading to policy change and compassion fatigue risk
prevention.
Trends were revealed in this study indicating that hospice nurses may be
migrating from an acute care setting to seek a better working environment in end-of-life
care. It is unknown what kind of stress levels they have as they change from a curative to
a palliative end-of-life method of care. It is also unknown, if hospice organizations are at
an advantage or disadvantage when hiring new graduates without any nursing experience
to work in this specialty. Their stress levels may easily rise once they finish orientation,
and become independent. These new graduate nurses may be especially at risk if
orientation does not encompass the risks of CF and burnout. Additionally, a cost analysis
study could be instituted, based on these studies, in order to ascertain the financial impact
compassion fatigue risks may have on the health care industry. Evidence-based practices,
101
preventive measures, and policy changes can then be implemented in the hope of
decreasing the risks of this clinical phenomenon.
Giving Nurses a Voice
It was enlightening to read nurses’ unsolicited comments written on the survey
instruments because it brought the data to life and revealed the personal effects of stress.
One nurse indicated that caseloads are high and there is “poor gratitude from [the]
employer”. Some of the other comments included, “… no relief when you have multiple
deaths in a day or week, not recognizing the nurse needs time to refill the well.” “Why
doesn’t the survey ask about anger? I think this is a big symptom of burnout”. “When I
was a charge nurse I had increased weight, increased blood pressure, and worked 50-60
hours a week. I also had a child at home and a sick mother”. “No, but when I get them
[headaches], they are migraines that require an ER visit.” “…more mature people with
varied life experiences handle this job more capably.”
The review of literature revealed that there were few studies using qualitative
data, and a greater effort needs to be made giving nurses, in this specialty, a voice to
validate their concerns, bring this phenomenon to life, and to formulate theory from
qualitative data.
Implementing quantitative and qualitative approaches could result in an analysis
of the findings that provides more comprehensive, and perhaps, humanistic information
on a group of professionals who have been entrusted with the responsibility of providing
quality end-of-life care to the terminally ill.
Summary
This study began with an identification of gaps in literature with respect to the
prevalence of compassion fatigue risk among Florida’s hospice nurses. The literature
revealed that there were adequate theories and empirical research on the effects of stress
and compassion fatigue among health care workers. Specifically, theorists, Neuman,
Selye and Figley, have provided sufficient evidence to test their theories as they applied
to the current research context, and to inspect the previous synthesis of their theories
grounded in this study’s data. Therefore, the researcher was justified to use a tier three
prediction design in this study.
102
Although there was adequate research on compassion fatigue risk among health
care disciplines, the effects of this phenomenon on hospice nurses were virtually ignored
in the literature. The results of this study revealed that hospice nurses are at risk for
compassion fatigue, and that there are factors that can predict compassion fatigue risk
among this population. This study was able to fill some of the gaps in literature; however,
others still exist, such as comparison studies of larger populations, analysis of existing
policies in relation to compassion fatigue risk, and implementation of qualitative studies
to provide greater depth to quantitative findings. This study was the first step on a path to
identify those at risk for the debilitating effects of compassion fatigue inherent in a high
risk nursing specialty.
103
APPENDIX A
HUMAN SUBJECTS APPROVAL LETTERS
104
105
106
APPENDIX B
HOSPICE ORGANIZATION SUPPORT LETTERS
107
108
109
APPENDIX C
PROFESSIONAL QUALITY OF LIFE
COMPASSION SATISFACTION AND FATIGUE
SUBSCALES – REVISION III (PROQOL-RIII)
(MAIL)
110
111
APPENDIX D
DEMOGRAPHIC INSTRUMENT
(MAIL)
112
113
114
APPENDIX E
INITIAL E-MAIL TO HOSPICE ADMINISTRATORS
115
116
APPENDIX F
OUTLINE OF STUDY
117
118
119
120
APPENDIX G
E-MAIL INQUIRY ATTACHMENT TO HOSPICE ADMINISTRATORS
121
122
APPENDIX H
ADMINISTRATOR RESEARCH PACKET COVER LETTER
123
124
APPENDIX I
HOSPICE NURSE EXPLANATORY COVER LETTER
(MAIL)
125
126
127
APPENDIX J
FOLLOW UP LETTER / PHONE CALL
128
129
APPENDIX K
HOSPICE NURSE EXPLANATORY COVER LETTER
(CONFERENCE)
130
131
132
APPENDIX L
DEMOGRAPHIC INSTRUMENT
(CONFERENCE)
133
134
135
136
APPENDIX M
PROFESSIONAL QUALITY OF LIFE
COMPASSION SATISFACTION AND FATIGUE
SUBSCALES – REVISION III (PROQOL-RIII)
(CONFERENCE)
137
138
REFERENCES
AbuAlRub, R. F. (2004). Job stress, job performance, and social support among hospital
nurses. Journal of Nursing Scholarship, 34(1), 73-78.
Adams, J.P., Hershatter, M.J., & Moritz, D.A. (1991). Accumulated loss phenomenon
among hospice caregivers. The American Journal of Hospice & Palliative Care,
8(3), 29-37.
Agency for Health Care Administration. (2002). 2002 Certificate of need annual report.
Tallahassee, FL: Author. Retrieved June 15, 2004 from
http://www.fdhc.state.fl.us/MCHQ/CON_FA/Annual_Rpt/index.shtml#top
Bene, B., & Foxall, M.(1991). Death anxiety and job stress in hospice and medical
surgical nurses. The Hospice Journal, 7(3), 25-41.
Brewer, J.K., & Workman, D.R. (2003). Introductory statistics for researchers (7th ed.).
Boston: Pearson Custom Publishing.
Center to Advance Palliative Care (2002). CAPC Manual: Acuity. Retrieved March 2005
from http://64.85.16.230/educate/content/elements/inpatientunitacuity.html
Clark, M.L., & Gioro, S. (1998). Nurses, indirect trauma, and prevention. Image:
Journal of Nursing Scholarship, 30(1), 85-87.
Cohen, J. (1990). Statistical power analysis for the behavioral sciences ( 2nd ed.).
Mahwah, NJ: Lawrence Erlbaum Associates.
Collins, S., & Long, A. (2003). Too tired to care? The psychological effects of working
with trauma. Journal of Psychiatric and Mental Health Nursing, 10, 17-23.
Davis, S. (2003). Can caregivers care too much? Studies assess relatively new condition
known as compassion fatigue that can emotionally drain benevolent practitioner.
DVM Newsmagazine, 34(8), 58-59.
Dean, R.A. (1998). Occupational stress in hospice care: Causes and coping strategies.
The American Journal of Hospice & Palliative Care, 15(3), 151-154.
Duffy, S.A., & Jackson, F. (1996). Stressors affecting hospice nurses. Home Healthcare
Nurse, 14(1), 54-60.
139
Figley, C.R. (Ed.) (1995). Compassion fatigue : Coping with secondary traumatic stress
disorder in those who treat the traumatized. New York: Brunner/Mazel.
Figley, C.R. (1997). Burnout in families: The systemic costs of caring. Boca Raton, FL:
CRC Press.
Figley, C.R. (2002). Compassion fatigue: Psychotherapists’ chronic lack of self care.
JCLP/In Session: Psychotherapy in Practice, 58(11), 1433-1441.
Figley, C.R., & Stamm, B.H. (1996). Psychometric review of compassion fatigue self
test. Retrieved March 22, 2004 from
http://www.isu.edu/~bhstamm/pdf/figleystamm.pdf. In B.H. Stamm(Ed),
Measurement of Stress, Trauma and Adaptation. Lutherville, MD: Sidran Press.
Florida Center for Nursing. (2003). Nursing supply and demand: Synthesis and
evaluation of existing Florida data. Author. Retrieved July 10, 2004 from
http://www.flcenterfornursing.org/index.htm
Florida Center for Nursing (2004). Statewide strategic plan for nursing workforce in
Florida. Author. Retrieved February 2005 from
http://www.flcenterfornursing.org/strategic/StratPlan-November04.pdf
Freese, B.T. (2002). Betty Neuman: Systems model. In A.M. Tomey, & M.R. Alligood
(Eds.), Nursing theorists and their work (5th ed., pp. 299-335.). St Louis, MO:
Mosby.
Gaydos, H.L. (2004). The living end: Life journeys of hospice nurses. Journal of Hospice
and Palliative Nursing, 6(1), 17-26.
Gigliotti, E. (1999). Women’s multiple role stress: Testing Neuman’s flexible line of
defense. Nursing Science Quarterly, 12(1), 36-44.
Gray-Toft, P.A., & Anderson, J. G. (1986-87). Sources of stress in nursing terminal
patients in a hospice. OMEGA, 17(1), 27-39.
Gregg, A., Brunell, M.L. (2003). Nursing supply & demand in Florida: Analysis of
nursing licensure data. Florida Center for Nursing. Retrieved July 10, 2004 from
http://www.flcenterfornursing.org/research/FLSupply.pdf
Hall, D.S. (2004). Work-related stress of registered nurses in a hospital setting. Journal
for Nurses in Staff Development, 20(1), 6-14.
Hospice Association of America (2002). Hospice Facts and Statistics. Author. Retrieved
February 14, 2004 from http://www.nahc.org/Consumer/hpcstats.html
140
Hospice Foundation of America. (2002, January ). Hospice Medicaid education project
final report (Contract No. COR84). Miami, FL: Health Council of South Florida,
Inc.
Hospice & Palliative Nurses Association. (2004). HPNA position paper: Value of the
professional nurse in end-of life care. Author. Journal of Hospice and Palliative
Nursing, 6(1), 65-66.
Hospice & Palliative Nurses Association. (2003). HPNA position statement: Shortage of
registered nurses. Author. Retrieved February 17, 2004 from
http://www.hpna.org/position_nurseshortage.asp
Huggard, P. (2003). Compassion fatigue: how much can I give?. Medical Education. 37,
163-164.
Joinson, C. (1992). Coping with compassion fatigue. Nursing 92, 22(4), 116, 118-119,
121.
Kees, N.L., & Lashwood, P.A. (1996). Compassion fatigue and school personnel:
Remaining open to the affective needs of students. Educational Horizons, 75, 4144.
Keidel, G.C. (2002). Burnout and compassion fatigue among hospice caregivers.
American Journal of Hospice & Palliative Care, 19(3), 200-205.
Kulbe, J. (2001). Stressors and coping measures of hospice nurses. Home Healthcare
Nurse, 19(11), 707-711.
Lambert, V.A., Lambert, C.E., & Yamase, H. (2003). Psychological hardiness, workplace
stress and related stress reduction strategies. Nursing and Health Sciences. 5, 181184.
Larsen, D., Stamm, B.H., & Davis, K. (2002). Telehealth for prevention and intervention
of the negative effects of caregiving. Traumatic StressPoints, 16, (4). Retrieved
from http://www.istss.org/publications/TS/Fall02/telehealth.htm
Ma, C., Samuels, J.E., & Alexander, J.W. (2003). Factors that influence nurses’ job
satisfaction. Journal of Nursing Administration, 33(5), 293-299.
Mallett, K., Price, J.H., Jurs, S.G., & Slenker, S. (1991). Relationships among burnout,
death, anxiety, and social support in hospice and critical care nurses.
Psychological Reports, 68, 1347-1359.
Masterson-Allen, S., Mor, V., Laliberte, L., & Monteiro, L. (1985). Staff burnout in a
hospice setting. The Hospice Journal, 1(3), 1-14.
141
Maytum, J.C.., Heiman, M.B., & Garwick, A.W. (2004). Compassion fatigue and burnout
in nurses who work with children with chronic conditions and their families.
Journal of Pediatric Health Care, 18, 171-179.
McCance, K.L., & Shelby, J. (1996). Stress and disease. In S.E. Huether, & K.L.
McCance (Eds.), Understanding pathophysiology (pp. 215-228.). St Louis, MO:
Mosby-Year Book.
McCann, L., & Pearlman, L.A. (1990). Vicarious traumatization: A framework for
understanding the psychological effects of working with victims. Journal of
Traumatic Stress, 3(1), 131-149.
Meyers, T.W., Cornille, T.A. (2002). The trauma of working with traumatized children.
In C.R. Figley (Ed.), Treating compassion fatigue (pp. 39-55). New York:
Brunner-Routledge.
McVicar, A. (2003). Workplace stress in nursing: a literature review. Journal of
Advanced Nursing, 44(6), 633-642.
National Center for Health Workforce Analysis. (2002a). The registered nurse
population: Findings from the 2000 national sample survey of registered nurses.
U.S. Department of Health and Human Services, 1-135.
National Center for Health Workforce Analysis. (2002b). Projected supply, demand and
shortages of registered nurses: 2000-2020. U.S. Department of Health and Human
Services, 1-22.
National Hospice and Palliative Care Organization (2003). NHPCO Facts and figures.
Author. 1-7. Retrieved January 10, 2004 from
http://www.atlanticare.org/hospice/answers/outline/FactsFiguresMay02.pdf
National Hospice and Palliative Care Organization (2004). NHPCO Facts and figures.
Author. 1-4. Retrieved April 5, 2004 from
http://www.nhpco.org/files/public/Facts%20Figures%20Feb04.pdf
National Hospice and Palliative Care Organization (2004). National Trend Summary
Report. Author. 1-24. Retrieved March 1, 2005 from
http://www.nhpco.org/files/public/NDS00_03TrendsStats101904.pdf
Neuman, B. (1995). The Neuman Systems Model. In B. Neuman (Ed.)., The Neuman
systems model (3rd ed., pp. 3-76.). Norwalk, CT: Appleton & Lange.
Norwood, S.J. (2000). Research strategies for advanced practice nurses. Upper Saddle
River, NJ: Prentice Hall Health.
142
Olofsson, B. & Bengtsson, C. & Brink, E. (2003). Absence of response: a study of
nurses’ experience of stress in the workplace. Journal of Nursing Management,
11, 351-358.
Page, G.G., & Lindsey, A.M. (2003). Stress response. In V. Carrieri-Kohlman, A.M.
Lindsey, & C.M West (Eds.), Pathophysiological phenomena in nursing (3rd ed.
pp. 275-295.). St. Louis, MO: Saunders.
Payne, N. (2001). Occupational stressors and coping as determinants of burnout in female
hospice nurses. Journal of Advanced Nursing, 33(3), 396-405.
Stamm, B.H. (2005). The ProQOL Manual. The professional quality of life scale:
Compassion satisfaction, burnout, compassion fatigue/secondary trauma scales, 124. Retrieved March18, 2005 from
http://www.isu.edu/irh/documents/proqol/ProQOL_Manual.pdf
Riggio, R.E., & Taylor, S.J. (2000). Personality and communication skills as predictors of
hospice nurse performance. Journal of business and Psychology, 15(2), 351-359.
Roberts, S.B., Flannelly, K.J., Weaver, A.J., & Figley, C.R. (2003). Compassion fatigue
among chaplains, clergy, and other respondents after September 11th. The Journal
of Nervous and Mental Disease, 191(11), 756-758.
Schwam, K. (1998). The phenomenon of compassion fatigue in perioperative nursing.
AORN Journal, 68(4), 642-645.
Selye, H. (1956) The stress of life. New York: McGraw-Hill.
Selye, H. (1976). The stress of life, revised edition. New York: McGraw-Hill.
Sontag, M.A. (1996). Hospices as providers of total care in one western state. The
Hospice Journal, 11(3), 71-94.
Stamm, B.H. (2002). Research information on the ProQOL-CSF-R-III. Retrieved March
22, 2004 from http://www.isu.edu/~bhstamm/testsProQOL_psychometric.htm
United States General Accounting Office. (2001). Testimony before the committee on
health, education, labor and pensions, U.S. Senate. Nursing workforce:
Recruitment and retention of nurses and nurse aides is a growing concern. GAO01-750T, 1-8.
Vachon, M.L.S. (1987). Occupational stress in the care of the critically ill, the dying, and
the bereaved. New York: Hemisphere Publishing Company.
Wastell, C.A. Exposure to trauma: The long-term effects of suppressing emotional
reactions. The Journal of Nervous and Mental Disease, 190(12), 839-845.
143
Wright, D.J. (2002). Researching the qualities of hospice nurses. Journal of Hospice and
Palliative Nursing, 4(4), 210-216.
Zimmerman, J. (2000). Compassion fatigue in HIV/AIDS care nurses. Unpublished
Master’s Thesis. Tallahassee, FL: Florida State University.
144
BIOGRAPHICAL SKETCH
Maryann Abendroth, graduated from Florida State University with a Bachelor of
Science in Nursing in April, 1997. She currently lives in Tallahassee with her husband,
John and their two children, Jennifer and John.
Maryann is also a hospice nurse, and a graduate teaching assistant who is a recent
recipient of Florida State University’s Outstanding Teaching Assistant Award. She plans
to pursue a career in nursing education with an emphasis on end-of-life care following
completion of graduate school.
Maryann’s personal interests include hiking, reading, and gardening. She plans to
publish her thesis manuscript to heighten awareness of the risk of compassion fatigue
among hospice nurses, and to promote preventive measures to lessen the risk of this
clinical phenomenon.
145