* h

Ji'!.53§§.‘j '1.

4 1W5}
4"

l'

i"; fig¢

I‘z:

'2':
, ..
3‘.
r. n
. :.~....'* ~
.g».\.
.u

.
-.
mo
.3: '
-

 

3h .-

E?‘ ,‘szw .
9' VII

. .5! 4..

(

V 1L»:

6 .
i _»
l :33;

This is to certify that the
dissertation entitled

THE ASSOCIATION OF FREQUENCY, INTENSITY, AND
DISTRESS OF FATIGUE, PAIN AND INSOMNIA FOR
CHEMOTHERAPY PATIENTS

presented by

JACQUELYN ANN KEEHNE-MIRON

has been accepted towards fulfillment
of the requirements for the

degree In Nursi&

 

 

takes/:44.

Major Professor’s Signature

5'07— 2007

 

Date

MSU is an aflirmative-actrbn, equal-opportunity employer

_ -._._.—.--.—._ _

..-.—.-.-.-a-.-.—._ -.- —.-.-.-.-.—.--.-.-u---.-.-.-.-.-.—‘--.---—c-.—A_

 

LIBRARY
Michigan State
University

 

 

 

PLACE IN RETURN BOX to remove this checkout from your record.
TO AVOID FINES return on or before date due.
MAY BE RECALLED with earlier due date if requested.

 

DATE DUE DATE DUE DATE DUE

$196M gaze mm? 2010

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

6/07 p:/ClRC/DaleDue.indd-p.1

THE ASSOCIATION OF FREQUENCY, INTENSITY AND DISTRESS OF FATIGUE,
PAIN AND INSOMNIA FOR CHEMOTHERAPY PATIENTS
By

Jacquelyn Ann Keehne-Miron

A DISSERTATION
Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of
DOCTOR OF PHILOSOPHY
Department of Nursing

2007

Abstract
THE ASSOCIATION OF FREQUENCY, INTENSITY AND DISTRESS OF FATIGUE,
PAIN AND IN SOMN IA FOR CHEMOTHERAPY PATIENTS
By

Jacquelyn Ann Keehne-Miron

Purpose/Objectives: The purpose of this research was to examine the dimensions
of frequency, intensity, and distress of the co—occurring symptoms of fatigue, pain and
insomnia as they occur at two different data collection points in two randomized clinical
trials (RCT’S) of a cognitive behavioral intervention. This study will answer two
questions. At baseline observation and at 10 weeks is there an association between the
dimension of fatigue and the dimension of pain and/or insomnia for adults receiving
chemotherapy? Also, can categories of response to fatigue management predict changes
in the dimensions of pain and insomnia at 10 weeks? Co-variates examined include: age,
site and stage of cancer, sex, and co-morbid conditions.

Setting/Methods: Seven cancer centers throughout the Midwest and East coast
accrued patients. Descriptive, cross-tabs, t-tests and regression analyses were conducted
using SPSS version 13.0.

Sample: Adults receiving chemotherapy for solid tumors or non-Hodgkin’s
lymphoma (n=671) participated. Seventy percent were female and 86.3% were
Caucasian. Thirty-five percent had breast cancer and 21% had lung cancer. The mean
age in this study was 57.6 years (range of 25-90).

Findings: At baseline and 10 weeks there is an association between the three

dimensions of fatigue and the same dimensions for pain and insomnia. Categories of

response to fatigue management were capable of predicting changes in the dimensions of
pain and insomnia at 10 weeks. For all dimensions of pain and insomnia, with the
exception of 10 week frequency and distress of insomnia, younger age enhanced the
dimension of pain or insomnia over and above the effect of fatigue. Co—morbid conditions
also enhanced the dimensions of pain at both baseline and 10 weeks over and above the
effect of fatigue. However, co-morbid conditions only influenced frequency of insomnia
at baseline over and above the effect of fatigue. This study supported the appropriateness
of the use of fatigue management categories as a research technique. Future studies
should be directed toward the use of these categories to compare interventions aimed at
various dimensions of multiple co-occurring symptoms such as fatigue, pain and

insomnia in RCT’S.

Dedication

This dissertation is dedicated to my husband and family, and to the cancer patients I have

had the privilege of working with.

iv

ACKNOWLEDGMENTS

I would like to thank the following organizations for their support of the study — The
National Cancer Institute (R01 CA 79280 & R01 CA 30724), The Walther Cancer
Institute, The Oncology Nursing Society Foundation and Michigan State University
College of Nursing. I would also like to thank the following individuals: Drs. Barbara
and Charles Given for their encouragement, support, leadership and wisdom to guide this
project; Dr. Alla Sikorskii for her statistical expertise, responsiveness, and support; Dr.
Barbara Conley for her oncology expertise and Dr. Susan Hoppough for her expertise,
enthusiasm and friendship; and my husband and family for their never ending patience

and support.

TABLE OF CONTENTS

CHAPTER 1 .......................................................................................... 1
OVERVIEW .......................................................................................... 1
FATIGUE ............................................................................................. 5
PAIN AND INSOMNIA ASSOCIATED WITH FATIGUE ................................ 8
RESEARCH QUESTIONS ..................................................................... 10
RESEARCH QUESTION 1 ................................................................. 10
RESEARCH QUESTION 2 ................................................................. 10
RESEARCH QUESTION 3 ................................................................. ll
RESEARCH QUESTION 4 ................................................................. l 1
ANT ECEDENTS .................................................................................. 12
AGE ............................................................................................. l2
STAGE OF DISEASE ........................................................................ l3
SEX .............................................................................................. 14
CO-MORBID CONDITIONS / SITE OF CANCER ................................... l4
PURPOSE OF THE RESEARCH .............................................................. 15
CHAPTER 2 ........................................................................................ l7
CONCEPTUAL FRAMEWORK ............................................................... l7
INTERVENTION ADAPTATION OF THE MODEL ..................................... 17
THE SYMPTOM EXPERIENCE MODEL .................................................. 20
ANTECEDENTS .............................................................................. 20
SYMPTOM PRODUCTION ............................................................... 21
SYMPTOM PERCEPTION ................................................................. 22
FREQUENCY ................................................................................. 23
INTENSITY .................................................................................... 24
DISTRESS ..................................................................................... 24
MEANING .................................................................................... 25
MULTIPLE SYMPOTMS .................................................................. 26
SYMPTOM EXPRESSION ................................................................. 27
CONSEQUENCES ........................................................................... 27

vi

ADAPTATION OF THE SEM .................................................................. 28

INTERVENTION ............................................................................ 28
TIME ........................................................................................... 30
SUMMARY ......................................................................................... 33
CHAPTER 3 .......................................................................................... 35
LITERATURE REVIEW ........................................................................ 35
ANTECEDENTS ................................................................................... 35
AGE ............................................................................................. 35
SEX .............................................................................................. 37
STAGE AND SITE OF CANCER ........................................................... 39
CO-MORBID CONDITIONS ............................................................... 40
SYMPTOM PERCEPTION ..................................................................... 42
FATIGUE ........................................................................................... 43
CONCEPTUAL DEFINITION OF FATIGUE ............................................ 44
CO-OCCURRING SYMPTOM STUDIES INCLUDING FATIGUE .................. 46
INTERVENTIONS FOR FATIGUE, PAIN AND INSOMNIA ........................ .54
TIME ................................................................................................ 60
SUMMARY ........................................................................................ 62
CHAPTER 4 ....................................................................................... 64
DESIGN AND METHODS ..................................................................... 64
RESEARCH QUESTIONS ..................................................................... 64
RESEARCH QUESTION 1 .................................................................. 64
RESEARCH QUESTION 2 .................................................................. 65
RESEARCH QUESTION 3 .................................................................. 65
RESEARCH QUESTION 4 .................................................................. 65
SAMPLE AND SETTING ....................................................................... 67
THE INTERVENTION ........................................................................... 68
EXPERIMENTAL VARIABLES .............................................................. 71
MEASURES ......................................................................................... 7]

vii

DATA COLLECTION SCHEDULE AND PROCEDURES .............................. 72

DATA ANALYSIS AND INTERPRETATION ............................................. 73
POWER ............................................................................................. 76
PROTECTION OF HUMAN PARTICIPANTS ............................................ 76
DATA SECURTIY ................................................................................ 77
RECRUITER TRAINING ....................................................................... 77
THE RECRUITMENT PROCESS ............................................................ 78
RECTUITER QUALITY ASSURANCE ..................................................... 79
INTERVETNION NURSE TRAINING ...................................................... 79
THE INTERVENTION PROCESS ............................................................ 80
INTERVENTION NURSE QUALITY ASSURANCE .................................... 81
INTERVIEWER TRAINING .................................................................. 81
THE INTERVIEW PROCESS .................................................................. 82
INTERVIEWER QUALITY ASSURANCE ................................................. 83
WOMEN AND MINORITY INCLUSION IN CLINICAL RESEARCH .............. 83
FACILITIES AND RESOURCES ............................................................. 84
SUMMARY ......................................................................................... 84
CHAPTER 5 ........................................................................................ 85
FINDINGS .......................................................................................... 85
SAMPLE ............................................................................................. 85
MEASURES ........................................................................................ 89
GROUP MEMBERSHIP AND A'ITRITION ................................................ 90

viii

RESULTS ........................................................................................... 91

RESEARCH QUESTION 1 .................................................................. 94
RESEARCH QUESTION 2 .................................................................. 98
RESEARCH QUESTION 3 ................................................................. 101
RESEARCH QUESTION 4 ................................................................. 103
POWER ............................................................................................ l l 1
DISCUSSION ..................................................................................... 114
PAIN AND FATIGUE ....................................................................... 115
AGE ............................................................................................ l 16
CO-MORBID CONDITIONS ............................................................... 119
INSOMNIA AND FATIGUE ............................................................... 121
LUNG CANCER .............................................................................. 121
THE DIMENSION OF DISTRESS AND INSOMNIA ................................. 123
SUMMARY OF FINDINGS FOR RESEARCH QUESTIONS 1-3 .................. 124
FINDINGS FOR RESEARCH QUESTION 4 .......................................................... 126

Frequency ofPain... ....126
Intensity ofPain... ....127
Distress ofPain... ......127

Frequency ofInsomnia... ....128

Intensity ofInsomm'a... ....128
Distress ofInsomm'a... 129
SUMMARY OF FINDINGS FOR RESEARCH QUESTION 4 ....................... 129
STUDY LIMITATIONS ........................................................................ 131
IMPLICATIONS FOR CLINICAL PRACTICE ......................................... 132
IMPLICATIONS FOR RESEARCH ......................................................... 137
CONCLUSION ................................................................................... 139
APPENDIX A ..................................................................................... 142
REFERENCES ..................................................................................... 144

ix

LIST OF TABLES

 

Title Page
Table 1: Descriptive Statistics of Sociodemographic and Disease
Characteristics of Patients ............................................................... 88
Table 2: Number / % Statistics of Cancer Site by Sex ............................. 89
Table 3: Number / % Statistics of Co-morbid Conditions ......................... 90
Table 4: Mean Values, Standard Deviations and Possible Ranges of the
Symptoms per Dimension .............................................................. 91
Table 5: Regression Analyses for Relating Frequency of Pain to Frequency
of Fatigue and Other Covariates at Baseline and 10 Weeks ....................... 95
Table 6: Regression Analyses for Relating Frequency of Insomnia to
Frequency of Fatigue and Other Covariates at Baseline and 10 Weeks... 97
Table 7: Regression Analyses for Relating Intensity of Pain to Intensity of
Fatigue and Other Covariates at Baseline and 10 Weeks .......................... 99
Table 8: Regression Analyses for Relating Intensity of Insomnia to Intensity
of Fatigue and Other Covariates at Baseline and 10 Weeks ....................... 100
Table 9: Regression Analyses for Relating Distress of Pain to Distress of
Fatigue and Other Covariates at Baseline and 10 Weeks .......................... 102
Table 10: Regression Analyses for Relating Distress of Insomnia to Distress
of Fatigue and Other Covariates at Baseline and 10 Weeks ....................... 104
Table 11: Frequency and % of Fatigue Management Group Membership. . 106
Table 12: Regression Analysis for Frequency, Intensity, & Distress of Pain
and Fatigue Management Category at 10 Weeks ..................................... 108
Table 13: Regression Analysis for Frequency, Intensity, & Distress of
Insomnia and Fatigue Management Category at 10 Weeks ........................ 112
Table 14: Power to Detect Effects of Covariates in the Models for Pain or
Insomnia & Fatigue Frequency, Intensity, & Distress at Baseline & 10
Weeks .................................................................................... 1 1 3

LIST OF FIGURES

 

Title Page
Figure 1: The Symptom Experience Model ........................................... 19
Figure 2: Adapted SEM .................................................................. 32

Figure 3: Frequency Distribution of the Variable of Age for Study
Participants ................................................................................. 86

xi

Chapter 1
Overview
The American Cancer Society projects that there will be 1,444, 920 new cases of

cancer diagnosed in the United States in 2007 (Jemal, Siegel, Ward, Murray, Xu, &
Thun, 2007). It is estimated that 559,650 Americans will die of cancer in 2007, making it
the second most common cause of death in the United States (J emal et al., 2007).
However, due to progress in early diagnosis and the implementation of new and
improved treatments 5-year survival is at 65% for cancers diagnosed between 1995 and
2001, up fiom a 50% rate for cancers diagnosed between 1974 and 1976 (American
Cancer Society, 2006). Although prevalence of cancer is high in the United States,
survival rates are improving. The absolute number of cancer deaths in the United States
decreased for the years of 2005 and 2006 by more then 3,000 cases from 2003 and 2004
(Jemal et al., 2007). The National Cancer Institute estimates that 10.5 million Americans
are alive with some history of cancer (National Cancer Institute, 2007). Thus, cancer and
the treatment of cancer is a major health concern in the United States.

The cancer treatment experience will be unique to each individual. For some patients
a one-time surgical procedure may cure them of their disease. For others, cancer
treatment may involve one or any combination of the following: surgery, chemotherapy
treatments, bone marrow or stem cell transplantation, radiation therapy, biologic therapy
and/or gene therapy. However, once experienced, it is not likely that one will forget the
emotional and physical challenges associated with cancer.

Unfortunately, with a cancer diagnosis and treatment comes the advent of symptoms

that can be troublesome to the patient, their caregivers and their practitioners. Rutledge

and McGuire (2004) supported that symptoms are a phenomena that are
multidimensional, complex, and subjective changes in biopsychosocial functioning,
sensations, or cognition. Symptoms may be present prior to diagnosis, during treatment,
as well as upon the completion of treatment. Treatment may consist of any or a
combination of the modalities listed previously. The treatment itself may cause
symptoms, i.e., fatigue caused by chemotherapy or a skin reaction caused by radiation
therapy. The cancer itself may also cause symptoms, i.e., pain caused by a colon cancer
mass pressing on a nerve. It is not uncommon for patients to experience both symptoms
caused by the disease process as well as the treatment regimen designed to fight the
cancer. For some patients the best treatment option may be palliative care, which may
also contain one of the previously mentioned treatment modalities. Even comfort
measures such as pain medicines have the potential to produce unwanted symptoms on
top of symptoms that may be present from the disease process itself. Symptoms are both
related to the disease process as well as the treatment strategies designed to eliminate the
cancer. Many patients will experience more then one symptom concurrently during the
course of their disease treatment or after they have finished active treatment for their
cancer.

It is due to the multiple nature of symptoms that patients are most likely to seek care
from the health care team (Cleeland, Mendoza, Wang, Chou, Harle, Morrissey, &
Engstrom, 2000; Rutledge & McGuire, 2003). Based on a 2001 study of newly
diagnosed cancer patients (n=841), all 65 years of age or over, those diagnosed with lung
cancer reported an average of 5.3 symptoms, breast cancer patients reported 3.6

symptoms, colon cancer patients reported 3.7, and prostate cancer patients reported an

average of 4.3 symptoms (Given, Given, Azzouz, Kozachik, & Stommel, 2001). The
complex and dynamic nature of concurrent multiple symptoms pose significant
challenges for patients and caregivers, as well as health care providers. The symptom
experience will be unique to each patient. Each patient will experience varying degrees
of frequency, intensity and distress of the symptoms (Dodd, J anson, F acione, Faucett,
Froelicher, Humphreys, et al., 2001a; Lenz, Pugh, Milligan, Gift, & Suppe, 1997).

The dimensions of frequency, intensity and distress of symptoms are typically
measured, researched and treated as they relate to each individual symptom. The
majority of our research to date has focused on individual symptoms (Dodd et al., 2001a).
The effect of the dimensions of symptoms as well as the symptoms themselves changing
over time is of significant concern to patients, caregivers and members of the health care
team. Patients may experience symptoms related to the disease and/or the treatment for
their entire life. Although cancer/chemotherapy symptoms “wax and wane” over the
course of their existence (Sama, 1993), particular patterns are unknown.

Different dimensions of the symptom experience such as fi'equency, intensity, and
distress, will also change over time. For example, Tishelman, Degner, Rudman,
Bertilsson, Bond, Broberger, et a1. (2005) found in a symptom study of lung cancer
patients, completed over six different data collection points following diagnosis, that
symptom intensity (frequency) and symptom distress (how disturbing or troubling a
symptom was perceived to have been) were not equivalent over time. Actual symptom
presentation as well as patient perceptions of the symptoms will change over time.
Multiple, co-occurring symptoms will further complicate this scenario for patients as well

as practitioners.

Researchers have descriptively studied multiple co-occurring symptoms throughout
the chemotherapy experience (Cleeland et al., 2000; Dodd, Miaskowski, & Paul, 2001b;
Kim, McGuire, Tulman, & Barsevick, 2005). In addition, researchers have begun to
examine the effect of single or multiple symptoms as they change over time and their
impact on outcomes such as physical functioning or quality of life (Bender, Ergyn,
Rosenzweig, Cohen, & Sereika, 2005; Cooley, Short, & Moriarty, 2003; Gaston-
Johansson, Fall-Dickson, Bakos, & Kennedy, 1999; Gift, Stommel, J ablonski, & Given,
2003; Given, Given, Azzouz, Stommel, & Kozachik, 2000). The research that has been
conducted on multiple, co-occurring symptoms has typically focused on symptoms
and/or disease states identified a priori (i.e., pain and fatigue for only breast or lung
cancer patients) and their relationship with an outcome (Bender, et al., 2005; Gift et al.,
2003; Given, Given, McCorkle, Kozacik, Cirnprich, Rahbar, & Wojcik, 2002).

As stated above, many patient present with a diagnosis or cancer or at the start of
chemotherapy treatment with multiple co-occurring symptoms. The symptoms may
“wax and wane” throughout chemotherapy or radiation treatment and even persist after
the completion of therapy. Nursing interventions and studies have traditionally focused
on individual symptoms. No randomized clinical trials (RCT) containing a nursing
intervention have been identified that associate the dimensions of the fi'equency,
intensity, and distress of one symptom with these same dimensions for additional
symptoms that are co-occurring. No studies have been found that examine the potential
of a change in dimension of one symptom being able to predict the co-occurrence of

additional symptoms and/or changes in the dimensions of the other symptoms over time.

This study will examine multiple co-occurring symptoms in adults with various types
of solid tumors as well as non-Hodgkin’s lymphoma. The focus will be on three of the
most common 00ng symptoms associated with a cancer diagnosis and
chemotherapy treatment; fatigue, pain and insomnia. This study will examine the
fiequency, intensity and distress for each symptom and associate these dimensions with
the other symptoms presenting concurrently. Thus, as changes occur in one dimension of
one of the symptoms of fatigue, pain, or insomnia, other dimensions of the other
symptoms will be examined for possible changes. Due to the fact that fatigue is such a
prevalent and persistent symptom with chemotherapy, the influence of fatigue and the
management of fatigue as it is associated with the presence of and dimensions of the
other symptoms of pain and insomnia will be of particular interest.

Fatigue

Fatigue has been one of the most widely studied cancer/chemotherapy symptoms. The
symptom of fatigue has been defined as being subjective in nature with patients recalling
sensations of weakness, lack of energy and/or tiredness (Cella, Lai, Chang, Peterman, &
Slavin, 2002; Stone, Richards, & Hardy, 1998). Fatigue in the cancer patient is clinically
different than fatigue experienced in the general healthy population. Cella et al. (2002)
noted in a study comparing the general healthy United States population (n=1010) with
non-anemic cancer patients (n=113) as well as anemic cancer patients (Hemaglobin < 11
g/DL) receiving chemotherapy (n=2369) that fatigue scores at baseline and upon
completion of the study were significantly worse among the three groups respectively

(p<.001).

However, Cella et al. (2002) noted that anemia is not the only variable that contributes
to fatigue in the cancer/chemotherapy patient. Other possible causes include cytokine
production, altered muscle metabolism, sleep deprivation, stress and depression. Nieboer
et al. noted that joint (p < .0001) and muscle pain (p < .0283) were associated with
greater amounts of fatigue in breast cancer patients (n=885) (N ieboer, Buij s, Rodenhuis,
Seynaeve, Beex, van der Wall, et al., 2005). In addition, strong associations have been
noted between increased severity of fatigue and poor performance status (p < .001),
gastrointestinal symptoms (p < .001), pain (p < .05), and insomnia (p < .05) for patients
receiving chemotherapy for non-Hodgkin’s lymphoma (Wang, Giralt, Mendoza,
Engstram, Johnson, Peterson, et al., 2002).

Fatigue is multifaceted and multidimensional. Fatigue is the most common symptom
associated with cancer and its treatment (Cella, Davis, Breitbart, & Curt, 2001).
Prevalence of fatigue for various cancer survivors either undergoing therapy or upon
completion of therapy ranges from 25% to 99% as reported in different studies (Blesch,
Paice, Wickharn, Harte, Schnoor, Paul, et al., 1991; Irvine, Vincent, Graydon, Bubella, &
Thompson, 1994; Servaes, Verhagen, & Bleijenberg, 2002). Fatigue has been recognized
as the most distressing and the most unrelieved symptom of cancer as well as cancer
treatment (Bower, 2005; Cella et al., 2002). Many studies have supported the negative
effect of fatigue on personal, professional and social activities resulting in physical,
psychologic, and economic issues influencing quality of life (Cella, 1997; Cella et al.,
2001; Cella, Tulsky, Gray, Sarafian, Linn, Bonomi, et al., 1993; Ferrell, Grant, Funk,
Otis-Green, & Garcia, 1996; Irvine et al., 1994; Nail & Jones, 1995; Yellen, Cella,

Webster, Blendowski, & Kaplan, 1997). The Fatigue Coalition reported that 81% of

patients who have undergone chemotherapy reported diminished energy as the most
common physical complaint. In addition, 78% of these patients stated that they
consistently required more Sleep and rest. The consequences of this fatigue was reported
to have caused 88% of the respondents to alter their daily routines and 75% of those
employed were required to change their work status due to the fatigue (Bemdt, Kallich,
McDermott, Xu, Lee, & Glaspy, 2005).

There is little doubt that the presence of fatigue is widespread and burdensome to
patients undergoing chemotherapy. Cella et al. (2001) support that fatigue is typically the
initial symptom experienced by patients. In addition fatigue increases with the
progression of cancer and treatment (Cella et al., 2001). Tishelman et al. (2005) found in
a study of lung cancer patients (nflOO) that difficulty breathing, pain, and fatigue were
the symptoms associated with the greatest amounts of distress (how disturbing or
troublesome the symptom is to the patient). In this same study, fatigue remained the
most distressing symptom throughout the entire study period (six time points over 1 year
after diagnosis).

Gift et al. (2003) found in a study of newly diagnosed lung cancer patients (n= 112)
that fatigue was the symptom with the highest ranking for incidence rate at baseline
(upon diagnosis), three months later, and six months following diagnosis. Fatigue is
problematic at baseline and may remain throughout and after the treatment period for
cancer patients. However, fatigue is seldom a symptom appearing in isolation. Past
research has supported that fatigue may effect the severity levels of additional symptoms

experienced by the cancer patient (Given et al., 2001; Miaskowski & Lee, 1999). Pain

and insomnia are two symptoms that have been noted to co-occur with fatigue among
cancer patients.
Pain and Insomnia Associated with Fatigue

Sarna (1993), noted in a study of women with lung cancer (11 = 69) that the three most
prevalent and distressing symptoms included fatigue, pain and insomnia. In fact, 41% of
the women experiencing fatigue concurrently experienced pain and 31% experienced
concurrent insomnia. Degner and Sloan (1995) validated this finding in their study with
434 newly diagnosed cancer patients with a variety of solid tumors. Degner and Sloan
found that the most frequently occurring symptoms include fatigue, pain, loss of appetite,
cough, and insomnia.

Beck and Schwartz (2000) found in a cross-sectional study (n=84) that pain intensity
levels negatively influenced fatigue reports as well as “sleep quality.” Likewise, Given et
al. (2001) supported in a study of solid tumor patients (n= 841) age 65 and older, that
those experiencing fatigue alone reported 4.4 additional symptoms. Those experiencing
pain alone reported 3.8 additional symptoms. However, patients who experienced fatigue
and pain concurrently reported an average of 6.3 “other” symptoms. Given et al. (2001)
also found that 50% (n=161) of the patients that concurrently experienced pain and
fatigue likewise experienced insomnia.

Fatigue, pain and insomnia were the three symptoms selected to examine in this study.
The presence of multiple symptoms having a synergistic effect on outcomes such as
morbidity has been discussed in the literature (Dodd et al., 2001a; Rutledge & McGuire,
2004). Although this study will not be examining morbidity as an outcome, the

associations among three symptoms (fatigue, pain and insomnia) will be examined.

Fatigue appears to be a dominant symptom that may influence the occurrence of other
symptoms, especially pain and insomnia due to their prevalence in the cancer and/or
chemotherapy patient. To best describe and evaluate these symptoms the dimensions of
fiequency, intensity and distress were used to define each symptom cited above.
Frequency was defined as the number of days out of seven days a patient reported the
symptom as being present. Intensity was defined as a patient reported ranking of the
symptom on a scale of 0 meaning symptom not present to 10 meaning the intensity of the
symptom was the worst possible that the patient could imagine. Distress was defined as a
self reported scale with 0 meaning the symptom does not bother the patient in any way to
10 meaning the symptom causes the worst bother for the patient that they can imagine.
The higher the score, as reported by the patient in the 0-10 scale described above,
assigned to the dimension (frequency, intensity and/or distress) the more problematic the
symptom (fatigue, pain, and/or insomnia) is for the patient. It is assumed that appropriate
and effective interventions may cause a change in the reported score for a dimension,
thus, the score would decrease. Of course, the opposite may happen as well, if a patient’s
symptoms are not being properly managed or their condition is worsening their reported
scores may increase. Due to the fact that this study contained an intervention component
it was vital to examine changes in the symptom dimensions over time.

Due to the focus on the symptom of fatigue in this study it is important to examine the
effect of fatigue management on the symptoms of pain and insomnia. To accomplish
this, categories of response to fatigue management (none / mild, non-response, partial
response, and full response), as established in the literature (Given, Given, Jeon,

Sikorskii, in press), were examined as they were related to the dimensions of frequency,

intensity and distress for all three symptoms. These concepts will be further described in
the Research Questions section.
Research Questions
Symptoms associated with cancer and chemotherapy seldom present individually.
Multiple co-occurring symptoms may have a synergistic effect on each other (Rutledge &
McGuire, 2004). Three of the most common symptoms associated with
cancer/chemotherapy include fatigue, pain, and insomnia. Patients will experience
varying degrees of frequency, intensity and distress associated with each symptom.
These dimensions for each symptom and their associations among the symptoms of
fatigue, pain and insomnia were explored in detail. The following research questions
were addressed in this study.
Research Question 1
At baseline observation, prior to entry into a clinical trial, and at 10 weeks is there an
association between the dimension of fiequency of fatigue and frequency of pain, as well
as fiequency of fatigue and fi'equency of insomnia for adults receiving chemotherapy? Is
this association influenced by; age, site and stage of cancer, sex, and co-morbid
conditions? The dimension of frequency is the number of days out of the past seven that
a patient reports the symptom as present. This will be a continuous variable with a
possible range of 0-7.
Research Question 2
At baseline observation, prior to entry into a clinical trial, and at 10 weeks is there an
association between the dimension of intensity of fatigue and intensity of pain, as well as

intensity of fatigue and intensity of insomnia for adults receiving chemotherapy? Is this

10

association influenced by; age, site and stage of cancer, sex, and co-morbid conditions?
The intensity dimension is defined as a ranking of the symptom on a scale of 0 meaning
not present to 10 meaning the intensity is the worst possible. This will be a continuous
variable with a possible range of 0-10.
Research Question 3

At baseline observation, prior to may into a clinical trial, and at 10 weeks is there an
association between the dimension of distress related to fatigue and distress related to
pain, as well as distress related to fatigue and distress related to insomnia for adults
receiving chemotherapy? Is this association influenced by; age, site and stage of cancer,
sex, and co-morbid conditions? The dimension of distress will be defined as a scale with
0 meaning the symptom does not bother the patient to 10 meaning the symptom causes
the worst bother the patient can imagine. This will be a continuous variable with a
possible range of 0-10.

Research Question 4

Using categories of response to fatigue management (none / mild, non-response,
partial response and full response) established in the literature, determine how changes in
the dimensions of frequency, intensity, and distress of pain and insomnia are predicted by
categories of response to the management of fatigue between baseline and 10 weeks? Is
this association influenced by; age, site and stage of cancer, sex, and co-morbid
conditions?

Question 4 allowed the researcher to determine if the change in frequency of fatigue
was predictive of a change in the dimensions of frequency, intensity and/or distress for

the associated symptoms of pain and insomnia. For this question, fatigue intensity levels

11

were converted fi'om continuous variables to categorical variables to facilitate
comparison with fatigue response categories (Mendoza, Wang, Cleeland, Morrissey,
Johnson, Wendt, et al., 1999). Fatigue response categories, as reported in the literature
(Miaskowski, Dodd, West, Paul, Schumacher, Tripathy, et al., in press), define
categorically the change in fatigue intensity levels after an intervention has been
conducted. Response to fatigue management categories include: none / mild, non-
response, partial response and full response. This question examined how changes in the
dimensions of frequency, intensity, and distress of pain and insomnia at 10 weeks are
predicted by a change in fatigue categories from baseline to 10 weeks. Specific measures
will be further clarified in Chapter 4 of this dissertation. All four research questions
contain the association of patient characteristics of age, site and stage of cancer, sex, and
co-morbid conditions with the symptoms of fatigue, pain, and insomnia. These
antecedents will be described in detail below.
Antecedents
Age
Demographic and/or disease characteristics may influence the research questions. A
brief discussion supporting the selection of each variable; age, site and stage of cancer,
sex, and co-morbid conditions follow. The variable of age as it effects a cancer diagnosis
and treatment has received heightened awareness since the development of the National
Cancer Institute Surveillance, Epidemiology, and End Result programs (N CI SEER).
Insights from this program have determined that increasing age leads to the development

of tumors, pharmacology may differ in an older population, as well as the effect of oo-

12

morbidity, previous illness, and disabilities on the management of an older patient with
cancer (Yancik & Ries, 2000).

Tishelman et al. (2005) noted that age (younger) and sex (women) were variables
negatively influencing intensity (frequency) levels of fatigue throughout a study of adults
with inoperable lung cancer (n=400). Age was examined in a study by Degner and Sloan
(1995) (n = 434) with lung cancer to find that older patients experienced less symptom
distress vs. younger counterparts.

Dodd et al. (2001b) studied the symptom cluster of pain, fatigue and sleep
insufficiency in mainly breast and colon/rectal cancer patients during the first three
chemotherapy treatments (n=93). Dodd et al. (2001b) found that age explained 11.8% of
the change in functional status (p < 0.001) for these patients between baseline and the end
of the third cycle of chemotherapy. Pain was able to explain 10.7% of the change (p =
0.002), fatigue explained 7.3% of the change in functional status (p = 0.011), and sleep
insufficiency was statistically not significant (p = 0.344). On the contrary, Given et al.
(2001) noted no relationship between advancing age and pain or fatigue in a sample of
solid tumor chemotherapy patients all over the age of 65. Therefore, this conflicting
information is rationale for the selection of age as an antecedent variable.

Stage of Disease
The stage of disease was noted by Gift et al. (2003) to be predictive of the number of
symptoms that co-occurred with fatigue six months following a diagnosis of lung cancer
(n=112). Gift et al. (2003) used a subset of lung cancer patients from the Given et al.
(2001) study cited above. Given et al. (2001) reported in the larger data base including

all tumor types that age, co-morbid conditions, and site of cancer failed to interact among

13

each other to produce a significant prediction of any combination of pain or fatigue.
However, stage of disease when examined categorically as early or late stage was
significant for predicting those who reported pain and fatigue vs. only those reporting
pain alone (Given et al., 2001).
Sex

Sex has become a variable of interest in the cancer literature. Discussion related to
sex ranges from examining how sex influences treatment choices to demographic
findings supporting that male sex can be considered and independent negative prognostic
factor for lung cancer survival (Fu, Kau, Severson, Kalemkerian, 2005; Moynihan, 2002).
Numerous authors have found that women are more likely to experience pain and fatigue,
or each alone when compared to neither symptom (Degner & Sloan, 1995; Given et al.,
2001; Tishelman et al., 2005). Thus, the sex variable may indicate tolerance for therapy
as well as provide prognostic value.

Co—morbid Conditions / Site of Cancer

The presence of co-morbid conditions were found to positively correlate with
symptom distress and was found to be a significant predictor of symptom distress in
women with lung cancer as reported by Sama (1993). Given et al. (2001) noted that
“patients with three or more comorbid conditions were more likely to report pain, fatigue,
or their combination, when compared with those reporting neither symptom” (p. 462).
Given et al. (2001) also noted mean number of symptom variations per differing sites of
cancer. The variables of age, stage, sex, co-morbid conditions and site of cancer will be
important variables in examining the effect of associations among the dimensions of

frequency, intensity, and distress for fatigue, pain and insomnia.

l4

It is essential to note that participation in a cognitive behavioral intervention is also a
variable included in the data base for this study. This study examined data from 2 NCI
supported longitudinal panel studies of adult cancer patients undergoing chemotherapy
(ROI CA-79280 {study number one} & ROI CA-30724 {study number two}). Both
studies contained cognitive behavioral interventions. These interventions will be
described in detail in Chapter 4.

Purpose of the Research

The purpose of this research was to examine the dimensions of frequency, intensity,
and distress of the co-occurring symptoms of fatigue, pain and insomnia as they occurred
at two different data collection points in two randomized cognitive behavioral
intervention clinical trials. This study answered the questions: At baseline observation
and at 10 weeks is there an association between the dimension of fi'equency, intensity,
and distress of fatigue and frequency, intensity and distress of pain and/or insomnia for
adults receiving chemotherapy? Each dimension was analyzed in a separate question as
described earlier. All research questions examined the association between the symptoms
as possibly being influenced by the variables of age, site and stage of cancer, sex, and co-
morbid conditions.

This research studied these variables within the context of an adapted Symptom
Experience Model (SEM) (Armstrong, 2003). The SEM has been adapted to include the
dimensions of time and an intervention to accommodate the chronic nature of the cancer
diagnosis as well as the longitudinal, randomized control/intervention design of the
studies used as the database for this study. In Chapter 2, the conceptual model and the

adapted Symptom Experience Model by Armstrong (2003) will be presented and

15

discussed. The support for the adaptation of the model to include the dimensions of time
and an intervention will be explained. In Chapter 3, a review of the oncology multiple,
co-occurring symptom literature including the symptoms of fatigue, pain and insomnia
will be described. The sample, recruitment process, measures, and methods for data
collection will be presented in Chapter 4. The research findings will be presented in
Chapter 5. Implications for clinical practice, study limitations and future direction for
nursing research based on this work will also be presented in Chapter 5. Ultimately, this
research may lead to future studies intended to design and test effective interventions to
manage and/or prevent the individual and/or multiple, co-occurring symptoms of fatigue,
pain and insomnia, in adult chemotherapy patients, thus, improving fitnctional status,

quality of life and/or economics (Kim et al., 2005).

16

Chapter 2
Conceptual Framework

In this chapter an adapted version of the Symptom Experience Model (SEM) by
Armstrong (2003) will be presented and described. The rationale for selecting the SEM
will be presented, followed by an explanation of the need to adapt it to this research. The
adapted SEM includes the addition of a time and intervention component. The addition
of the time and intervention components will be described in the context of the
relationship of the co-occurring symptoms of fatigue, pain and insomnia within the
cancer disease and treatment trajectory. The measurements of symptoms in this research
occurred at two different data collection points; the effect of a cognitive behavioral
intervention was also evaluated.

Intervention Adaptation to the Model

The addition of the intervention component to the SEM model will allow the model to
guide this project. Two distinct interventions, a nursing intervention (Nurse Assisted
Symptom Management, NASM) and/or an automated telephone symptom monitoring
system (Automated Telephone Symptom Management, ATSM) intervention was
implemented in the original studies that served as the database for this study. These
interventions will be described in Chapter 4. The addition of the intervention will also
promote utility of this adapted SEM for future studies as nursing research moves beyond
descriptive symptom studies into randomized clinical trials designed to demonstrate the
effect of antecedents and interventions on outcomes.

To begin, a brief look at the development of the SEM by Armstrong (2003). Figure 1

describes the SEM as created by Armstrong (2003). Armstrong used the Walker and

17

Avant (1995) method of concept analysis to examine the effect of multiple symptoms on
the cancer patient’s symptom experience. An extensive literature review was conducted
examining the concept of the symptom experience, defining attributes, antecedents,
consequences, a model case and other cases, and conclusions and implications for nursing
and research. Four common themes guided by research conducted by Leventhal and
Johnson (1983), Rhodes and Watson (1987), Lenz et al. (1997), and Lenz, Suppe, Gift,
Pugh, & Milligan (1995) provided the foundation for the creation of the SEM. The
themes included: symptoms are subjective in nature, symptoms are a departure from
normal functioning, they are multidimensional, and they include an emotional response
(Armstrong, 2003).

Armstrong used the classic Rhodes and Watson (1987) definition of symptoms as the
foundation for her model; “ . . . subjective, phenomena regarded by individuals as an
indication of a condition departing from normal function, sensation, or appearance or as
perceived indicators of change in normal functioning as experienced by patients” (2003,
p. 601). In addition, the Armstrong model is rooted in the premise that symptoms occur
concurrently and refers “. . . to the experience of multiple symptoms as the ‘symptom
experience’” (2003, p. 601). Similar to the work of Lenz et al. (1995, 1997) and Dodd
(2001a), Armstrong included the dimensions of frequency, intensity, distress and
meaning of the symptoms in the model. These dimensions are referred to as concepts
that further delineate each symptom that make up the symptom experience. Armstrong
applies the dimensions to each symptom in “. . . a multidimensional experience that can

be measured separately or in combination with other symptoms” (2003, p. 602).

18

 

 

 

 

 

 

 

 

 

 

 

e

 

52:35. 2533558”
58:: 5.9383. 5.23.
33 3.2388. ea 8:8

 

 

2 8.53:8

 

 

mac—can 3.2.2.8

 

         
     

   

    

    

  
 

$332..» Basso

magmas. 3823

  

          
    

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

258336 _ 33:83 «5:83 33:83
28:33 33.38 96332.
5.3333..." 3332.38”
89 332. 333. 38...... o
l 88. 2.83. 3.9 8:83? «5:22: > 383303
ea «8.88338 was.» «89.32
53.53
h 9396 _ 22333 8 =58”
2.823
233.. 25333333 gusaV \3323. n 95.3 a =3
33 ea «88. .35 a 2.8? N m z. a
32:. man 832% 38.8. v Jung...” "Mafia 8
23 gm .
ea 25.8. .883 2.9.5....» 233$ 35253. «ES
.5828 38:26 288.. 28888:

 

 

 

_ mags". _

19

Emacs N. .25 $8an mxgngoa Koda—

dfi mEEoB mermgoo“ > 00:83 team—5mm... 3 >338? am; 88. 05888, Zfimgm mode? mo Q5. u. mow. 003453

was 3, 05855, 2:35 moans»

The SEM is designed to serve as a guide for supporting practice and research related
to multiple, co-occurring symptoms. This model was originally created for application
with an oncology population. These two statements represent rationale for selection of
the SEM for this study. Additional rationale will be described while examining each
component of the model. In addition, the elements of time and an intervention will be
discussed. The following sections of this chapter will describe the model (fi‘om lefi to
right) with application for this research identified within each component of the model.

The Symptom Experience Model
Antecedents

Various antecedents initiate the symptom experience when viewing the SEM fi'om lefi
to right. Antecedents are used to clarify both the attributes and the context in which the
symptom experience is found (Armstrong, 2003; Walker & Avant, 1995). Antecedents
are characteristics that the patient brings to the symptom experience such as demographic
characteristics, disease characteristics, and individual characteristics (Armstrong, 2003).
Theoretically, antecedent components include: physiologic, psychological, and
situational factors that are the initial contributors to the symptom experience (Rhodes &
Watson, 1987). Likewise, Lenz et al. (1997) incorporated physiologic antecedents into
their classic symptom model, the middle range Theory of Unpleasant Symptoms (TOUS).
Armstrong (2003) labeled the antecedents in the SEM as demographic characteristics,
disease characteristics, and individual characteristics.

An additional classic model, the revised Symptom Management Model (SMM) by
Dodd et al. (2001a) likewise depicts various domains that influence the symptom

experience, management strategies and outcomes. An example from the SMM that is

20

closely replicated in the SEM is the environment domain that “ . . . refers to the aggregate
of conditions or the context within which a symptom occurs; that is, it includes physical,
social, and cultural variables” (Dodd et al., 2001a, p. 671). Demographic characteristics
in the antecedent section of the SEM include age, sex, marital status, race, culture, role,
education, and socioeconomic status. Disease characteristics include type and state of the
disease, type of treatment and co-morbid medical and clinical factors. As an example, in
breast cancer the disease characteristics such as tumor size, disease stage, and HERZ/neu
status assists in determining the type of treatment a patient receives, thus, influencing the
symptoms the patient may experience. Additional characteristics that are considered
antecedents include such areas as health knowledge, values, past experiences, and sense
of coherence. An example of the influence of health knowledge, values and past
experiences may include the breast cancer patient who chooses a bilateral mastectomy
although disease is found at an early stage in only one breast. This patient may feel very
strongly per her health belief system that prophylaxis is an optimum approach for dealing
with a possible future cancer, in addition, she may have an extensive family history of the
disease that has influenced her values and past experience. Recall that the numerous
antecedents cited in the SEM (demographic, disease characteristics, and individual
characteristics) support the idea that multiple, co-occurring symptoms may or may not
share the same etiology (Kim et al., 2005). The etiologies, as well as other antecedents,
play a role in the production of symptom(s).
Symptom Production
Antecedents contribute to the production of the symptom. Symptom production is the

next component of the model (moving lefi to right, after antecedents). Theoretically,

21

Armstrong (2003) cites the work of Leventhal and Johnson (1983) to define symptom
production as an objective, concrete representation of the disease. Although not clearly
articulated by Armstrong, the SEM likely depicts symptom production as the actual
display of a physical response or a psychological reaction that takes into account the
symptom antecedents. As noted in the model, antecedents lead to the production of a
symptom and the perception of the symptom by the patient.
Symptom Perception

Symptom perception is multidimensional in nature. Theoretically, dimensions that are
common across various symptoms and populations delineate symptom perception, for
example, the frequency or intensity of the symptom. Dodd et al. (2001a) report that the “.
. . gold standard for the study of symptoms is based on the perception of the individual
experiencing the symptoms and his/her self-report” (p. 669). Within the Dodd et al.
(2001a) SMM the symptom experience is composed of an individual’s perception of the
symptom as well as evaluation of the meaning and response to the symptom. McCorkle
and Young (1978) provide an excellent example of the individualized perception of
symptom assessment. These authors describe a nurses’ perception of a lung cancer
patient as struggling to breath and coughing, however, the patient does not acknowledge
that these symptoms pose a problem. In this example, this patient’s dyspnea has “become
a way of life for him” (p. 375). Thus, patient perception of the distress created by the
symptom is crucial for an accurate assessment and ultimately providing an optimal
intervention and outcome. Patient perceptions of symptoms are typically assessed via

three dimensions; frequency, intensity and distress (Armstrong, 2003; Lenz et al., 1997).

22

Similar to the SMM and TOUS models cited above, the SEM includes the three
dimensions of the symptom experience; fi'equency, intensity and distress. Lenz et al.
(1997) hypothesize that the dimensions of timing (frequency), intensity, and distress are
“. . . assumed to be separable but related to one another” (p. 15). In addition, the concept
of symptom meaning is considered a dimension of the symptom perception phase of the
SEM model. The dimensions of frequency, intensity and distress will be applied to an
adapted version of the SEM to be used with this research and described below.

Frequency

Conceptually, frequency has been defined by Lenz et al. (1997) as a component of a
time dimension that notes the occurrence of the symptom in terms of being intermittent,
persistent, or a combination of both. An example would appear as: “During the past
seven days, how many days did you experience (the symptom)?” In this study frequency
was represented as the number of days out of seven that each symptom was present. The
range for fi'equency for each symptom was 0-7 with the exception being research
question number four that converted the continuous variable into a categorical variable as
described in Chapter 4. Note that this study was longitudinal in design; therefore, it is
appropriate to evaluate fi'equency in terms of the “last seven days.” The question was
worded as: “During the past seven days, on how many days did you experience pain?”
Frequency refers to an actual count of number of days that the patient experienced the
symptom. In the case of this research the maximum value for the range will be seven,

representing days experiencing the symptom in the past week.

23

Intensity

The second dimension evaluated as a component of the symptom perception in the
SEM is intensity. Intensity was defined by Lenz et al. (1997), as the severity, strength, or
amount of the symptom being experienced. In this study intensity was ranked by the
patient describing the severity of the symptom on a scale of 0 — 10. Zero would mean
that the symptom was “not present” to 10 meaning “worst possible.” Again this is the
patient’s perception of how strong or severe the symptom is, for example; “On a scale of
0 = not present to 10 = the worst it could be, how severe is pain for you?” Intensity was
the level of severity assigned to the symptom. This severity score was based on the
patients’ perception of past experiences with no severity (a 0) associated with a symptom
with no intensity value versus their perception of what the worst possible level of severity
may be (a 10).

Distress

The concept of distress was conceptually defined by Lenz et al., as “. . . the degree to
which the person is bothered by it” (1997, p. 16). This concept of distress has been used
to define the “degree of discomfort” a patient is experiencing (McCorkle & Young,
1978). This study and the SEM viewed the distress dimension as a measurement of the
patient’s perception of the burden or interference that the symptom causes in their
everyday life. In this study an example is; “How much did (the symptom) interfere in
your life?” Patients rank their response on a scale of O — 10, 0 represents no interference
to 10 meaning completely interfering. Interference in this study is in the context of; to
what degree the symptom disrupted the patient’s ability to do what they wanted to do.

For example; “On a scale of 0 = did not interfere to 10 = completely interfered, overall,

24

how much did pain interfere in your life?” Armstrong (2003) noted that one’s ability to
perceive physical or mental distress is influenced by various factors; age, socioeconomic
levels, culture, family role, education, health knowledge, values and past experiences.
Distress is commonly associated with dimension of meaning.

Meaning

The dimensions of frequency, intensity and distress were included in this study. The
dimension of meaning, although it is depicted in the SEM, will not be included in this
study. The study was a secondary data analysis; the original data set did not contain
items that specifically addressed situational and existential meaning. The concept of
“meaning” is discussed only as a component of the SEM conceptual model. For the
purpose of this study the dimension of distress is sufficient to cover the concept of
interference or burden with daily life activities. These concepts are similar to situational
and existential meaning.

Armstrong defines meaning as being assigned to physical sensations and “. . . may
have profound implications for their physical and psychological health and, therefore
their quality of life” (Armstrong, 2003, p. 602). The Armstrong SEM supports two types
of meaning, situational and existential, this fiom the work of Richer and Ezer (2000).
Situational meaning refers to the “. . . perception of a new event and their capacity to
handle it” (Armstrong, 2003, p. 603). An example of situational meaning is the lack of
ability to go to work due to fatigue caused by chemotherapy. Existential meaning refers
to “. . . global representations of their places in the world” (p. 603). An example of

existential meaning provided by Armstrong is the patient developing a sense of

25

vulnerability and mortality due to symptoms that remind them of the cancer diagnosis (p.
603). The dimension of meaning was not evaluated in this research.

Encapsulated within the meaning circles of the SEM, symptoms are perceived either
individually or related to each other, refer to Figure l. The SEM depicts multiple, co-
occurring symptoms as three overlapping circles that include the underlying dimensions
of frequency, intensity, distress, and meaning assessed for each symptom. The SEM
depicts three overlapping circles not only as a means of defining that the dimensions must
be assessed for each symptom, however, also as a representation of multiple, co-
occurring symptoms being experienced by the patient.

Multiple Symptoms

Multiple symptoms co-occur, as noted by overlapping circles, see Figure 1. Each
symptom individually possesses the dimensions of frequency, intensity and distress.
However, the circles representing each individual symptom are connected with arrows
and encapsulated as a group, thus, the theoretical perspective of multiple, co-existing
chemotherapy/cancer symptoms. For this study fatigue, pain and insomnia will each
have their own circle. The dimensions of frequency, intensity and distress will be
identified for each symptom respectively. The associations between the symptoms will
be evaluated as depicted by the arrows connecting the symptoms in the SEM, refer to
Figure 2.

The numerous antecedents in the SEM may serve to clarify the etiology of the co-
occurring symptoms. As an example the patient may present with pain, fatigue, and
insomnia all co-occurring. The overlapping nature of the symptoms (represented by

circles in the SEM) encapsulated in the symptoms and existential meaning circles depict

26

the stable and independent nature of the co-occurring symptoms. Arrows between the
symptoms (circles) represent the strong relationship among the symptoms that are co-
occurring. Understanding the symptom experience requires nurses to view symptoms as
influencing and being influenced by other symptoms (Armstrong, 2003). This research
examined changes in the frequency, intensity, and distress of symptoms (fatigue, pain and
insomnia) that co-occur over time and have been affected by antecedents such as age,
stage and site of cancer, sex, and co-morbid conditions.
Symptom Expression

The SEM depicts symptoms as influencing the expression of the symptom and
ultimately the consequences section of the model, see Figure 1. Armstrong provided no
definition for the symptom expression component of the model. However, the symptom
expression component may be described as the physical, psychological, and emotional
response to a symptom. In this study, one component of symptom expression was the
response to fatigue management. Response to fatigue management was measured
categorically as none / mild, non-responder, partial responder, or full responder. These
categories will be described in detail in chapter 4. Symptom expression leads to
consequences of the symptom.

Consequences

The SEM consequences include the concepts of: adjustment to illness, quality of life,
mood, functional status, disease progression, and survival. Each concept cited as a
consequence has its own theoretical definition. For example, quality of life may
theoretically be described as the patient’s perception of their ability to participate

physically, emotionally, and socially in activities that promote well being (Cella &

27

Cherin, 1988). This study did not evaluate any consequences of the symptoms. This
study focused on select antecedents and their influence on the frequency, intensity and
distress associated with fatigue, pain, and insomnia.
Adaptation of the SEM

The SEM has been adapted to include only the elements needed to conceptually guide
this research and the addition of the components of time and an intervention. The
antecedents used in this study included: age, stage and site of cancer, sex, and co-morbid
conditions. Refer to Figure 2 for the revised SEM including the antecedents listed above.
As in the original model, symptom production will follow (from left to right) beginning
with the antecedents. Symptom perception will follow symptom production and include
the three overlapping symptom circles. Each circle will represent each of the symptoms
of fatigue, pain, and insomnia. The dimensions of fiequency, intensity and distress are
represented for each symptom. The circles are connected by arrows and overlap
representing the concept of multiple symptoms that co-occur.

Intervention
An additional adaptation for the SEM model is to include an intervention. As

supported by the 2005 ONS White Paper for Nursing-Sensitive Patient Outcomes,
“. . . nursing interventions play a vital role in preventing or minimizing symptoms and
complications during all phases of cancer care (positive outcomes sensitive to nursing
care)” (Given & Sherwood, 2005). The intervention for this study will be described in
detail in Chapter 4. The intervention designed for the studies that were the source of data
for this study were targeted directly at each symptom. For example, if a patient

experienced fatigue, pain, and insomnia they would receive instruction per the nurse on

28

self care management for each symptom. Due to the fact that the intervention directly
targets the symptom perception the intervention section of the model will follow (from
left to right) the symptom perception component of the revised model, see Figure 2.

As stated previously, it is rare that a single symptom will present in the oncology
population. The Symptom Experience in Time (SET) model by Henly, Kallas, Klatt, &
Swenson (2003) describes interventions as “. . . operations that affect the target symptom
or the symptom constellation. . .” (p. 413). The intervention may differ for each
individual symptom or possibly one intervention may be appropriate to manage multiple,
co-occurring symptoms. Interdisciplinary strategies may be used to intervene with
symptom pairs or clusters due to the complex nature of co-occurring symptoms (Parker,
Kimble, Dunbar, & Clark, 2005). Interdisciplinary involvement necessitates the need for
conceptually derived clear documentation of the specifications of the intervention. The
specifications will assist in the replication of studies that include interventions (Dodd et
al., 2001a). Interventions are displayed in the Dodd et al. (2001a) SMM as symptom
management strategies with “specifications” such as when, how much, and to whom.

In response to the need to include an intervention and specifications for use with the
study data base, the SEM has been adapted to include an intervention component that
consists of a set of thorough descriptors (what [nature of the strategy], when, where, why,
how much [dose], to whom [recipient], and how [delivery means]) (Dodd et al., 2001a).
Refer to Figure 2. The descriptors describe the intervention as well a means of

facilitating study replication (Dodd et al., 2001a).

29

Time

The researcher is challenged as to the appropriateness of one time only measurements
for co-occurring symptoms due to the chronic nature of cancer, treatment approaches that
are delivered over time, and the knowledge that symptoms change over time. Issues to
consider in examining symptoms over time include the timing of the symptom report in
relation to the diagnosis and/or treatment plan. Patients diagnosed with advanced stage
disease will likely present with a different symptom profile versus an early stage disease
patient. The tumor type will also play a significant role. As stated earlier, Given et al.
(2001), noted in a study (n=841) of newly diagnosed cancer patients that individuals
diagnosed with lung cancer presented with an average of 5.3 symptoms versus those
diagnosed with colon cancer presented with an average of 3.7 symptoms. In this study,
the co-occurring symptoms of fatigue, pain and insomnia was evaluated over 2 different
points in time. In addition, antecedents such as age, site and stage of cancer, sex, and co-
morbid conditions were evaluated in relation to how they were associated with symptoms
over time.

The site and stage of cancer were crucial pieces of data in this study. It is essential to
understand the clinical context of the symptom (Barsevick, Whitmer, Nail, Beck, &
Dudley, 2006). Symptoms may be indicative of the disease state, the treatment, or both.
For example, fatigue may be associated with the lung cancer diagnosis or the myleo-
suppressive effects of the chemotherapy. It may not even be possible to determine the
etiology of the symptom. If the symptom is associated only with the treatment it is likely
to predictably appear, remain, peak and/or dissipate (Barsevick et al., 2006). The

symptom may be representative of a late effect of prior chemotherapy or radiation

30

therapy. For example patients with mantle field radiation for Hodgkin’s disease may
experience respiratory complications/symptoms later in life as a result of their field of
radiation years prior. It is possible that some symptoms may be permanent, for example,
hair loss due to radiation therapy. Therefore, the symptom assessment instrument as well
as timing of the administration of this assessment instrument within the overall study
design is crucial. These examples reinforce the need to acknowledge the concept of time
in the adapted SEM model to guide this research.

Studying multiple symptoms requires examining the temporal nature of the symptoms
across the specified study trajectory (Dodd et al., 2001a). Time plays a key role in
symptom presentation as well as management for the oncology patient. Symptoms may
change over time depending on if they are treatment or disease induced. As an example,
dyspnea and loss of appetite due to a diagnosis of non-Hodgkin’s lymphoma may be
resolved after a few cycles of CHOP chemotherapy. The patient’s bulky neck and chest
disease may be decreased and the Prednisone may increase a patients’ appetite. However
this same patient may go on to experience fatigue and nausea as a result of the CHOP
chemotherapy. Symptoms linked to the treatment (i.e., chemotherapy) may be dose
dependent and/or schedule dependent. For example, a patient receiving a 28 day cycle of
chemotherapy may experience less fatigue than a patient receiving dose dense (every 14-
day) chemotherapy. Factors such as concurrent radiation therapy, growth factor
administration, and if the chemotherapy is neo-adj uvant, adj uvant, or palliative may also
impact the symptoms. Thus, antecedents such as stage of disease and treatment approach
are very important, as well as the treatment and disease trajectory in effecting the

identification and stability of the symptoms over time.

31

 

 

 

 

 

 

 

 

 

 

 

 

 

  

  

 

     

 

      

 

 

 

   
    

 

 

 

 

 

 

 

 

 

     

 

 

 

 

 

>24m0m0m24fl m<2=udog m<31403 _2.—.m_~<m2._._02 fl<g1403
1500.50.20: 1m50m14=vz mX15mm@_Oz
>mo
mSmG <55»
max 33:38.. in:
00583:? 5.035 «$88
9.8 cm 0309. 653mm 23,
mos. Zea:
luv fio $508
$92
5832; of
20:0 \ 7&5
33.53% Tense—3% 20.: Womvonafl.
38:me 58:me dem— ”8325.
930mm . 9.38m we: Womwosaan
H dam

 

 

Ewes“ N. >339” wmg

32

This SEM was adapted to include a time arrow reflective of a feedback loop from the
symptom expression section of the model back to the symptom perception circles of the
model, refer to Figure 2. The addition of the time component was influenced by the work
of Henly et al. (2003) as stated above. Also, the TOUS by Lenz et al. (1997) was one of
the first nursing models to represent multiple, co-occurring symptoms as well as a
feedback loop from the outcome of performance back to the physiologic, psychologic,
and situational factors and the symptoms. The Symptom Management Model depicts a
feedback loop represented as a bi-directional arrow representing that the patient
experiencing an outcome fi'om the symptom necessitates that one will “modify” or
“refine” intervention strategies (Dodd et al., 2001a).

The addition of a time component to the SEM allowed the researcher to evaluate a
nursing intervention in relation to a particular symptom or group of symptoms at any
given point in time along the treatment or disease trajectory. As applied to this
longitudinal oncology nursing research, the time arrow would allow the researcher to
evaluate the co-occurring symptoms of fatigue, pain, and insomnia at baseline and 10
weeks into therapy. The time arrow turns the previous static SEM model into a dynamic
one that facilitates the evaluation of symptoms over time as well as the impact of a
nursing intervention on the symptom dimensions (fi'equency, intensity, and distress).

Summary

The concepts of time and interventions are crucial when studying a treatment
trajectory within a chronic illness, such as cancer, with multiple cycles of chemotherapy
as a treatment. The adapted SEM, with the concepts of time and an intervention added

describes multiple symptoms as well as nursing interventions over time. Timing of the

33

 

measurement of the symptoms as well as the intervention is dependent upon the clinical
context of the subject group and research hypothesis. The clinical context determines
when symptoms are likely to appear, peak and/or dissipate based on landmarks of the
treatment and/or intervention and may be dependent upon how homogeneous the sample
is (Barsevick et al., 2006). Thus, the revised SEM contains the essential elements to
capture timing of the intervention as well as its effect on future symptoms.

The study of multiple symptoms requires a conceptual framework that matches the
disease trajectory, over time, as well as interventions and their effects on future
symptoms. It is essential for oncology nursing research to utilize a conceptual model that
not only complements the research design, however, offers the closest representation to
the clinical context of the multiple symptoms. The need for oncology nursing research as
well as practice to comprehensively monitor and manage symptom clusters over the
disease trajectory necessitates the use of a conceptual model that includes a time and
intervention component (Kim et al., 2005). The SEM as adapted in this study may best
address this need to monitor and manage multiple symptoms in the context of this study

as well as future studies and clinical practice.

34

Chapter 3
Literature Review
This chapter will present a thorough review of the oncology literature published to
date that is related to fatigue, pain, and insomnia as these three symptoms have the
potential to occur together over time for the patient receiving chemotherapy. The adapted
SEM will be used to guide this literature review. The components of the adapted SEM;
antecedents (age, stage and site of cancer, sex, co-morbids) symptoms perception
(fatigue, pain, insomnia) and the intervention and time components will serve as the
headings for this literature review. Refer to Figure 2 for the adapted SEM. Descriptive
studies as well as longitudinal randomized clinical trials (RCT’s) including interventions
will be included. The selection of studies/literature was based on scientific merit,
congruence with the adapted SEM model, and similarity to the research questions.
Antecedents
When viewing the adapted SEM fi'om left to right the first component is the
antecedents. The antecedents of interest in this study include; age, sex, stage and site of
cancer, and co-morbid conditions. These variables are frequently examined in the
demographic sections of both descriptive as well as RCT’s. The following literature
review will examine each antecedent listed above as it relates to the symptoms of fatigue,
pain, and insomnia, as well as the concept of co—occurring symptoms.
Age

This section will focus on the impact that age has on the symptoms of fatigue, pain,

and insomnia as well as co—occurring symptoms in general. Physiologically, age has

been linked to fatigue by factors such as decreased muscle mass, muscle morphology, as

35

well as changes in energy metabolism and neuromuscular activation (Ratel, Lazaar,
Williams, Bedu, & Duche, 2003). Sama (1993) was one of the first researchers to
examine the effect of age on symptom distress. Sama used the Symptom Distress Scale
(SDS) to study 69 women diagnosed within 2 years with lung cancer, 43% were actively
receiving treatment during the study period. The average age of women in this study was
61 years (SD = 11). The most prevalent and distressing symptoms were fatigue, pain and
insomnia. Sama found that symptom distress was not correlated with age when
examined as a continuous variable or as a categorical variable with age 65 as the cut point
(1993).

Degner and Sloan (1995) studied 434 newly diagnosed cancer patients, mean age 59.3
years (SD = 13.9), 52% were male and patients were diagnosed as having solid tumors or
lymphomas. A weak correlation was noted between age and symptom distress, as
evaluated by the SDS (r = 0.11). Younger patients experienced a greater reported amount
of symptom distress vs. their older counterparts (p = .02).

Dodd et al. (2001b) studied a convenience sample for the identified symptom cluster
of pain, fatigue and sleep insufficiency in mainly breast (45%) and colon/rectal (27%)
cancer patients during the first three chemotherapy treatments (n=93). The average age
of this sample was 55.4 years (SD = 14.6). The authors found that age explained 11.8%
of the change in functional status (p < 0.001) for these patients between baseline and the
end of the third cycle of chemotherapy. As an additional finding, pain was able to
explain 10.7% of the change (p = 0.002) in functional status and fatigue explained 7.3%
of the change (p = 0.011), sleep insufficiency was statistically not significant (p = 0.344)

(Dodd et al., 2001b).

36

Given et al. (2001) studied an inception cohort of 841 patients from various
community oncology centers throughout Michigan and one site in Indiana. All patients
were age 65 or older and 31% of the patients were over the age of 75. Patients were
receiving chemotherapy and were diagnosed with breast, lung, prostate or colon cancer.
This study used The Symptom Experience Scale (SES), developed by the authors for this
research project, to assess patients fatigue and pain. Although age was presented
categorically, it was entered into the model as continuous data to reveal no relationship
between age and fatigue or age and pain.

Tishelman et al. (2005) studied 400 patients in Sweden with inoperable lung cancer
following diagnosis through the course of one year for a total of six different data
collection periods. Two of the most prevalent symptoms identified in this sample were
fatigue and pain. The purpose of the Tishelman et al. study was to determine if patients
reported patterns of symptom frequency would be similar to patterns of symptom
intensity. Tishelman et al. (2005) used the SDS instrument and reported that the mean
age of their subjects was 64.5 years (SD = 10.0). The authors did report that their sample
was younger than data provided by the cancer registry (mean age 68.8 years) and that the
ages for males, 52% of the sample (mean age 66.2) and females (mean age 62.6) were
statistically different (p = .001 ). Tishelman et al. (2005) noted that age (younger) and sex
(women) were variables that negatively influenced intensity (frequency) levels of fatigue
throughout their study.

Sex
Tishelman et a1. (2005) found that female patients in their study reported significantly

higher levels of fatigue and pain at the final data collection point (one year post

37

diagnosis) vs. their male counterparts. Women were more likely to be divorced and
widowed vs. men. In addition, women were more likely to have received chemotherapy
in this study, 84% vs. 72% (p = 0.006) of the men receiving some form of chemotherapy.
No significant differences were noted for education level, type of lung cancer or stage of
lung cancer.

Degner and Sloan (1995), noted that overall, women reported greater symptom
distress than the male subjects in this lung cancer study (t = -2.05, p = 0.04). The Given
et al. (2001) study, cited above, found that after controlling for other demographic
variables such as co-morbid conditions, age, and site and stage of cancer, that women
were more likely to experience pain with fatigue or each symptom alone when compared
with neither symptom versus their male counterparts.

Cooley, Short and Moriarty, (2003) completed a secondary analysis from three
different data sets (n=117) that used the SDS to study adults with lung cancer. The
purpose of the Cooley et al. (2003) study was to describe which symptoms were reported
as being the most distressing, what was the prevalence of these symptoms, how did the
symptoms change over time, and examination of patient/clinical characteristics. The
Cooley et al. (2003) study contained 54% men and found fatigue and pain to be the most
distressing symptoms for all patients. Cooley et al. failed to support sex as a patient
characteristic that was related to symptom distress over time. However, at baseline sex
(being female) was noted to predict greater symptom distress at baseline (data was
collected at baseline, three months and six months post diagnosis) (Cooley et al., 2003).
Based on the data presented above, females experience higher levels of pain and fatigue.

To further examine this association between sex and symptoms this current study

38

examined the association between sex and the symptoms of fatigue, pain and insomnia
over time.
Stage and Site of Cancer

Early work in examining the effect of site of cancer on prevalence and intensity of
symptoms was conducted by Vainio and Auvinen, (1996). Vainio and Auvinen together
with the international members of the Symptom Prevalence Group prospectively studied
1840 cancer patients in seven hospices in the United States, Europe and Australia. The
purpose of the Vainio and Auvinen study was to estimate the prevalence of pain and eight
other common symptoms in a population with advanced cancer upon admission to a
palliative care/hospice unit. Severe pain (using a scale with four grades: none, mild,
moderate, severe) was noted in 51% of the sample. Statistically significant (p = 0.003)
differences were noted in pain intensity according to primary site of the cancer. Severe
pain was most commonly reported by prostate cancer patients. Gynecological cancer and
head and neck cancer patients reported the greatest prevalence of moderate and severe
pain. Overall, Vainio and Auvinen fount that“. . . there were statistically significant (p <
0.0001) differences between primary sites in the prevalence of all other symptoms except
constipation, insomnia, and confusion” (1996, p. 6).

Dodd et al. (2001b) noted in their sample of 93 outpatients with various solid tumors,
that the presences or absences of metastases did not have an effect on the severity of
fatigue, pain or insomnia. However, stage of disease was noted by Gift et al. (2003) to be
predictive of the number of symptoms that co-occurred with fatigue six months following
diagnosis of lung cancer (n=112). Gifi et al. stated; “The stage of cancer at diagnosis is

the best predictor of symptoms later in the disease” (2003, p. 393).

39

Gift et al. (2003) used a subset of lung cancer patients from the Given et al. (2001)
study cited above. Given et al. (2001) noted in the larger data base, including all tumor
types, that site of cancer failed to interact among other demographic variables to produce
a significant prediction of any combination of pain or fatigue. However, stage of disease
when examined categorically as early or late stage was significant for predicting those
who reported pain and fatigue vs. only those reporting pain alone. Given et al. (2001)
also noted mean number of symptom variations per differing sites of cancer. Patients
diagnosed with lung cancer reported an average of 5.3 symptoms, breast cancer patients
reported 3.6 symptoms, colon cancer patients reported 3.7, and prostate cancer patients
reported an average of 4.3 symptoms (Given et al., 2001).

The effect of histology on symptom distress was examined by Cooley (2002). Cooley
found that histology of the lung cancer did influence symptom distress at baseline.
Adults with non-small cell lung cancer experienced more symptom distress than their
small cell lung cancer counterparts. Thus, site and stage of cancer significantly
influenced symptoms experienced by the cancer patient and was explored in this study.

Co-morbid Conditions

The presence of the co-morbid condition of respiratory disease was found to positively
correlate with a high level of symptom distress (chi square, 5.14; p = 0.023) in women
with lung cancer as reported by Sama (1993). Given et al. (2001) also noted that
“patients with three or more co-morbid conditions were more likely to report pain,
fatigue, or their combination, when compared with those reporting neither symptom” (p.

462). Refer to descriptions of both of these studies above.

40

Kurtz, Kurtz, Given & Given (1993) likewise noted a correlation between co-morbid
conditions and symptoms, as well as with physical functioning. Kurtz et al. used a
convenience sample of patients with various solid tumors as well as lymphoma (n=279)
to examine the trajectory of symptoms and loss of physical functioning over a six month
time span for patients initiating chemotherapy treatment. The most frequently appearing
symptoms in this study were fatigue, pain, insomnia and nausea. These authors found
that “. . . although co-morbidity was only modestly correlated with symptoms and loss of
function for the total sample, it was highly correlated with both symptoms and loss of
physical functioning for the younger patients (those younger than 60 years of age)”
(Kurtz et al., 2003, p. 275). Co-morbidity consisted of a count of the number of physical
co-morbid conditions as reported by the patient, range 0-13, examples include; arthritis,
hypertension, diabetes, etc. (Kurtz, Kurtz, Stommel, Given, & Given, 1997).

Bower, Ganz, Desmond, Rowland, Meyerowitz & Belin (2000) studied the effect of
co-morbid conditions in breast cancer survivors (n=1,927) as a component of a larger
study to identify and describe the effects of fatigue on this population. To be eligible for
this study women had to have stage II disease or lower within the past five years and
received adjuvant therapy. Subjects in the study had to be free of disease at the time of
the study to participate. The symptom assessment instrument used for this study was the
Breast Cancer Prevention Trial Symptom Checklist. A logistic regression was
completed to reveal that the co-morbid conditions of arthritis (r = 1.35, p = .03) and high
blood pressure (r = .99, p = .02) emerged as significant predictors of fatigue group

membership (Bower et al., 2000).

41

The presence of co-morbid conditions has been noted in the literature to impact
symptom severity and distress. This study examined co-morbid conditions as
antecedents. The next section of the adapted SEM model includes symptom production.
Antecedents contribute to the production of the symptom. This section of the model is
rooted in the work of Leventhal and Johnson (1983). Leventhal and Johnson define
symptom production as an objective, concrete representation of the disease. As noted in
Chapter 2, symptom production is not clearly articulated by Armstrong (2003).
However, the Armstrong SEM likely depicts symptom production as the actual physical
response or psychological reaction that takes into account the symptom antecedents and
leads to symptom perception. The definition of the term symptom used for this research
is from the work of Henly, Kallas, Klatt, & Swenson (2003): “Symptoms as experienced
by individuals are unpleasant sensations or perceived changes in normal functioning or
appearance” (p. 410). It is important to note that “symptoms are seldom solitary”
(Kroenke, 2001, p. 848).

Symptom Perception

Symptom perception is multidimensional in nature. McCorkle and Young (1978)
were among the first nursing authors to fully examine and develop a scale to research
symptom perception. The scale was called the Symptom Distress Scale (SDS) and has
been used in numerous studies; examples have been cited throughout this dissertation.
McCorkle and Young define symptom distress as “. . . the degree of discomfort reported
by the patient in relation to his/her perception of the symptoms being experienced” (1978,
p. 373). This symptom perception section of the literature review will focus on research

that has examined patient perceptions of the symptoms of fatigue, pain and insomnia.

42

Emphasis will be placed on the symptom of fatigue due to the significance of this
symptom within the research questions.
Fatigue

The multidimensional nature of the concept of fatigue makes this concept one of the
most interesting, yet difficult, areas of study. Fatigue is important to study for various
physical, psychological and economic reasons. It is also important to study due to its
impact on other chemotherapy symptoms, in particular pain and insomnia. Arronson,
Teel, Cassmeyer, Neuberger, Palikkathayil, & Pierce, stated that “Fatigue is a universal
symptom not only associated with most acute and chronic illnesses, but also with normal
healthy functioning and everyday life”(l999, p. 45). Empirical evidence supports,
patients with diagnoses such as: chronic obstructive pulmonary disease (COPD), human
immunodeficiency virus disease (HIV), Addison’s Disease, rheumatoid arthritis, multiple
sclerosis, and cancer all experience fatigue, and most experience distress as a result
(Aaronson, et al., 1999; Shaver, 1999; Breitbart, Rosenfeld, Kaim, & Funesti-Esch, 2001;
Cleare, 2003; Theander & Unosson, 2004). This section will focus on the concept of
fatigue fiom the perspective of oncology care.

The concept of fatigue is examined in the discipline of nursing via a bio-behavioral
approach. This bio-behavioral approach is reflective of a level of understanding that is
made up of the mental (psychological, cognitive, emotional, affective) and
physiochemical elements that drive human interaction with the environment (Shaver,
1999). Fatigue is composed of behavioral, cognitive, somatic and affective domains
(Stein, Jacobsen, Blanchard, & Thors, 2004). Congruent with the bioobehavioral

approach, fatigue is classified as a symptom (Shaver, 1999).

43

Fatigue is the most common symptom associated with cancer and the treatment of
cancer (Cella et al., 2001). Simon and Zittoun reported in their review of international,
empirical chemotherapy studies that 60% to 96% of chemotherapy patients experienced
fatigue (1999, p.244). Specific studies of the impact of fatigue as a symptom included
the work of Ashbury, Findlay, Reynolds, & McKerracher (1998) who noted that of their
913 Canadian cancer patients surveyed, 78% reported fatigue and 52% actively sought
out information to manage it. In addition, Vogelzang, Breitbart, Cella, Curt, Groopman,
Homing, et al. (1997) reporting on behalf of The Fatigue Coalition, found that 78% of
radiation and/or chemotherapy cancer patients experienced fatigue, and 32% of them
experienced it on a daily basis. Fatigue is one of the most common symptoms associated
with chemotherapy for both men and women. The prevalence and distress of fatigue for
patients, caregivers and practitioners associated with fatigue due to cancer and cancer
treatment necessitates continued study and conceptual understanding of this symptom.
The conceptual definition of fatigue will follow.

Conceptual Definition of Fatigue

As noted above, fatigue is a serious, multifaceted, problematic symptom for many,
especially cancer patients. A literature review of the concept provides numerous papers,
reviews, and research findings that all share a common description of fatigue; it is
complex, multi—dirnensional, and universally experienced. However, no universal
definition of fatigue was noted in the literature (Trendall, 2000). Most authors on this
subject agree that the concept of fatigue involves biologic, psychosocial, and behavioral

manifestations (Aaronson et al., 1999).

Fatigue may be viewed as a biological approach applied to a physical subsystem. An
example of this would be “fatigue” of a muscle. In this context fatigue refers to the lack
of ability of a muscle to contract after intense use (Shaver, 1999). Based on this
biological and physiological approach the phenomena of fatigue may be referenced as
functional organ failure (Berger, McCutcheon, Soust, Walker, & Wilkinson, 1991).
Typically, the physiological context of fatigue is of concern to disciplines outside of
nursing practice such as: biology, physiology, and medicine.

Disciplines outside of nursing studying fatigue examine mechanisms such as cytokine
and irnmunoregulatory factors, the depletion of hormones, the role of neutrotransmitters,
or dietary measures at the cellular level. An example of fatigue studied physiologically is
the pathophysiological perspective of the influence of cortisol and the hypothalamic-
pituitary-adrenal (HPA) axis. Cleare (2003) noted that low circulating cortisol levels
mediate symptoms associated with fatigue syndromes. Cleeland, Bennet, Dantzer,
Dougherty, Dun, Meyers et al. (2003) noted that cancer related symptoms such as fatigue
and pain may involve actions of proinflarnmatory cytokines such as interleukin (IL) -1,
tumor necrosis factor — (1 (INF - a), IFN- 0., and IL-2. The use of a physiological
definition alone is limited for nursing clinical practice and research due to the applied
nature of our practice and need for a bio-behavioral approach.

A psychological approach may be used to define the concept of fatigue. Lee, Hicks,
and Nino-Murcia (1991) defined the state of psychological fatigue as one of weariness
related to reduced motivation. Aaronson et al. (1999) defined fatigue within a
psychological framework as a response to internal or external demands exceeding

available resources. A qualitative study conducted in Britain compared cancer and

45

COPD patient responses to questions related to physical and psychological aspects of
fatigue. Two psychological themes emerged for both groups; impact of fatigue on
functionality and perceived control, and the emotional effect of the disease management
(Ream & Richardson, 1997). Berger and Walker-Nobel (2001) also support the
psychological approach to the concept in reference to mood and depression as
components of mental health that influence the perception of fatigue. Fatigue is a
complicated symptom consisting of physiological, psychological, and social components.
The symptom of fatigue is thought to provide a possible synergistic effect on other
chemotherapy symptoms. Fatigue as it presents with and possibly influences other
symptoms will be examined below.
Co-occurring Symptom Studies Including Fatigue

Although symptom distress has been reported in the literature for three decades, it
was Sama (1993) who published one of the first oncology research papers to include
fatigue as it was related to other co-occurring symptoms. Sama hypothesized that some
symptoms may “wax and wane during a course of treatment” (1993, p. 22). Sarna used
the SDS in women with lung cancer (n=69). It was noted by Sama that 61% reported
more than one distressing symptom, in fact, 23% reported that they experienced four or
more distressing symptoms concurrently. In addition, 41% experiencing fatigue also
experienced pain.

Sama was also one of the first authors to associate antecedents with the production of
multiple symptoms. Sama was interested in determining if multiple co-occurring
symptoms could be predicted by the location and stage of the cancer. It was noted that

antecedents such as treatment status, type of lung cancer, time since diagnosis, metastatic

46

 

disease, co-morbid conditions, education, marital status, and age did not effect ones
perception of symptom distress in this study. However, the design of this study as a
cross-sectional study with a small sample size of women at differing points in their
disease and/or treatment process was a limitation.

Hickok, Morrow, McDonald, and Bellg (1996) described multiple co-occurring
symptoms among male and female adult radiation therapy lung cancer patients (n=50) via
a medical record audit. Hickok et al. noted that concurrently 78% of these patients
reported fatigue, 80% reported pain, and 12 % had documented depression. Although
this work was descriptive and retrospective, it did provide an early attempt to examine
men and women with greater then two concurrent symptoms. The link between fatigue,
pain and depression among these patients, regardless of sex, provided a significant
replication of the findings of Sama and lead to the future consideration of these multiple
co-occurring symptoms a priori in study design.

Soon after, Gaston-Johansson et al. (1999) in their sample of 127 post adjuvant
chemotherapy women with stage II-IV breast cancer, designed a study to specifically
examine pain, fatigue and depression as co-occurring symptoms impacting the outcome
of health status. Gaston-Johansson et al. found that fatigue, pain, and depression were
present for 91%, 47% and 54% respectively. Fatigue, pain and depression were all
significantly correlated to each other as well as participant’s perception of their total
health status. The pain-fatigue correlation was .34. Pain and depression correlated at .25
and accounted for 63% of the variance in total health as measured by the SF-36.
Depression and fatigue correlated at .58 and accounted for 42% of the variance in the

outcome of health status. Gaston-Johansson et al. supported a priori selection of

47

symptoms noted in past studies, offered a larger number of subjects and was one of the
first to examine correlations among multiple co-occurring symptoms.

Beginning in the year 2000, Drs. Barbara and Charles Given began to publish work
that supported the conceptual application of antecedents, the symptoms experience and
outcomes/consequences. The Given work examined the impact of cancer and/or
chemotherapy symptoms on physical functioning. The Given et al. (2001) study
consisted of an inception cohort of 907 newly diagnosed solid tumor patients. The
Symptom Experience Scale (SES) instrument was used. Four different patient interviews
were completed at 6 to 8, 12 to 16, 26 to 30 and 52 weeks after the cancer diagnosis.

This study examined antecedents such as tumor type, stage of disease, age, education
and marital status via a retrospective medical record audit. A prospective examination of
the immediate, cumulative, and longer-term effects of chemotherapy on the dimensions
of the symptom experience (frequency and severity) of multiple symptoms was
completed at the time periods mentioned above. The outcome or consequence of
physical functioning was measured by the SF -36. Given et al. noted that symptoms did
not mediate between treatments and patient fimctioning. In this Given et al. (2001) study,
the number of symptoms produced a significant effect, independent of surgery, radiation,
and chemotherapy, for every additional symptom a decline of 2 units of physical
functioning was noted. The symptoms of pain and fatigue and the total number of
symptoms experienced by the subject were predictors of loss of function. Antecedent
variables such as age, sex, co-morbid conditions, and the time of observation also had

significant effects.

48

Kurtz, Kurtz, Stommel, Given and Given (2000) examined only the lung cancer
patients from the above stated Given et al. (2001) data set. This sample of 133 patients
was used to examine the effects of antecedents on the presence of multiple co-occurring
symptoms. Kurtz et al. demonstrated that fatigue was the most universally reported
symptom in this population regardless of age. Regardless of sex, five of the top six
reported symptoms were the same (fatigue, cough, night urination, difficulty breathing,
pain, and weakness), with some variation in order and fiequencies. The antecedents of
treatment type and stage of disease were responsible for the greatest variation in these six
symptoms.

Later studies by Given et al. (2001) would use this same data set to study co-
occurrence and patterns of change of pain and fatigue (n=841). Given et al. reported “. . .
if patterns are found, then future work can focus on elaborating how levels of intensity or
distress may moderate these patterns over time” (2001, p. 45 8). Findings of this study
revealed that the occurrence of pain and fatigue were predictors of other reported
symptoms among chemotherapy patients. Patients in this study who experienced pain
and fatigue concurrently reported an average of 6.3 additional symptoms.

Building on the work of Gaston-Johansson et al. (1999) as well as the Given et al.
(2001) work presented above, Dodd et al. (2001b) published one of the first studies based
on a conceptual framework supporting a symptom cluster as defined by the authors. The
Symptom Management Model as mentioned in Chapter 2 was the conceptual framework
used for this study. Pain, fatigue and sleep insufficiency were determined a priori to be
the symptom cluster of interest. The Kamofsky Performance Scale (KPS) measured the

effect of this cluster on the outcome of functional status. A hierarchical multiple

49

regression model was used to explain variance in the KPS at the end of the patient’s third
cycle of chemotherapy adjusting for KPS at baseline. The authors found that pain
explained only 10.7% of the change in KPS (p=0.002) and fatigue explained 7.3%
(p=0.011). Sleep insufficiency failed to produce statistically significant findings with
explanation of only 1% of the change overall (p=0.344). In addition, the Dodd et al.
(2001b) study did not produce strong correlations between the three symptoms (pain to
fatigue r = 0.22, pain to sleep insufficiency r = - 0.06, and fatigue to sleep insufficiency r
= -0.13). The authors acknowledged this work as the “beginning insights” into the effects
of symptom clusters.

An additional Given et al. (2002) study used a subset of the data presented in their
studies that began in 2000 to include patients reporting concurrent pain and fatigue at
baseline (n= 53 in the experimental arm, n=60 in the control arm) and a 10 contact
nursing intervention over 20 weeks. The Given et a1. (2002) publication was significant
due to the emphasis on the randomized clinical trial design and inclusion of an analysis of
the data related to the implementation of a nursing intervention for multiple symptoms.
Review of this Given et a1. (2002) study also supports the adaptation of the SEM to
include a timeline as well as an intervention component.

The Given et al. (2002) prospective, randomized clinical trial (n=1 13) was part of the
larger study described above that examined up to 14 different symptoms. The study
focused on subjects who experienced both pain and fatigue at baseline. Given et al.
(2002) used a cognitive-behavioral framework to guide the nursing intervention.
Oncology Certified Nurses (OCN®) were provided study specific training to assess and

intervene with identified symptoms in the experimental group of this study. The nursing

50

interventions consisted of patient problem solving techniques, the acquiring of
information, symptom self-care management, and emotional and social support for
patients. The effectiveness of supportive nursing interventions on symptom management,
physical role impact and social functioning were examined. Nurse sensitive outcomes for
this study included a reduction in the number of reported symptoms (measured by the
Symptom Experience Scale) as well as social and physical functioning (measured by the
SF -36).

The Given et al. (2002) longitudinal study contained baseline, 10 week, and 20 week
observations. Effects for the group were noted as related to symptom count at the 20-
week observation. At the 20-week observation, patients in the experimental group
reported 3.3 (SD = 2.6) symptoms on average vs. the control group reporting 4.4 (SD =
2.7) symptoms. Similar statistically significant reductions were noted in physical role
impact scores at 20 weeks. The experimental groups score mean was 50 vs. the control
group mean score of 31 (Given et al., 2002). Social functioning produced similar
findings with the experimental group having a mean score of 76 vs. the control group
mean score of 63 (Given et al., 2002). These findings demonstrated that patients
reporting pain and fatigue at baseline who received the nursing intervention reported
fewer symptoms and improved impact on their physical and social role functioning.
Work presented in the preliminary studies cited above shaped the development of the
ROI CA-30724 and ROI CA-79280 studies, both of which were used as the database for
this research.

In addition, a retrospective analysis completed by Cooley et al. (2003) studied

prevalence, level of distress, and how symptoms change over time in 117 lung cancer

51

patients. The significance of the addition of the timeline to the SEM is also supported by
this study. Cooley et al. (2003) combined three different data sets, all using a repeated
measures design with use of the SDS to complete secondary data analysis. It was
reported that fatigue, pain, and insomnia were among the most distressing symptoms at
baseline and remained distressing at the three and six-month points of the study for
patients receiving a combination of treatment modalities (n=51). Percentages of patients
reporting the presence of the symptoms at the three different time periods are as follow:
fatigue (65%, 61%, 43%), insomnia (45%, 27%, 22%), and pain (49%, 33%, 41%).
Decreasing symptom reports over time for fatigue and insomnia were present, however,
pain slightly increased by the 6-month point of the study. Cooley et al. (2003) imply that
retrospective, longitudinal, descriptive studies of this nature are a starting point for
studying multiple symptoms and have great relevance to the discipline of nursing and the
development of nurse sensitive outcomes.

Bower et al. (2000) examined fatigue in breast cancer survivors. They compared a
“fatigued” sample, all women scoring at or below 50 for the energy/fatigue subscale of
the RAND 36-item Health Survey (mean age = 54.87, SD=11.91, mean years since
diagnosis = 2.92, SD=1.18) to a “non-fatigued” sample of breast cancer survivors (mean
age = 56.17, SD=11.04, mean years since diagnosis = 2.89, SD=1.19). The women with
fatigue reported significantly more “severe” and “disabling” pain (p <.0001), greater
menopausal symptoms (p < .035), and significantly greater levels of depression as
measured by the CES-D (score > 16) (p < .0001). Logistic regression was completed on
the total sample (n=1,927) to reveal that depression was the strongest predictor of fatigue

group membership (odds ratio 1.13, p <.0001), followed by pain (odds ratio 0.97,

52

p <.0001) (Bower et al., 2000).

Two studies conducted with breast cancer patients and multiple co-occurring
symptoms used different study designs. The first, Wilrnoth, Coleman, Smith, and Davis
(2004) provided an example of a priori establishment of a symptom cluster of fatigue,
weight gain, and altered sexuality and examined it via a literature review of breast cancer
publications. The second, Bender et al. (2005) used an exploratory secondary analysis
approach to compare the prevalence of multiple symptoms associated with breast cancer.
A pooled analysis was conducted by Bender et al. examining baseline assessments flour 3
independent studies (combined n = 154 fiom the three studies). It is essential to note that
each of the 3 studies contained women with differing stages of disease. Study 1 (n=40)
were all early stage disease patients who had only completed surgery at the time of the
study. Study 2 (n=88) were stage I, II, or 111 patients who had undergone surgery,
adj uvant chemotherapy and were currently receiving hormonal therapy. Finally, study 3
(n=26) were all women with metastatic breast cancer. Bender et al. found that in all three
studies a common “symptom cluster” existed that was composed of fatigue, cognitive
impairment and mood problems.

We note fiom the above examples that cancer/chemotherapy patients exhibit both
single and multiple, co-occurring symptoms at diagnosis, throughout treatment and even
upon completion of therapy (Cleeland et al., 2000; Given et al., 2001). Fatigue is one of
the most prevalent symptoms and is often associated with pain and insomnia (Cooley et
al., 2003; Dodd et al., 2001b; Given et al., 2001; & Sarna, 1993). Although the

significance of these co-occurring symptoms has been established, randomized clinical

53

trials involving interventions have been less plentiful. The following section will focus
on the intervention component of the adapted SEM.
Interventions for Fatigue, Pain and Insomnia

Although the actual intervention for this study will be described in detail in Chapter 4.
This section will review conceptual issues related to intervention studies that support the
symptoms of fatigue, pain and insomnia. Recall that the interventions discussed here will
target the symptom(s) directly. The adapted SEM draws from the SMM to include the
intervention “specifications” of : what, when, where, why, how much, to whom, and
how; to facilitate the literature review as well as study replication (Dodd et al., 2001a).

The Given et al. (2002) work as cited above (longitudinal, RCT with three observation
periods), was one of the most significant intervention based nursing research studies
conducted to date related to the symptoms of fatigue and pain in chemotherapy patients.
Given et al. (2002) used a subset of the data presented above to include patients reporting
concurrent pain and fatigue at baseline (n= 53 in the experimental arm, n=60 in the
control arm) and a 10 contact nursing intervention over 20 weeks.

A description of the nursing intervention will follow. The intervention was comprised
of evidence-based intervention strategies that were delivered to the nurse intervention
group. Each intervention nurse had the same stepped cancer-nursing intervention
software loaded onto a laptop computer. The Given et al. (2002) study software housed
problem-specific, evidence-based intervention strategies that the nurse and patient could
mutually elect for the patient to implement on his or her own behalf to move the problem

toward resolution.

54

The nurse intervention was targeted to assess and intervene with previously identified
symptoms. At each visit, the nurse assessed all symptoms. Any symptom that reached a
severity level of 5 or higher on a lO-point scale or a reported impact of 3 or higher on a 5-
point scale with respect to the impact on patients’ quality of life were posted to the
problem list. These patient assessments were part of the nursing intervention. Patient
responses were selected fi'om the options available in the intervention program and
related to the impact that a particular symptom had on quality-of-life indicators (e.g.,
sleep, mobility, appetite). Patients were asked to rate their responses fi'om 0 (no impact)
to 10 (a great impact). For example, pain would be assessed according to its onset,
duration, maximum severity, impact on daily activities, and other associated problems,
such as fatigue or insomnia. All symptoms and functional health indicators that reached
a threshold (i.e., either the current intensity self-rating of 5 or higher or quality-of-life
indicators self-rating 3 or higher) were posted to the plan of care.

Once a symptom was posted the nurse and patient addressed it until the symptom was
controlled or the intervention ended. At each intervention encounter, the nurse would ask
the patient to evaluate the efficacy of the intervention strategy and the status of the
problem resolution. Intervention strategies were modified, changed, or deleted
depending on the result. Using a computer-assisted protocol, the intervention nurse was
able to document in real time the interventions for each patient problem at each encounter
in which the nurse and patient focused on the problem. Intervention fidelity was
monitored by the nurse coordinator reviewing every encounter for each patient, as well as
weekly meetings between the PPS and nurse interveners (Santacroce, Maccarelli, & Grey,

2004).

55

At baseline the experimental group in the Given et al. (2000) study presented with 7.3
(SD 2.8) symptoms and the control group 6.8 (SD 2.1). By the second observation (10
weeks into chemotherapy) the experimental group reported an average of 5.2 (SD 2.9)
symptoms vs. 6.1 (SD 3.2). Upon completion of the nursing intervention the
experimental group reported an average of 3.3 (SD 2.6) symptoms vs. 4.4 (SD 2.7)
symptoms reported by the control group (p. 952).

The above data base was also used by this investigator to examine if symptom severity
can be predicted by chemotherapy regimen 20 weeks following baseline assessment, after
controlling for age, co-morbid conditions and experimental vs. control group membership
in breast and lung cancer chemotherapy patients. The sample consisted of 71 females
with breast cancer, 21 females with lung cancer and 23 men with lung cancer. A binary
logistic regression was completed to support that co-morbid conditions and group
assignment (experimental vs. control group membership) reliably predicted symptom
severity. Chemotherapy category did not predict symptom severity at 20 weeks following
baseline assessment. Symptoms related to toxicity profiles (i.e., nausea, mucositis, fever,
diarrhea) of chemotherapy agents were not as prevalent as symptoms linked to the
chemotherapy experience (i.e., fatigue, pain, insomnia). These data supported the
significance of the co-morbid conditions, as mentioned above. However, the most
significant contribution was the validation of significance of the impact of the nursing
intervention on the outcome of decreasing symptom severity in the Given et al. (2002)
study (Keehne-Miron, Given, Given, vonEye, & Doorenbos, 2005).

Yates, Aranda, Hargraves, Mirolo, Clavarino, McLachlan, et al. (2005) likewise

implemented a nursing intervention in a RCT for cancer patients. Yates et al. studied

56

stage I or 11 breast cancer patients and fatigue. Subjects were recruited and randomized
prior to their first chemotherapy treatment (n=109) to the control group (standard of care)
or the intervention arm. The intervention was a psycho-educational intervention aimed at
improving one’s self-care skills and behaviors to minimize fatigue. Trained nurses used
protocols to guide the structure, process and content of the intervention. All experimental
group members also received a fatigue management booklet published by the Oncology
Nursing Society. A baseline fatigue assessment was conducted on all patients with their
first chemotherapy visit. The first intervention session took place for the experimental
group in a face to face fashion at the clinic upon the subject’s second course of
chemotherapy. Two booster sessions were provided to the experimental group members
one week post the second cycle of chemotherapy and the second booster took place two
weeks post this same second cycle of chemotherapy.

It is important to note that no group differences existed at baseline between the groups.
An immediate effect for the nursing intervention was noted upon the first follow-up
assessment; this took place after the second booster session, prior to cycle three of
chemotherapy. Estimated marginal change scores for severity scores for the experimental
group from time one (pre chemotherapy) to time two (post cycle two chemotherapy and
post intervention for the experimental group) was 1.0, for the control group, the mean
change score was 2.6 (p = .01). This change score difference supported the nursing
intervention. However, this finding was not supported in the remaining follow-up
evaluations. Estimated marginal change scores for severity scores for the experimental
group from time one (pre chemotherapy) to time three (post cycle three chemotherapy)

was 1.6, for the control group the mean change score was 1.9 (p = .59). Time one to time

57

four (post cycle four of chemotherapy) was likewise non significant with the
experimental group posting change scores of .6 and control group members with scores
of 1.3 (p = .26). Thus, as stated by the authors, “no significant differences were
identified for any pre-or post-test change scores for confidence with managing fatigue,
cancer self-efficacy, anxiety, depression, or quality of life” (p. 6027). The effect of the
nursing intervention did not withstand the time period of the study. Nursing interventions
that were successful for reducing symptom severity between cycle one and cycle two
were not successful at maintaining the reduction of symptom severity by cycle three. The
importance of assessment, intervention and measurement of symptoms over time for
chemotherapy patients is supported in this study. Variables influenced by time include
the multiple cycles of chemotherapy and their possible influence on firture cycles,
chemotherapy dosing schedules (every 14 days vs. every 21 days), as well as timing of
the measurements.

Likewise, Anderson, Lohen, Mendoza, Guo, Harle, and Cleeland, (2006) conducted a
RCT to evaluate the efficacy of 3 brief cogrritivebbehavioral techniques (relaxation,
distraction, and positive mood interventions) on cancer related pain. Patients were
recruited from an outpatient clinic (n=57), patients had a solid or hematological
malignancy and experienced daily pain reported between 4 and 7 on a 0-10 scale for
intensity as their “usual” level of pain. All patients were using opioid medications and
have reported a positive response (decrease in pain intensity by 1 point after the start of
the opioid) to that therapy.

The interventions for the Anderson et al. (2006) study are described below. Patients

in the relaxation group received a 20 minute audiotape containing positive mood

58

statements and positive imagery suggestions. The relaxation group received a 20 minute
audiotape that contained progressive muscle-relaxation instruction. The distraction group
selected an audiotape on various topics of their choice, i.e., history, geography, foreign
language. The control group received standard of care.

The Anderson et al. (2006) study used the MD Anderson Symptom Inventory
(MDASI) to measure cancer related symptoms and the Brief Pain Inventory (BPI) was
used to measure pain at four different assessment points throughout the 9 week study
period. The mean age of the sample was 52 years (range 30-80) with the majority (79%)
of the participants being women, 58% of the sample had metastatic disease. The
relaxation group reported a mean pain reduction level of .90 (CI: 0.16-1.65; p = .023)
from time one to time two. The distraction group also reported a mean pain reduction
level of 1.16 (CI: 0.47-1.85; p = .004) fiom time one to time two. The positive mood
intervention actually caused a slight increase in pain intensity of 0.08 (CI: -1.53 — 0.138;
p = .91) after the initial session. However, at all time periods beyond time two, no
significant differences in pain intensity were noted between any of the groups. Thus,
significant reports of pain reduction were only noted upon the first assessment
immediately following either the relaxation and distraction interventions, over time the
intervention was not effective.

This finding in the Anderson et al. (2006) study offers additional support for the
diminished endurance of specific interventions over time for the cancer patient. It is
important to note in this study that the control group members were more likely to be
receiving chemotherapy (p = .04) as well as a non-significant difference in baseline

severe pain levels between the intervention groups (81%) with the control group (55%) (p

59

= .09). What is demonstrated in the intervention studies described above is that when
studying symptoms in cancer patients it is vitally important to monitor the symptoms as
well as the effectiveness of the interventions over time. As noted above, an intervention
can be considered effective and significant upon immediate implementation however,
over time this effect may deteriorate. Thus, it is crucial to examine the effects of
interventions within the context of the disease and treatment trajectory, over time.

This effect of the time element leads us to our final section of this chapter. According
to the adapted SEM following the intervention is the symptom expression component of
the model. The symptom expression is the physical display of the symptom(s) post
intervention. The symptom express component is linked with a time arrow back to the
symptom perception component of the adapted SEM, refer to Figure 2. We note
following the intervention that the symptom(s) may stay the same, change in one or any
of the dimensions, or no longer be perceived by the patient. The final concept of time
will be reviewed below.

Time

The research presented above supports the appropriateness of measurements of
symptoms over time due to the chronic nature of cancer, treatment approaches that are
delivered over time, and the knowledge that symptoms change over time. Many cancer
patients may experience symptoms for some time prior to diagnosis. The actual
diagnosis of cancer can often take a significant amount of time to obtain with numerous
visits to various specialists, miss diagnosis, and the numerous tests and procedures
required to arrive at an accurate diagnosis. Often for patients symptoms continue or

worsen during this time. Likewise, the treatments for cancer are administered over time.

60

Even surgery, although it may be considered a one time event, requires an extensive
amount of time to recover from. Recovery of the blood cell values after surgery should
be back to normal as well as incisions should be healing prior to the initiation of the next
treatment such as chemotherapy or radiation therapy. Again, the variable of time needs
to be considered as well as possible new or enhanced dimensions of symptoms due to the
procedure and/or extension of time.

The concepts of dose and time are essential for optimum chemotherapy delivery.
Optimum doses of chemotherapy due to body surface area and past clinical trials are
intended to be delivered within specific time periods. These time periods are typically
depicted as cycles. Standard time periods or cycles of chemotherapy are every week,
every 14 days, every 21 days or every 28 days. Symptoms will be influenced by the
selection of agents, the route of administration and the timing of the administration. It is
also essential to recall that the symptoms associated with chemotherapy typically remain
to various degrees upon completion of the treatment. Fatigue is one such symptom that
has been found to persist over time following the completion of treatment for cancer.

Curt, Breitbart, Cella, Groopman, Homing, Itri, et al. (2000) offer support for this
concept in their report for the Fatigue Coalition. The coalition conducted a telephone
interview of 379 cancer patients having had a history of chemotherapy. Recruitment was
from 6,125 households in the United States, the median patient age was 62 with the
majority (79%) being women. Seventy six percent of this sample reported experiencing
fatigue “at least a few days each mon ” during their most recent chemotherapy; 30%
experienced the symptom on a daily basis. Women were more likely to experience daily

fatigue vs. their male counterparts (33% vs. 22%, p < .05). Sixty two percent of patients

61

reporting fatigue during chemotherapy reported it as “lasting longer than four days” (p.
355). Curt et al. (2000) noted that of the patients who experience fatigue and pain, and
either nausea or depression (a total of three symptoms) (n=198) the majority (58%)
reported that fatigue was the symptom that had lasted the longest. This descriptive study
emphasizes the importance of monitoring and accurately measuring symptoms, especially
fatigue in combination with other co-occurring symptoms, over time.

Summary

Various studies cited in the sections above support the importance of the concept of
time. The Curt et al. (2000) study exemplifies the “clinical context” of the symptom as
described by Barsevick et al. (2006) in Chapter One. The Anderson et al. (2006); Given
et al. (2002); and Yates et al. (2005) studies support the impact of interventions over
time. Thus, the intent of this literature review was to not only examine the symptoms of
fatigue, pain and insomnia, with an emphasis on studies that examine them as multiple,
co-occurring symptoms, however, to also support the adaptation of the SEM as described
in Chapter 2.

The purpose of this research is to examine the dimensions of frequency, intensity, and
distress of the co-occurring symptoms of fatigue, pain and insomnia as they occur at two
different data collection points in two randomized clinical trials that include a cognitive
behavioral intervention. The literature review above provided an opportunity to discover
some inconsistent findings in past work as well as provide conceptual and historical
guidance leading to the development of the research questions for this study. These

studies have also provided direction for the development of the research design and

62

methods that will be used for this study. In Chapter 4 a through description of the

research design and methods will be provided.

63

Chapter Four
Design and Methods
This chapter will present and describe the sample and setting, the intervention,
experimental variables, instruments, data analysis plan, and human subject protection
plan for this research. The purpose of this research is to exarrrine the dimensions of
frequency, intensity, and distress of the co-occurring symptoms of fatigue, pain and
insomnia as they occur at two different data collection points in two randomized
cognitive behavioral intervention clinical trials. The two trials are both NCI supported
longitudinal panel studies of newly diagnosed adult cancer patients (ROI CA-79280
{study number one} & ROl CA-30724 {study number two}).
Research Questions
Overall, this study will address the question: At baseline observation and at 10 weeks
is there an association between the dimensions of fi'equency, intensity, and distress of
fatigue and frequency, intensity and distress of pain and/or insomnia for adults receiving
chemotherapy? The effect of fatigue management will be evaluated as well as each
question will be examined for a possible influence on the association by the variables of
age, site and stage of cancer, sex, and co-morbid conditions. The specific research
questions will follow.
Research Question I
At baseline observation, prior to entry into a clinical trial, and at 10 weeks is there an
association between the dimension of frequency of fatigue and frequency of pain, as well
as fiequency of fatigue and frequency of insomnia for adults receiving chemotherapy? Is

this association influenced by; age, site and stage of cancer, sex, and co-morbid

64

conditions? The dimension of frequency is the number of days out of the past seven that
a patient reports the symptom as present. This will be a continuous variable with a
possible range of 0-7.
Research Question 2
At baseline observation, prior to entry into a clinical trial, and at 10 weeks is there an
association between the dimension of intensity of fatigue and intensity of pain, as well as
intensity of fatigue and intensity of insomnia for adults receiving chemotherapy? Is this
association influenced by; age, site and stage of cancer, sex, and co-morbid conditions?
The intensity dimension is measured on a rating scale of 0 meaning not present to 10
meaning the intensity is the worst possible. This will be a continuous variable with a
possible range of 0-10.
Research Question 3
At baseline observation, prior to entry into a clinical trial, and at 10 weeks is there an
association between the dimension of distress related to fatigue and distress related to
pain, as well as distress related to fatigue and distress related to insomnia for adults
receiving chemotherapy? Is this association influenced by; age, site and stage of cancer,
sex, and co-morbid conditions? The dimension of distress will be measured on a rating
scale with 0 meaning the symptom does not bother the patient to 10 meaning the
symptom causes the worst bother the patient can imagine. This will be a continuous
variable with a possible range of 0-10.
Research Question 4
Using categories of response to fatigue management (none / mild, non-response,

partial response and full response) established in the literature, determine how changes in

65

the dimensions of frequency, intensity, and distress of pain and insomnia are predicted by
categories of response to the management of fatigue between baseline and 10 weeks. Is
this association influenced by; age, site and stage of cancer, sex, and co-morbid
conditions?

Question 4 will examine if a change in fiequency of fatigue is capable of predicting a
change in the dimensions of frequency, intensity and/or distress for the associated
symptoms of pain and insomnia. For this question, fatigue intensity levels will be
converted from continuous to categorical variables to facilitate the comparison with
fatigue response categories (Mendoza, Wang, Cleeland, Morrissey, Johnson, Wendt, et
al., 1999). The fatigue response categories will be 0 = none 1 = mild fatigue, 2-4
moderate fatigue, 5-10 severe fatigue.

Response categories for the management of fatigue will be determined by comparing
the baseline fatigue severity category with the 10 week fatigue severity category.
Response categories for the management of fatigue will be as follows: None/mild =
patient stays within the same none or mild category at the baseline measurement as well
as the 10 week measurement. A patient can move fi'om none to mild or from mild to
none in this category. Non-responder = a patient who stays within the same category
(excluding none/mild) from the baseline measure of fatigue to the 10 week measurement,
(i.e., severe at baseline and severe at 10 weeks). Partial responder = A patient who goes
from severe to the moderate level of fatigue between baseline and 10 weeks. Full
responder = A patient who goes to the none or mild category of fatigue fiom moderate or
severe between baseline and 10 weeks (Given et al., in press; Miaskowski, et al., in press;

Paul, Zelman, Smith, & Miaskowski, 2005).

66

Sample and Settings

The two randomized cognitive behavioral intervention clinical trials database from
which will be used for this research started accruing patients in 2004. Subject accrual
was completed in June of 2006 at which time a total of 671 cancer patients participated in
both studies. All patients were over the age of 21, undergoing chemotherapy for a solid
tumor or lymphoma and being treated at one of seven community or academic cancer
centers throughout the Midwest and East coast. Study sites included: Yale University in
Connecticut, Indiana University in Indiana, and Michigan State University/Breslin
Cancer Center/Great Lakes Cancer Institute, Saint Joseph Mercy Oakland Hospital, Saint
Mary’s Health Care, William Beaumont Hospital, and Holy Cross Cancer Center all in
Michigan. Subjects were accrued to one of four different arms (groups) that made up the
two studies, study one (n= 235) and study two (n=437). See Appendix A for the Study
Schema. Potential subjects were recruited at each center by trained recruiters.

Once patients were enrolled and consented, in either study, they completed twice-
weekly automated telephone calls for up to six weeks assessing chemotherapy symptoms.
In study one, patients who had advanced disease and reported a 2 out of 10 for both pain
and fatigue or a 3 for either pain or fatigue entered the trial, received a baseline interview,
and were randomized to either the NASM group to receive an eight week, six contact
stepped-approach cognitive behavioral intervention implemented by an oncology nurse as
well as a symptom management toolkit, or to the group involving a similar number of
encounters with a trained non-nurse intervener (non nurse symptom management
[NNSMD who provided a self-management intervention based on a symptom

management toolkit.

67

In study two, patients who reported a 2 or higher for a severity rating for one or more
symptoms entered the trial, received a baseline interview, and were randomized to one of
two arms (groups) of the study. Patients in group one received six weekly calls by an
automated system (ATSM) with reference to a symptom toolkit based on the severity of
the symptoms reported. Patients in group two received an eight-week, six contact
stepped-approach cognitive behavioral intervention implemented by an oncology nurse as
well as the toolkit (N ASM). Patients never reaching a 2 or higher on any symptom at
screening did not enter the trial, and were sent a letter thanking them for participation.

Both randomization procedures balanced the groups with respect to site of cancer and
recruitment location. All patients fi'om both studies, all groups, will be used in this
research. The rationale for the various academic and community sites involved in this
study is to provide a mix of academic and community cancer center patients, rural and
urban participants, and to promote ethnic and cultural diversity based on geography.

The Intervention

Study number one contained two arms, each with an intervention. Upon
randomization patients and caregivers could be placed in the NNSM group or the NASM
group. The NNSM group intervention consisted of the patients and caregivers receiving
a symptom management toolkit (book that provided detailed instruction and reference for
management of common chemotherapy/cancer symptoms) and six calls over an eight
week time period by a research assistant who would listen to concerns and advise
participants to reference the appropriate section in the toolkit.

Study number two contained an ATSM group as one of the two groups of this study.

This group likewise received the toolkit, however, instead of a research assistant making

68

the call, as in the example cited above; an automated telephone system contacted the
patient for six calls over eight weeks at times per the patient’s pre-determined preference.
The automated telephone system required patients to respond to symptom assessment
questions by ranking the symptoms 0-10 on their telephone keypad, based on responses
the automated system would direct them to appropriate sections of the toolkit and/or refer
the patient to their oncologist if the symptom met a threshold pre-deterrnined by their
oncologist.

Both studies contained NASM groups. Participants in these groups also received the
toolkit and six calls over eight weeks by an oncology nurse. The nursing intervention
was comprised of evidence-based intervention strategies that were delivered to the nurse
intervention group. Each intervention nurse had the same stepped cancer-nursing
intervention software loaded onto a laptop computer. This software housed problem-
specific, evidence-based intervention strategies that the nurse and patient could mutually
elect for the patient to implement on his or her own behalf to move the problem toward
resolution.

The intervention for all of the groups described above was targeted to assess and
intervene with previously identified symptoms. At each visit, the nurse, research
assistant (non-nurse), or automated telephone system assessed all symptoms. In addition
to severity rated on a scale fiom 0 = not present to 10 meaning the worst severity
possible, patients were asked to rate their responses from 0 (no impact) to 10 (great
impact) on quality of life indicators (e.g., sleep, mobility, appetite). Any symptom that
reached a level of 4 or higher in severity or impact on quality of life indicators was

posted to the problem list. In the NASM groups, once a symptom was posted the nurse

69

and patient addressed it until the symptom was controlled or the intervention ended. In
the NNSM and ATSM groups, the patients were directed to the toolkit as described
above.

At the next intervention encounter, the patients were asked to evaluate the efficacy of
the intervention strategy and the status of the problem resolution. In the NASM groups
the intervention strategies were modified, changed, or deleted depending on the result.
Using a computer-assisted protocol, the intervention nurse was able to document in real
time the interventions for each patient problem at each encounter in which the nurse and
patient focused on the problem. Intervention fidelity was monitored by the nurse
coordinator reviewing every encounter for each patient, as well as weekly meetings
between the FPS and nurse interveners (Santacroce, 2004).

Patients in both studies were very similar. The distributions of age, race/ethnicity and
level of education did not differ between the two studies. Study number one had a higher
percent of lung cancer patients, lower percent of breast cancer patients, and higher
percentage of late stage cancer patients. Because of disease severity, patients in study
number one had higher severity symptoms at intake, including pain and fatigue.
Preliminary analysis revealed that both studies demonstrated decreased summed
symptom severity at the ten week analysis with no differences by group in each study.
The data from both studies will be combined. This study will not be looking for a group
effect. Demographic data will be presented in Chapter 5. All analyses will adjust for
group and study membership. The main interest of this study is the frequency, intensity
and distress of fatigue, pain and insomnia, the effect of a fatigue intervention, and the

testing of other covariates. Refer to Appendix A for the study schema.

70

Due to the extensiveness of the data and the large number of subjects, unique aspects
of this study include: the impact of variables such as age, site and stage of disease, sex,
and co-morbid conditions, the ability to examine symptoms for a change in pattern of
frequency, intensity, and distress over time, and, the ability to assess the relationships
among symptoms.

Experimental Variables

The outcome variables for this study arise from the adapted SEM and include the
frequency, intensity, and distress of the symptoms of fatigue, pain, and insomnia, refer to
Figure 2. All results will be adjusted for study membership and group. Additional
covariates that will be examined include: age, site and stage of disease, sex, and co-
morbid conditions. The randomized clinical trial design was closely monitored
throughout the original studies resulting in a minimal amount of missing values.

Measures

Demographic data such as site of cancer, stage of disease, and treatment protocol were
obtained from a medical record audit after patients signed consent forms and completed
the study.

The Symptom Experience Scale (SES). The SES was deve10ped by Drs. Barbara and
Charles Given for the Family Home Care for Cancer: A Community -Based Model
study (1993), it is based on the Symptom Distress Scale (SDS) by McCorkle & Young
(1978). The SES is an interview guide that prompts trained interviewers to ask patients if
they had experienced symptoms commonly associated with cancer or their cancer
treatment in the previous two weeks. Frequency was recorded as the number of days out

of the past seven a particular symptom was experienced, thus, used 0-7 scale. Intensity

71

was recorded as the severity level and was ranked on a scale from 0 = not present to 10 =
the worst it can be. Distress referred to the extent to which the symptom caused an
“interference with life” and also used a 0-10 scale. Cronbach’s alpha has been
established for this instrument as .90 in previous studies (Wyatt, Friedman, Given, Given,
& Beckrow, 1999). The SES has consistently measured symptoms within the theoretical
definition by Armstrong (2003), thus, establishing construct validity. Content validity
has been examined and established by reviewing the literature and surveying
chemotherapy symptom assessment/management experts as to the degree to which the
items cover the domain of chemotherapy/cancer symptoms.
Data Collection Schedule and Procedures

Nurse recruiters were trained from each of the participating sites to identify eligible
patients, explain the study to potential subjects, and obtain signed informed consents.
Upon receipt of signed consent, all patients were monitored and screened into the study
via the automatic telephone system as described previously. All study participants,
regardless of random assignment completed the telephone interviews. Trained personnel
who were not nurses conducted all of the telephone interviews. If a patient was
randomized to a NASM group the nurse was contacted by the project coordinator. The
intervention nurse then contacted the patient via telephone to introduce herself, review
the consent, and the patients’ role in the study and schedule a time to conduct the baseline
intervention session. All intervention contacts were conducted via the telephone.

Telephone interviews included the SES at baseline, 10 weeks and 16 weeks and were
conducted from a central location. Each interviewer was trained and followed an explicit

quality assurance protocol. The non-nurse intervention group received telephone calls of

72

similar frequency that provided an automated version of the SES. All patients received
conventional care as prescribed by their oncology team. Medical record audits were
completed at the end of the study.

Data Analysis and Interpretation .

The outcome measures for this study included symptom fiequency, intensity and
distress. The purpose of this research was to examine the dimensions of fiequency,
intensity, and distress of the co-occurring symptoms of fatigue, pain and insomnia as they
occur at two different data collection points in two randomized cognitive behavioral
intervention clinical trials. The following statistical approaches were applied to the
research questions.

Descriptive analyses of the outcome measures (symptom fiequency, intensity and
distress of each symptom) were obtained by group, and by age, site, stage of cancer, sex,
and co-morbid conditions at baseline and ten weeks. Distributions of the outcome
measures were evaluated. Regression analysis was conducted at baseline and 10 weeks
for each of the following research questions: Research Question 1: At baseline
observation, prior to entry into a clinical trial, and at 10 weeks is there an association
between the dimension of fiequency of fatigue and fiequency of pain, as well as
fiequency of fatigue and frequency of insomnia for adults receiving chemotherapy? Is
this association influenced by; age, site and stage of cancer, sex, and co-morbid
conditions? Research Question 2: At baseline observation, prior to entry into a clinical
trial, and at 10 weeks is there an association between the dimension of intensity of fatigue

and intensity of pain, as well as intensity of fatigue and intensity of insomnia for adults

receiving chemotherapy? Is this association influenced by; age, site and stage of cancer,

73

sex, and co-morbid conditions? Research Question 3: At baseline observation, prior to
entry into a clinical trial, and at 10 weeks is there an association between the dimension
of distress related to fatigue and distress related to pain, as well as distress related to
fatigue and distress related to insomnia for adults receiving chemotherapy? Is this
association influenced by; age, site and stage of cancer, sex, and co-morbid conditions?
Analysis conducted for baseline measurements were performed using the following
statistical models: 1) Frequency of pain at baseline related to frequency of fatigue at
baseline and 2) Frequency of pain at baseline related to fiequency of fatigue at baseline
add covariates. The covariates for all of the following models included age, site and
stage of cancer, sex, and co-morbid conditions. Models 1) and 2) were also fit with
frequency of insomnia as an outcome. Similar analyses were preformed for the intensity
and distress dimensions:
Intensity of pain at baseline related to intensity of fatigue at baseline;
Intensity of pain at baseline related to intensity of fatigue at baseline add covariates;
Intensity of insomnia at baseline related to intensity of fatigue at baseline;
Intensity of insomnia at baseline related to intensity of fatigue at baseline add
covariates;
Distress of pain at baseline related to distress of fatigue at baseline;
Distress of pain at baseline related to distress of fatigue at baseline add covariates;
Distress of insomnia at baseline related to distress of fatigue at baseline;
Distress of insomnia at baseline related to distress of fatigue at baseline add

covariates.

74

The essential parameters tested in these models were regression coefficients for the
frequency, intensity and distress of fatigue at baseline.

Analysis conducted at 10 weeks was performed using the same statistical models as at
baseline with baseline measures replaced by 10 week measures. Covariates in the 10
week analyses were the same as in the baseline analyses with study membership and
group added to adjust for their effect on the outcomes. The essential parameters tested in
these models were regression coefficients for the frequency, intensity and distress of
fatigue at 10 weeks.

Regression analysis was also conducted to answer research question number four:
Research Question 4: Using categories of response to fatigue management (none / mild,
non-response, partial response and full response) established in the literature, determine
how changes in the dimensions of frequency, intensity, and distress of pain and insomnia
are predicted by categories of response to the management of fatigue between baseline
and 10 weeks? Is this association influenced by; age, site and stage of cancer, sex, and
co-morbid conditions? Analysis were performed by fitting the following statistical
models: 1) Frequency of pain at 10 weeks related to fiequency of pain at baseline add
category of response to fatigue management and 2) Frequency of pain at 10 weeks
related to frequency of pain at baseline add category of response to fatigue management
add covariates. The covariates include age, site and stage of cancer, sex, co-morbid
conditions, study membership and group. The essential parameters tested in these models
are the regression coefficients for the variables representing different categories of
response to fatigue management. Following the analysis of frequency of pain, statistical

models 1) and 2) were fit for the intensity and distress of pain, and for the frequency,

75

intensity and distress of insomnia. Note that the above models included group and study
membership; therefore, the results were adjusted for these variables. Analyses were
performed using SPSS ® Version 13.0 Graduate Pack for Windows statistical software
package.
Power

Following completion of the data analysis post hoc power analysis was performed for
each research question. From the results of fitting each statistical model, the observed
effect sizes for the essential parameters tested in the models were determined, and based
on the magnitude of the effect size and the sample size, the statistical power to detect
each effect size was calculated. The final power analysis is displayed in a Table 14; see
Chapter 5, page 112.

Protection of Human Participants

This study (approval # 06-130, Michigan State University Institutional Review Board
[IRB]) as well as the two studies that this data analysis is based upon have successfully
completed full IRB review at Michigan State University (MSU). The two studies (ROI
CA-79280 {study number one} & ROl CA-30724 {study number two}) providing the
data for this research received full review at each of the 7 clinical sites recruiting for the
studies. Informed consent written materials were presented to prospective patients by
trained nurse recruiters at each clinical site. Upon receipt of signed written consent
patients were enrolled in the automated telephone screening system. Upon reaching
threshold for study participation patients were reminded of the study procedures and their

informed consent.

76

If patients wished to continue with the study at this time they were given a baseline
interview. Following the baseline interview patients were randomized to the study
groups by the site study coordinator. At any time a patient was allowed to withdraw from
the studies. All patients were given a study code, no names or other identifying
characteristics were used in the analytical datasets. The list of codes to identify patients
is kept in a locked file drawer with access to the key only available to the study PI’s.

Data Security

All patient data was recorded and stored via a web based data collection program.
This program was backed up on a nightly basis and stored on a secure server in a fire
retardant room located outside MSU. All members of the research team have completed
mandatory MSU IRB training related to patient consent, confidentiality, quality
assurance, and data safety and security on a yearly basis.

Recruiter Training

Nurses from the clinical trials offices of participating sites were hired and trained to
implement the recruitment protocol. The PI’s for the two studies Dr. Barbara Given and
Dr. Charles Given interviewed all recruiter candidates and made the final selection for
hire of one recruiter per site. All of the nurse recruiters were experienced in clinical trial
recruitment, most were oncology nurses familiar with the cancer centers practitioners and
their patients.

Prior to initiation of recruitment, all nurse recruiters were extensively trained in
interactive sessions the included discussion of the objectives and schema of the studies
(ROI CA-79280 [study one] and R01 CA-30724 [study two}), inclusion and exclusion

criteria, policies and procedures for identifying and recruiting participants,

77

confidentiality, MSU IRB policies, their local institution’s IRB policies, and completion
of the study forms, web-based tracking system, and symptom severity screening system.
Secure computer passwords were provided to each nurse recruiter for entering
sociodemographic information into the web-based tracking system. A recruiter manual
was used for training as well as provided for future reference to each nurse recruiter. The
recruiter manual contained information on the research team and contact information,
overall study goals and objectives, inclusion and exclusion criteria, screening and
recruitment procedures, and a recruitment script.
The Recruitment Process

The nurse recruiters evaluated patients based on the inclusion criteria. Inclusion
criteria consisted of: being over the age of 21 , diagnosed with a solid tumor cancer or
non-Hodgkins lymphoma, currently receiving chemotherapy, able to speak and read
English, and had a touchtone telephone. Nurse recruiters presented the study and
informed consent verbally and provided a study packet that included a description of the
studies, contact information, and the informed consent forms to potential subjects and
their caregivers. Nurse recruiters completed screening/enrollment forms for each patient
approached. Every screening/enrollment form was faxed to the PPS office at MSU.
Upon obtaining a signed informed consent, sociodemographic information was entered
into the web-based system. Patients were screened at this time for symptom severity.
Recall that patients needed to have a 2 or higher level of severity (range 0-10) on any of
the 13 symptoms assessed via the MD. Anderson Symptom Inventory (MDASI) to enter
into study one or study two or higher for both pain and fatigue or a 3 or higher for either

pain or fatigue on the MDASI to enter into study two (Cleeland et al., 2000). Symptom

78

severity screening took place via an automated voice response version of the MDASI.
Patients were called weekly for up to 6 weeks.

Patients that never met the symptom severity inclusion criteria described above were
thanked for participating and not entered into the trial. Patients that did meet all inclusion
criteria were randomized into one of the two studies based on their symptom severity
screening. Refer to Appendix A for the study schema. Randomization to one of the two
arms of each study was completed using a computer minimization program balancing
cancer site within each arm by recruitment location (Taves, 1974).

Recruiter Quality Assurance

The PI’s reviewed each screening/enrollment form submitted by the nurse recruiters.
The forms were reviewed for missing data and to verify inclusion criteria. Nurse
recruiters participated in regularly scheduled conference calls with the PPS to discuss and
troubleshoot recruiting issues. Regularly scheduled meetings at MSU were conducted in
which all recruiters were expected to attend. All recruiters were required to submit a
weekly web-based recruitment summary that was reviewed by PI Dr. Barbara Given. Dr.
Barbara Given personally called the nurse recruiter with any questions or concerns
related to patient recruitment.

Intervention Nurse Training

Experienced oncology nurses fi'om Michigan, were hired and trained to implement the
intervention protocols for the nurse intervention arms of both studies (ROl CA-79280
[study one] and ROI CA-30724 [study two}). The PI’s for the two studies Dr. Barbara
Given and Dr. Charles Given interviewed all intervention nurse candidates and made the

final selection for hire. Prior to initiation of the study, all intervention nurses were

79

extensively trained in interactive sessions the included discussion of the objectives and
schema of the studies (ROI CA-79280 [study one] and ROI CA-30724 [study two}),
policies and procedures for intervening and interviewing participants, confidentiality,
MSU IRB policies, and completion of the web-based data management system. Training
sessions involved classroom interactive sessions with the study PI’s as well as practice
taped telephone intervention sessions with MSU College of Nursing doctoral students
acting as patients.

Secure computer passwords were provided to each intervention nurse for entering
information into the web-based system. A computerized stepped intervention protocol
was used for training as well as study intervention. Study manuals were provided that
contained information on the research team and contact information, overall study goals
and objectives, inclusion and exclusion criteria, intervention policies, procedures, and
scripts.

The Intervention Process

The intervention nurses intervened with patients randomized to the nursing arms of
both studies based on the stepped nursing interventions contained in the web-based
system. A total of 17 symptoms were evaluated they included: fatigue, pain, dyspnea,
insomnia, distress, nausea, fever, difficulty remembering, lack of appetite, dry mouth,
vomiting, numbness and tingling, diarrhea, cough, constipation, weakness, and alopecia.
Six contacts were made over an 8 week time period of the study. Patients reporting a
symptom severity of 1 or higher were asked to respond to how that symptom interfered
with daily activities, emotions, enjoyment of life and relationships (Daut, Cleeland, &

F lanery, 1983). Patients randomized to the nurse intervention arm that reported a

80

symptom severity of 4 or higher (range 0-10) received the intervention from the nurse.
Up to 4 strategies were used for each symptom with a severity score of 4 or higher. The
strategies included: teaching, prescribing, communicating with the provider, and
counseling and support. Patients in the nurse intervention arms were also directed to
refer to their symptom management toolkit. At all later contacts the intervention nurse
evaluated the strategies used. The nurse asked if the strategy was tried or not and if
helpful or not. If the strategy was not helpful it was altered and/or a new strategy
proposed.
Intervention Nurse Quality Assurance

Throughout the training periods PI Dr. Barbara Given listened to each practice tape
and monitored the corresponding web-based documentation and provided feedback to the
intervention nurses. Throughout the study The PI’s randomly reviewed tapes and the
web-based documentation completion by the intervention nurses. The PI Dr. Barbara
Given randomly reviewed the web-based data management system to review nursing
interventions for appropriateness and documentation based on the study protocols.
The intervention nurses participated in regularly scheduled conference calls with the PPS
to discuss and troubleshoot issues. Regularly scheduled meetings at MSU were
conducted in which all intervention nurses were expected to attend.

Interviewer Training

Interviewers were hired and trained to implement the interview protocols for all arms
of both studies (ROl CA-79280 [study one] and ROI CA-30724 [study two}). The PI’s
for the two studies Dr. Barbara Given and Dr. Charles Given interviewed all interviewer

candidates and made the final selection for hire. Prior to initiation of the study, all

81

interviewers were extensively trained in interactive sessions the included discussion of
the objectives and schema of the studies (ROI CA-79280 [study one] and ROI CA-
30724 [study two}), policies and procedures for interviewing participants, confidentiality,
MSU IRB policies, and completion of the web-based data management system. Training
sessions involved classroom interactive sessions with the study PI’s as well as practice
taped telephone interviews with MSU College of Nursing doctoral students acting as
patients. A minimum of 3 practice interviews were conducted with different doctoral
students acting out the roles of a depressed and very distraught patient, a patient that is
very talkative and easily distracted, and a very angry and upset patient.

Secure computer passwords were provided to each interviewer for entering
information into the web-based system. Study manuals were provided that contained
information on the research team and contact information, overall study goals and
objectives, inclusion and exclusion criteria, interview policies, procedures (ie., probing
and clarification), and scripts. Many of the interviewers were MSU graduate students in
a health care related field.

The Interview Process

Interviews were conducted via telephone calls by the trained interviewers at baseline
(after symptom screening and prior to intervention), 10 weeks and 16 weeks.
Interviewers used interview software to direct the interview process. Each interview took
approximately 45 minutes to complete. Interviews were scheduled at the patient’s

convenience.

82

Interviewer Quality Assurance
The PI Dr. Barbara Given listened to each of the 3 taped practice interviews for
content, technique, and confidentiality. The web-based documentation for each of the 3
practice interviewer sessions was also evaluated by Dr. Barbara Given. Dr. Given
provided feedback and coaching to each interviewer based on this evaluation.
Throughout the study every 10‘“ interview with the patient’s audible permission was
reviewed by the PI Dr. Barbara Given. Dr. Barbara Given critically evaluated the quality
of the interview (pace of the interview, probing and clarification techniques, and attention
to distress experienced by the patient). Feedback was provided to the interviewer.
Women and Minority Inclusion in Clinical Research

The past NCI (ROl NR/CA 01915) funded study completed by Drs. Barbara and
Charles Given (Given et al., 2000; Given et al., 2001; Given et al., 2002) served as a
benchmark for subject recruitment and retention. This past study included 763 patients at
its preliminary study point, 367 of these patients were male and 337 were female.
However, minority recruitment was low in this past study with approximately 92% of the
patients being white. This value was consistent with the socio-demographic profile of
Michigan and the participating centers in this study. Therefore, some different clinical
sites were recruited for participation in the two current studies (ROI CA-79280 & ROI
CA—30724) to improve minority recruitment. The addition of Holy Cross (a metro-
Detroit clinical site in Michigan) as well as the Yale academic site enhanced minority
recruitment for these two studies. Demographics for this research will be presented in

Chapter 5.

83

Facilities and Resources

All support for the facilities, computer, telephone, and human resources are provided
by two ROI grants fiom NCI, as cited above. This study involved analysis of data
generated by these two funded studies and was supported by The Oncology Nursing
Foundation Trish Green Research Grant and Michigan State University College of
Nursing. Nursing intervention and telephone interviews were conducted at a research
office on the campus of MSU; the automated telephone system is also housed there.
Trained recruiters used phones, faxes, and computers at the clinical sites that are property
of the studies.

Summary

This chapter presented the design and methods that were used for this study as well as
human subject protection and data safety. The purpose of this research was to examine
the dimensions of frequency, intensity, and distress of the co-occurring symptoms of
fatigue, pain and insomnia as they occur at two different data collection points in two
randomized cognitive behavioral intervention clinical trials. Descriptive demographic
statistics, generalized linear regression model analysis, and statistical power for this study
will be presented in Chapter 5. Chapter 5 will also include a discussion of the findings,

strengths and limitations of this study, and implications for clinical practice and research.

84

Chapter 5
Findings

This longitudinal study was designed to examine the dimensions of frequency,
intensity, and distress of the co-occurring symptoms of fatigue, pain and insomnia as they
occur at two different data collection points in two randomized clinical trials of a
cognitive behavioral intervention. This study will answer the questions: At baseline
observation and at 10 weeks is there an association between the dimension of frequency,
intensity, and distress of fatigue and frequency, intensity and distress of pain and/or
insomnia for adults receiving chemotherapy? Also, can categories of response to fatigue
management predict changes in the dimensions of fiequency, intensity and distress of
pain and insomnia at 10 weeks? Covariates examined included: age, site and stage of
cancer, sex, and co-morbid conditions. Participants were recruited fi'om seven academic
and community cancer centers.

Sample

A total of 671 adult chemotherapy patients participated in this study. The majority of
participants (37.4%, =251) were recruited from Indiana University and the Hosier
Oncology Group in Indiana. Grand Rapids, Michigan recruited 14.2% of the participants
(n=95) and 34.3% were from the Pontiac/Detroit city/Detroit suburban area (n=230).
Yale University in Connecticut recruited 8.0% of the participants (n=54) and the
remaining 6.1% (n=41) were recruited in the Lansing or Flint Michigan locations. A total
of 37.5% (n=252) of the participants were recruited from university cancer centers, the

remaining 62.5% (n=419) of the patients were recruited from community cancer centers.

85

The sociodemographic and disease characteristics of the participants are presented in
Table 1. Females made up 70% (n=467) of the patients in this study due to a large
number of breast cancer patients in this study. Thirty percent of the sample were male
(n=204). The mean age of the participants was 57.6 years (SD = 11.79, range 25-90
years). Age distribution is presented in Figure 3. The predominant race of participants
was Caucasian (86.3%, n =579), African Americans accounted for 9.8% (n = 66) of the
participants, and 2.7% (n = 18) were other races such as Mexican Americans, Chicanos,
Native American Alaskans, Oriental Asians, and Pacific Islanders. Race data was
missing for 8 patients (1.2%). Note: the race categories were taken directly from the
data collection instrument, categories were not altered or adapted to adjust for race or

ethnicity standards per NCI or other agencies.

 

30-
25—
[.

ZO-l n H

—l
0|
I
I
:l

 

Frequency

10- - r
w “r r

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

C
-

VIN

0°10-
Ow.—
bug-1
Org)“:
o-b‘

Age

 

Figure 3. Frequency Distribution of the Variable of Age for Study Participants

86

The majority of patients were married, (65.3%, n = 438), 10.3% were never married (11
= 69) and 15.4% (n = 103) were divorced or separated. Fifty patients (7.5%) in this study
were widowed. The education experience of this sample consisted of 118 patients
(17.6%) with a graduate or professional degree, 332 (49.5%) had completed college or
experienced some college or technical training as their highest level of education. One
hundred and sixty—one (24%) participants had completed their education through high
school and the remaining 60 (8.9%) completed grade school or some high school as their
highest level of education.

The most common tumor type of patients in this study was breast cancer (n = 234,
34.9%). There were no males with breast cancer in this study. Non-small cell lung
cancer was the second most common tumor type (11 = 113, 16.8%) and 27 (4.0%) of the
participants had small cell lung cancer. Eighty (11.9%) patients had colon cancer. Fifty-
one (7.6%) of the patients had a genitourinary malignancy and 33 (4.9%) experienced a
gastrointestinal malignancy. The remaining patients had a gynecological malignancy,
non-Hodgkin’s lymphoma, pancreatic cancer or other cancers. Nineteen (2.8%) patients
experienced some other form of cancer not mentioned above.

Cancer diagnosis as it relates to sex was also examined. Females with breast cancer
represented the largest percentage of the sample (34.9%, n = 234). Refer to Table 2 for
cancer site by sex data. Data representing cancer stage in this study is depicted as early
(stage one or two) or late (stage three or four). Overall, the majority (n = 569, 85.6%) of
the patients in this study experienced late stage disease. For 23.2% (n = 156) of patients

in this study this cancer diagnosis represented recurrent disease.

87

Table 1

Descriptive Statistics of Sociodemographic and Disease Characteristics of Patients

 

Characteristic N %
Patient Gender
Female 467 70.0
Male 204 30.0
Patient Ethnicity
Caucasian 579 86.3
African American 66 9.8
Other 18 2.7
Missing 8 1.2
Marital Status
Married 438 65.3
Divorced or Separated 103 15.4
Never Married 69 10.3
Widowed 50 7.5
Living Together 10 1.5
Education
Did not completed high school 60 8.9
Completed high school 161 24.0
Some college or technical training 202 30.1
Completed college 130 19.4
Completed graduate degree 118 17.6
Cancer Site
Breast 234 34.9
Non-small cell lung 113 16.8
Colon 80 11.9
GU 51 7.6
Gynecological 47 7.0
G] 33 4.9
NHL 38 5.7
Small cell lung 27 4.0
Pancreas 21 3.1
Mesothelioma 8 1 .2
Other 19 2.8
Stage of Cancer
Early (stage 1 or 2) 96 14.4
Late (stage 3 or 4) 569 85.6
Disease Recurrence
Yes 156 ' 23.2
No 515 76.8
Range Mean (SD) - Median
Patient Age 25-90 years 57.6 (11.79) 58.0
Number of Co-Morbid Conditions 0-9 2.07 (1.61) 2.00

 

88

Table 2

Number / % Statistics of Cancer Site by Sex

 

 

Cancer Site and Sex N %
Breast — Female 234 34.9
Lung — Female 77 11.5
Lung — Male 63 9.4
Colon — Female 53 7.9
Colon — Male 27 4.0
Other Cancer — Female 103 15.4
Other Cancer — Male 1 14 16.9

 

Co-morbid conditions were evaluated for this study. Eighty-one percent of the
participants in this study experienced one or more co-morbid conditions upon entry into
this study. The mean number of reported co-morbid conditions was 2.07 (SD 1.61 , range
0 — 9). Only 125 patients (18.6%) reported 0 co-morbid conditions. See Table 3 for a list
of co-morbid conditions identified in this study. The most commonly reported co-morbid
condition in this study was high blood pressure (11 = 285, 42.5%) followed by emotional
problems (11 = 175, 26.1%) and another cancer (n = 134, 20%), 118 (17.6%) reported
urinary incontinence and 116 patients reported heart problems (17.3%).

Measures

Descriptions of the measures used in this study were provided in Chapter 4. A
discussion of the mean values for the dimensions of frequency, intensity and distress for
the symptoms of fatigue, pain and insomnia at baseline and 10 weeks will be presented
here. Recall that fi'equency was described as the number of days out of the past seven
that the symptom was present. Intensity was described as a ranking of the symptom on a
scale of 0 meaning not present to 10 meaning the intensity is the worst possible. Distress

was recorded as a scale with 0 meaning the symptom does not bother the patient to 10

89

meaning the symptom causes the worst bother the patient can imagine. See Table 4 for
the mean values, standard deviations and possible ranges for each symptom (fatigue,
pain, and insomnia) at baseline and 10 weeks for all subjects (n = 671) regardless of
group membership.

Table 3

Number / % Statistics of Co-morbid Conditions

 

 

 

Co-morbid Condition N ‘Zg
High Blood Pressure 285 42.5
Emotional Problems 175 26.1
Other Cancer 134 20.0
Urinary Incontinence 1 18 17.6
Heart Problem 1 16 17.3
Diabetes 84 12.6
Cataract Surgery 70 10.4
Emphysema 67 10.0
Arthritis 62 9.2
Hearing Aid 41 6.1
Replace Joint 37 5.5
Chest Pain 25 3.7
Stroke 20 3.0
Fractured Hip 7 1.0
Other Major Health Problem 151 23.3

Group Membership and Attrition

Group membership was not expected to result in statistically significant differences
based on previous work with this same data set. Sikorskii, Given, Given, Jeon, Decker &
Decker (in press) noted that both the NASM and the ATSM groups achieved significant
reductions in symptom severity over baseline and at 10 weeks (post intervention) with
effect sizes exceeding .5 and no differences by group. This study will adjust for group

membership and examine group effect via generalized linear regression models.

90

Table 4

Mean Values, Standard Deviations and Possible Ranges of the Symptoms per Dimension

 

 

Fatigue Fatigue Pain Pain Insomnia Insomnia
Baseline 10 weeks Baseline 10 weeks Baseline 10 weeks
Frequency
mean 4.70 3.68 2.28 1.93 3 .07 1.62
(SD) (2.49) (2.85) (2.77) (2.77) (2.74) (2.41)
031186) (0-7) (0-7) (0-7) (0-7) (0-7) (0-7)
Intensity
mean 4.71 3.22 2.32 1.63 3.77 1.87
(SD) (2.69) (2.67) (2.81) (2.41) (3.27) (2.74)
(range) (0-10) (0-10) (0-10) (0-10) (0-10) (0-10)
Distress
mean 4.18 2.52 2.02 1.31 3.03 1.31

(SD) (3.06) (2.78) (2.88) (2.40) (3.08) (2.33)
(range) (0-10) (0-10) (0-10) (0-10) (0-10) (0-10)

 

A total of 1605 cancer patients were eligible and approached by nurse recruiters for
either of the two studies, 815 signed informed consent forms, and 806 patients initiated
the symptom screening (Sikorskii et al., in press). Seven hundred and twenty-eight met
the inclusion criteria to enter into one of the two trials. A total of 671 patients completed
baseline interviews, and 533 completed 10 week interviews. Sikorskii et al. (in press)
examined attrition with this data set; they found that patients who attrited did not differ
significantly on summed symptom severity at baseline. In study number 2 attrited
patients averaged 42.32 for summed symptom severity, NASM patients averaged 41.47
(p = 0.89).

Results
Regression analysis was conducted for each research question using SPSS ® Version

13.0 Graduate Pack for Windows statistical software package. Minimal missing data was

91

noted in the final data set, n=671, therefore, all cases were used for final analysis with no
missing data procedures used. All regression models for all research questions will
include group assignment (experimental or control) and study name (ATSM or Pain and
Fatigue [NASM]) to allow adjustment for these variables. Recall differing inclusion
criteria for the two different studies. To enter into the NASM group’s patients needed to
report a 2 out of 10 for frequency of pain and fatigue at baseline or a 3 out of 10 for pain
or fatigue on the MDASI screening instrument used for study participation. ATSM group
inclusion criteria required that the patient reported a 2 or higher on any one of the 13
symptoms assessed in the MDASI. Symptoms assessed in the MDASI include: fatigue,
pain, disturbed sleep, distress (emotional), nausea, shortness of breath, lack of appetite,
dry mouth, drowsy, emesis, numbness or tingling, bloated, sad (Cleeland et al., 2000).
Thus, in this study examining the symptoms of fatigue, pain, and insomnia patients in the
NASM study were expected to experience a greater effect of pain and fatigue at baseline
vs. their ATSM counterparts. It is possible that ATSM patients may not even experience
fatigue, pain or insomnia and participate in this study due to the inclusion criteria that
allowed any report of a 2 or higher for any one of the 13 symptoms from the MDASI
listed above.

The outcome variables examined in this study include frequency, intensity and distress
of fatigue, pain and insomnia as described previously. The covariates included: age, site
and stage of cancer, sex, and co-morbid conditions. Age will be the age of the patient at
baseline of the study and represents a continuous variable. Co-morbid conditions will be
represented by a continuous variable and be the total number of co-morbid conditions

that the patient reports at baseline. Stage of cancer was determined in the NASM and

92

ATSM studies to be a categorical variable. Early stage represents cancer stages 1 or 2
per the TNM staging system and late represents stages 3 or 4 per the TNM system at
baseline of the study as determined per the medical record audit. Site of cancer was
originally examined in this study as a categorical variable that combined sex and cancer
site, for example, females with colon cancer, males with lung cancer, females with an
“other” cancer, a total of 8 categories were developed. This variable was not statistically
significant in the regression models, with the exception of females with lung cancer being
different form the referent category on baseline insomnia. Based on this finding as well
as a review of the literature, the cancer site category for this study was represented as
lung cancer or other cancer (Cooley, 2002; Cooley & Moriarty, 2003; Fu et al., 2005;
Gift et al., 2003; Given et al., 2001; Given et al., 2002; Kurtz et al., 2000). Sex was a
categorical variable as male or female.

The following format guided the data input and interpretation of the generalized linear
regression model analysis for research questions 1-3. Frequency, intensity, or distress
level of pain or insomnia was the outcome for each model respective of the question or
portion of the question being examined. Covariates for all models included the
dimension of fatigue, group assignment, and study name. Covariates examined when the
portion of the question sought to examine influencing variables included: age, site and
stage of cancer, sex, and number of co-morbid conditions. Models were constructed for
baseline findings as well as 10 week findings. Results of the data analysis for research
questions 1-3 will be presented below. Research question 1 will be described in detail,

research questions 2 and 3 were conducted in the same manner, for the sake of avoiding

93

redundancy the results will be outlined. Research question 4 design and results will be
described in detail below.
Research Question 1

At baseline observation, prior to entry into a clinical trial, and at 10 weeks is there an
association between the dimension of frequency of fatigue and frequency of pain, as well
as fiequency of fatigue and frequency of insomnia for adults receiving chemotherapy? Is
this association influenced by; age, site and stage of cancer, sex, and co-morbid
conditions? As noted above, generalized linear regression model analysis was conducted
with frequency of pain at baseline as the dependent variable, study name, group
assignment and frequency of fatigue at baseline as covariates. This model supported that
there is an association between the dimensions of frequency of fatigue and frequency of
pain at baseline ([3 = .38, SE = .04, t = 9.40, p < .01, not in tables). Generalized linear
regression model analysis was also conducted with frequency of pain at 10 weeks as the
dependent variable, study name, group assignment and frequency of fatigue at 10 weeks
as the covariates. This model supported that there is an association between the
dimensions of frequency of fatigue and frequency of pain at 10 weeks ([3 = .27, SE = .04,
t = 6.75, p < .01 , not in tables). Refer to Table 5 for results of the regression analysis for
the pain and fatigue model with all covariates added to the model.

To examine the effects of the covariates a generalized linear regression model
examined pain at baseline as the dependent variable and the covariates as cited above.
The model included the frequency of pain at baseline as the dependent variable, the
covariates: sex, cancer site as a categorical variable of lung cancer or other cancer, stage

of disease as a categorical variable as early or late, patient age and number of co-morbid

94

conditions were continuous variables as well as frequency of fatigue at baseline. This
model demonstrated that there remains an association between the dimensions of
fiequency of fatigue and frequency of pain at baseline when the covariates are included in
the model (B = .35, SE = .04, t = 8.64, p < .01). Age (B = -.04, SE = .01, t = -3.78, p <
.01) demonstrated a negative association with this model and the number of co-morbid
conditions (B = .21, SE = .07, t = 2.98, p < .01) demonstrated a positive association.

Table 5

Regression Analyses for Relating Frequency of Pain to Frequency of Fatigue and Other

Covariates at Baseline and 10 Weeks

 

 

 

Baseline 10—Weeks
Parameter Est. St. T P- Est. St. T P-
Error value Error value
Fatigue
Frequency .35 .04 8.64 < .01 .29 .04 7.10 < .01
Age -.04 .01 -3.78 < .01 -.03 .01 -2.42 02

Comorbids .21 .07 2.98 <01 .24 .08 2.96 <'.01
Sex

Male .09 .22 .39 .70 -.17 .26 -.66 .51
Female 000 . . . 000

Stage
Early ‘ -.24 .29 -.83 .41 -.17 .31 -.55 .58
Late 000 . . . 000

Cancer
Lung .28 .25 1.1 1 .27 -.06 .31 -.20 .85

Non-Lung 000 . . . 000

 

Ten week data were also examined via a generalized linear regression model with the
effects of the covariates cited above. This model demonstrated that there remains an
association between the dimensions of frequency of fatigue and frequency of pain at 10
weeks (B = .29, SE = .04, t = 7.10, p < .01). Again, age (B = -.03, SE = .01, t = -2.42, p =

.02) demonstrated a negative association and the number of co-morbid conditions (B =

95

.24, SE = .08, t = 2.96, p < .01) demonstrated a positive association. Thus, being younger
and having more co-morbid conditions at baseline was associated with a higher frequency
of pain at baseline and 10 weeks over and above the frequency of fatigue at baseline or
10 weeks.

An additional model was developed that examined the fiequency of insomnia at
baseline as the dependent variable and the frequency of fatigue at baseline. This model
supported an association between the dimensions of frequency of fatigue and fiequency
of insomnia at baseline (B = .21, SE = .04, t = 4.92, p < .01, not in tables). The frequency
of insomnia at 10 weeks as the dependent variable and the frequency of fatigue at 10
weeks as the independent variable were also examined via generalized linear regression.
This model supported that there is an association between the dimensions of frequency of
fatigue at 10 weeks and frequency of insomnia at 10 weeks (B = .24, SE = .04, t = 6.70, p
< .01, not in tables). Refer to Table 6 for results of the regression analysis for the
insomnia and fatigue model with all covariates added to the model.

A generalized linear regression model was rrm with frequency of insomnia at baseline
as the dependent variable and included all of the covariates cited above. This model
demonstrated that there remains an association between the dimensions of frequency of
fatigue at baseline and frequency of insomnia at baseline that is not significantly affected
by the addition of the covariates (B = .18, SE = .04, t = 4.32, p < .01). Age demonstrated a
negative association with this model (B = -.04, SE = .01, t = -4.39, p < .01). The number
of co-morbid conditions (B = .20, SE = .07, t = 2.71, p < .01) demonstrated a positive
association. Differing fi'om the models with pain at baseline as the dependent variable

cited above, over and above other covariates in the model lung cancer patients had

96

significantly lower fi'equency of insomnia compared to non-lung cancer patients, (B = -
.52, SE = .26, t = -l .99, p = .05). See Table 6.

Table 6

Regression Analyses for Relating Frequency of Insomnia to Frequency of Fatigue and

Other Covariates at Baseline and 10 Weeks

 

 

 

Baseline lO-Weeks
Parameter Est. St. T P- Est. St. T P-
Error value Error value
Fatigue
Frequency .18 .04 4.32 < .01 .25 .04 7.03 < .01
Age -.04 .01 -4.40 < .01 -.02 .01 -1.56 .12
Comorbids .20 .07 2.71 < .01 .02 .07 .25 .81
Sex
Male .18 .23 .79 .43 .1 1 .23 .49 .63
Female 000 . . . 000
Stage
Early .13 .30 .44 .66 .17 .28 .63 .53
Late 000 . . . 000
Cancer
Lung -.52 .26 -l .99 .05 .1 1 .23 .49 .63
Non-Lung 000 . . . 000 .

 

The same model was run with fiequency of insomnia at 10 weeks as the dependent
variable; all other variables were the same as noted above. This model with frequency of
insomnia at 10 weeks as the dependent variable and all covariates demonstrated that there
remains an association between the dimensions of frequency of fatigue at 10 weeks and
frequency of insomnia at 10 weeks (B = .25, SE = .04, t = 7.03, p < .01). However, all
other covariates were not significantly associated with frequency of insomnia in this
model. Age, co-morbid conditions and lung cancer no longer had a significant
association with the outcome, frequency of insomnia, at 10 weeks. Thus, being younger,

having more co-morbid conditions, and not having lung cancer diagnosis at baseline was

97

associated with a higher frequency of insomnia at baseline over and above the frequency
of fatigue at baseline, however, age, number of co-morbid conditions, and site of cancer
did not influence insomnia over and above the frequency of fatigue at 10 weeks. See
Table 6.

Research Question 2

At baseline observation, prior to entry into a clinical trial, and at 10 weeks is there an
association between the dimension of intensity of fatigue and intensity of pain, as well as
intensity of fatigue and intensity of insomnia for adults receiving chemotherapy? Is this
association influenced by; age, site and stage of cancer, sex, and co-morbid conditions?
Generalized linear regression model analysis was conducted with intensity of pain at
baseline as the dependent variable. This model supported that there is an association
between the dimensions of intensity of fatigue and intensity of pain at baseline (B = .40,
SE = .04, t = 10.81, p < .01, not found in tables) and at 10 weeks (B = .31, SE = .04, t =
8.22, p < .01, not found in tables). Refer to Table 7 for results of the regression analysis
for the pain and fatigue model with all covariates added to the model.

Examination of the covariates resulted in a model that continued to support an
association between the dimensions of intensity of fatigue and intensity of pain at
baseline (B = .37, SE = .04, t = 9.61, p < .01). Similar to the findings noted above with
frequency of pain, age (B = -.04, SE = .01, t = -3.76, p < .01) demonstrated a negative
association within this intensity of pain model and the number of co-morbid conditions (B

= .19, SE = .07, t = 2.74, p < .01) demonstrated a positive association. See Table 7.

98

Table 7

Regression Analyses for Relating Intensity of Pain to Intensity of Fatigue and Other

Covariates at Baseline and 10 Weeks

 

 

 

Baseline 10-Weeks
Parameter Est. St. T P- Est. St. T P-
Error value Error Value
Fatigue
Intensity .37 .04 9.61 < .01 .30 .04 8.16 < .01
Age -.04 .01 -3.76 < .01 -.03 .01 -3.59 < .01

Comorbids .19 .07 2.74 < .01 .25 .07 3.62 < .01
Sex

Male .08 .23 .34 .74 -. 14 .22 -.62 .54
Female 000 . . . 000
Stage
Early -.03 .29 -.11 .92 -.32 .27 -l .19 .24
Late 000 . . . 000
Cancer
Lung .02 .26 .06 .96 -.17 .26 -.64 .52
Non-Lung 000 . . . 000

 

A model was also conducted with intensity of pain and fatigue at 10 weeks and all
covariates. This model supported an association between the two variables at 10 weeks
(B = .30, SE = .04, t = 8.16, p < .01). Age (B = —.03, SE = .01, t = -3.59, p < .01)
demonstrated a negative association and the number of co-morbid conditions (B = .25, SE
= .07, t = 3.62, p < .01) demonstrated a positive association. Younger age and having
more co-morbid conditions at baseline were associated with a higher intensity of pain
over and above the intensity of fatigue at baseline and 10 weeks. Refer to Table 7.

The intensity of insomnia and fatigue at baseline were also studied. An association
between the dimensions of intensity of fatigue and intensity of insomnia was supported at

baseline (B = .35, SE = .05, t = 7.76, p < .01, not in tables) and at 10 weeks (B = .40, SE =

99

.04, t = 9.73, p < .01, not in tables). See Table 8 for the regression analysis for the

intensity of insomnia and fatigue model with all covariates added.

The baseline model with all the covariates added demonstrated that an association

between the dimensions of intensity of fatigue and intensity of insomnia at baseline

remains (B = .32, SE = .05, t = 6.83, p < .01). Age was the only covariate that

demonstrated a significant association (B = -.04, SE = .01, t = -3.37, p < .01).

Table 8

Regression Analyses for Relating Intensity of Insomnia to Intensity of Fatigue and Other

Covariates at Baseline and 10 Weeks

 

 

 

Baseline lO-Weeks
Parameter Est. St. T P- Est. St. T P-
Error value Error value
Fatigue
Intensity .32 .05 6.83 < .01 .41 .04 9.95 < .01
Age -.04 .01 -3.37 < .01 -.02 .01 -1.98 .05
Comorbids . 12 .09 1.36 .176 .03 .08 .34 .73
Sex
Male -.02 .28 -.06 .95 -.O4 .25 -.18 .86
Female 000 000
Stage
Early .22 .35 .62 .54 .25 .30 .83 .41
Late 000 000
Cancer
Lung -.45 .31 -1.44 .15 -.20 .29 -.68 .50
Non- Lung 000 . . 000

 

Intensity of insomnia and intensity of fatigue with all covariates added at 10 weeks

was also studied. This model supported that there is an association between the

dimensions of intensity of fatigue and intensity of insomnia at 10 weeks (B = .41, SE =

.04, t = 9.95, p < .01). Age continued to demonstrate a negative association (B = -.02, SE

= .01, t = -1.98, p = .05). Younger age was associated with a higher intensity of insomnia

100

at baseline and 10 weeks over and above the intensity of fatigue at baseline or 10 weeks.
See Table 8.
Research Question 3

At baseline observation, prior to entry into a clinical trial, and at 10 weeks is there an
association between the dimension of distress related to fatigue and distress related to
pain, as well as distress related to fatigue and distress related to insomnia for adults
receiving chemotherapy? Is this association influenced by; age, site and stage of cancer,
sex, and co-morbid conditions? Generalized linear regression model analysis was
conducted with distress related to pain at baseline and distress of fatigue at baseline. This
model supported an association between the dimensions of distress of fatigue and distress
of pain at baseline (B = .38, SE = .03, t = 11.46, p < .01, not in tables). See Table 9 for
results of the regression analysis for the distress of pain versus distress of fatigue model
with all covariates added to the model.

The effects of the covariates were also examined. This model demonstrated that there
remains an association between the dimensions of distress of fatigue and distress of pain
at baseline with the covariates (B = .35, SE = .03, t = 10.20, p < .01). Age (B = -.04, SE =
.01, t = -3.70, p < .01) demonstrated a negative association and the number of co-morbid
conditions (B = .30, SE = .07, t = 4.23, p < .01) demonstrated a positive association with
level of pain at baseline.

Generalized linear regression model analysis was conducted with distress of pain
versus distress of fatigue at 10 weeks. This model supported an association between
distress of fatigue and distress of pain at 10 weeks (B = .34, SE = .04, t = 9.87, p < .01,

not found in tables). The effects of the covariates were examined. An association

101

between the dimensions of distress of fatigue and distress of pain at 10 weeks was

maintained with the covariates in the model (B = .34, SE = .04, t = 9.70, p < .01). Again,

age (B = -.03, SE = .01, t = -3.20, p < .01) demonstrated a negative association and

number of co-morbid conditions (B = .16, SE = .07, t = 2.38, p = .02) demonstrated a

positive association. Younger age and a greater number of co-morbid conditions at

baseline were associated with a higher distress of pain at baseline and 10 weeks over and

above the distress of fatigue at baseline or 10 weeks. See Table 9.

Table 9

Regression Analyses for Relating Distress of Pain to Distress of Fatigue and Other

Covariates at Baseline and 10 Weeks

 

 

 

Baseline 10-Weeks
Parameter Est. St. T P- Est. St. T P-
Error value Error value
Fatigue
Distress .35 .03 10.20 < .01 .34 .04 9.70 < .01
Age -.04 .01 -3.70 < .01 -.03 .01 -3.20 < .01
Comorbids .30 .07 4.23 < .01 .16 .07 2.38 .02
Sex
Male .14 .23 .62 .54 -.19 .22 -.87 .38
Female 000 000
Stage
Early -.1 1 .29 -.38 .71 -.25 .26 -.94 .35
Late 000 000
Cancer
Lung -.06 .26 -.22 .83 -.19 .22 -.87 .38
Non— Lung 000 000

 

An additional model was developed that examined the distress of insomnia and the

distress of fatigue at baseline. This model supported an association between the

dimensions of distress of fatigue and distress of insomnia at baseline (B = .3 7, SE = .04, t

= 10.07, p < .01, not in tables). Refer to Table 10 for results of the regression analysis for

102

the distress of insomnia and fatigue model with all covariates added. This same model
was run with all covariates. This model demonstrated that there remains an association
between distress of fatigue and distress of insomnia at baseline when all covariates are
included (B = .34, SE = .04, t = 9.18, p < .01). Age was the only covariate that was
significant in this model (B = -.04, SE = .01, t = -3.71, p < .01). See Table 10.

Distress of insomnia and the distress of fatigue at 10 weeks were also examined. This
model supported an association between the two variables (B = .33, SE = .03, t = 9.89, p
< .01, not found in tables). The covariates were added to this model to demonstrate that
there remained an association between the dimensions of distress of fatigue and distress
of insomnia at 10 weeks (B = .33, SE = .03, t = 9.82, p < .01). However, unique to this
question, no covariates had a significant association with distress of insomnia at 10
weeks. Younger age was associated with a higher distress of insomnia at baseline over
and above the distress of fatigue at baseline; however, age did not influence distress of
insomnia over and above the effect of fatigue at 10 weeks. Refer to Table 10.

Research Question 4

Using categories of response to fatigue management (none / mild, non-response,
partial response and full response) established in the literature, determine how changes in
the dimensions of frequency, intensity, and distress of pain and insomnia are predicted by
categories of response to the management of fatigue between baseline and 10 weeks? Is
this association influenced by; age, site and stage of cancer, sex, and co-morbid

conditions?

103

Table 10

Regression Analyses for Relating Distress of Insomnia to Distress of Fatigue and Other

Covariates at Baseline and 10 Weeks

 

 

 

Baseline 10-Weeks
Parameter Est. St. T P- Est. St. T P-
Error value Error value
Fatigue
Distress .34 .04 9.18 < .01 .33 .03 9.82 < .01
Age -.04 .01 -3.71 < .01 -.02 .01 -1.75 .08
Comorbids . 14 .08 1.75 .08 .10 .07 1.44 .15
Sex
Male .05 .25 .19 .85 .09 .21 .43 .67
Female 000 000
Stage
Early .30 .32 .94 .35 -.03 .26 -.12 .90
Late 000 000
Cancer
Lung -.31 .28 -1.08 .28 -.24 .25 -.95 .35
Non— Lung 000 000

 

Fatigue intensity levels (range 0-10) were converted from continuous to categorical

variables to facilitate comparison with fatigue response categories (Mendoza, Wang,

Cleeland, Morrissey, Johnson, Wendt, et al., 1999; Miaskowski, et al., in press).

Theoretically continuous variables provide greater power for statistical tests. However,

power was of less concern with a sample size of 671 versus clinical relevance for

analyses of data for this question. The measurement of intensity of fatigue on a scale of

0-10 is common in clinical practice. Guyatt, Norman, Juniper, and Griffith (2002)

suggest that a reduction in symptom severity that ranges from 33% to 50% is clinically

significant.

However, clinical interpretation of changes in intensity levels between measurement

periods indicate that the 0-10 fatigue scale is not necessarily considered an interval level

104

scale by practitioners. As an example, if a patient reports their fatigue intensity at a 9 at
baseline and a 6 at 10 weeks one would say that the patient experienced a 33%
improvement in fatigue intensity; however, clinically this patient is still experiencing a
severe level of fatigue that requires intervention. I doubt that a practitioner would accept
a 6 for fatigue intensity as being acceptable, it is an improvement, however, the symptom
remains unresolved. On the contrary, a patient experiencing a fatigue intensity level of 3
at baseline and a 2 at 10 weeks likewise represents a 33% reduction in fatigue intensity.
However, this 33% reduction in intensity level for fatigue has a completely different
clinical interpretation then the prior example. Thus, when using a 0-10 scale to measure
fatigue intensity both percent change and absolute value change values are not always
clinically meaningful.

Given et al. (in press) used data from 339 patients in this current studies database to
examine symptom response categories. It was noted by Given et al. that as intensity
(severity) of symptoms increased, interference (distress) level of the symptom did not
linearly increase. Differences in the associations between severity and distress were
noted to appear on the 0-10 scales between 1 and 2, and 4 and 5. Thus, the establishment
of cut points to create clinically meaningful fatigue categories.

The fatigue intensity categories for this current study were 0 = none, 1 = mild fatigue,
2-4 moderate fatigue, 5-10 severe fatigue. Response categories for the management of
fatigue were determined by comparing the baseline fatigue intensity category with the 10
week fatigue intensity category. Response categories for the management of fatigue are
as follows: None/mild = patient stays within the same none or mild fatigue intensity

category at the baseline measurement as well as the 10 week measurement. Non-

105

responder = a patient who stays within the same category (excluding the none or the mild
category) fiom the baseline measure of fatigue to the 10 week measurement, (i.e., severe
at baseline and severe at 10 weeks). Partial responder = a patient who goes fi'om severe
to the moderate level of fatigue between baseline and 10 weeks. Full responder = a
patient who goes to the none or mild category of fatigue fi'om moderate or severe
between baseline and 10 weeks (Given et al., in press; Paul, Zelman, Smith, &
Miaskowski, 2005). Refer to Table 11 for a frequency table of fatigue management
groups.

Table 11

Frequency and % of Fatigue Management Group Membership

 

 

Management of Fatigue Frequency Percent
None / mild 47 8.8
Non-responder 300 56.3
Partial responder 79 14.8
Full responder 107 20. 1

 

Generalized linear regression model analysis was conducted with frequency of pain at
10 weeks as the dependent variable and the management of fatigue category and pain at
baseline as covariates. This model supported an association between the dimensions of
fiequency of pain at baseline and frequency of pain at 10 weeks when management of
fatigue categories are included in the model (B = .43, SE = .04, t = 10.61, p < .01). The
only fatigue management category with a significant effect was the non-responders (B =
.93, SE = .28, t = 3.34, p = .01). Non-responders had significantly higher frequency of
pain at 10 weeks compared to fill] responders. See Table 12.

A generalized linear regression model was also conducted for frequency of pain at 10

weeks, the management of fatigue category, pain at baseline, cancer stage, site and sex,

106

age, and co-morbids. This model supported the association between fiequency of pain at
baseline and 10 weeks (B = .41, SE = .04, t = 9.91, p < .01). Other significant effects
were, the non-responder fatigue management category (B = 1.07, SE = .28, t =3.78, p <
.01) and co-morbid conditions (B = .15, SE = .08, t =1.95, p = .05). Therefore, non-
responders (as compared to full responders) and patients with a greater number of co-
morbid conditions had significantly higher frequency of pain at 10 weeks, see Table 12.
Generalized linear model regression analysis was conducted with intensity of pain at
10 weeks as the dependent variable, management of fatigue category, and intensity of
pain at baseline as the covariates. This model supported an association between the
variables (B = .31, SE = .04, t = 8.70, p < .01). Two fatigue management categories had
significant effects; the non-responders (B = .97, SE = .25, t = 3.88, p < .01) and the partial
responders (B = .75, SE = .33, t = 2.29, p = .02). Non-responders and partial responders
had significantly higher intensity levels of pain at 10 weeks compared to full responders.
Model analysis was conducted with the same variables plus all covariates. This model
supported an association between intensity of pain at baseline and 10 weeks (B = .29, SE
= .04, t = 7.96, p < .01). Two fatigue management categories had significant effects; the
non-responders (B = 1.05, SE = .25, t = 4.18, p < .01) and the partial responders (B = .65,
SE = .33, t = 1.95, p = .05). Age (B = -.03, SE = .01, t = -2.66, p = .01) and number of
co—morbid conditions also contributed to this model (B = .23, SE = .07, t = 3.43, p < .01).
Non-responders and partial responders had significantly higher intensity of pain at 10
weeks compared to full responders. Younger patients and patients with greater numbers
of co-morbid conditions experienced higher intensity levels of pain at 10 weeks. Refer to

Table 12.

107

hinge on

33:: oxammmxmmho: manhfiwkc‘. 3.323%. SEES. Am fihmaswmm oboe? 85 35.38 398353“ 06th9.! E h o 3&th

 

 

 

 

 

 

33358, 3:85me 338m
3.8.3084 mm". m”. H w- mm”. my a w- mm”. m”. H w-
max. <38 mien $98 mag $55
30%: 23. was 3 womozea go Womeosmo 8 mmmmca 385382: A? 738 m: 5 $5an
was .3 .3 3.3 A .3 .3 .3 who A .3 .NM .o3 who A .3
3 73m
298.202.. -.3 .3 -3 .ou he -.wm .3 -.mo .3 -.3 3o -hm .3
28-33% .8 .mm P3 A .3 .3 .3 who A .3 .3 .No Pom A .3
35-33% .3 .3 Bow .3 hm .3 who .om .mm .3 row .2
3530mm ooo . . . ooo . . . ooo . .
Zona— Sfl: was a 3.83% ea Wamooamo .8 mmmmco 333089: A3 33% m: _o 28.6 BE m: 005538
was .3 .3 0.3 A .3 .No .3 Sea A .3 .No .3 Mum A .3
3 73m
203928,.“ -.3 .3 -hh .3 -.3 .3 he .3 -.3 Lo -.mw .oo
205-33% 3.3 .mm who A .3 row .Nm 3.; A .3 3.3 .No .Co A .3
308.38% .3 :3 3m .3 .3 .3 mom .8 .3 .3 _.mo .3
mazgme ooo . . . ooo . . . ooo . . .
>mo -.3 .3 ha .3 -.om .3 boo .3 -.ou .3 boo A .3
003393 .3 .8 row .8 .Nu .3 93 A .3 .3 .3 P3 .8
max - 733 A: .3 -.om .mo -.3 .mm ..3 .3 -.No .8 -.mh be
moan—o ooo . . . ooo . . . ooo . . .
mSma .. m3? -.3 .No -.oN be -.Nm .uo -.om .3 A; .3 ..oo .mo
H38 ooo . . . ooo . . . ooo . . .
088$ - -.oo .8 -.3 ho -.oh .3 -.Nm .mo .5 .3 .3 .3m
95m
Zoe- 55m ooo ooo ooo .

 

108

Generalized linear model regression analysis was conducted with distress of pain at 10
weeks as the dependent variable, management of fatigue category, and distress of pain at
baseline as the covariates. This model supported an association between these variables
(B = .22, SE = .04, t = 5.98, p < .01). One fatigue management category contributed, the
non-responders (B = .94, SE = .26, t = 3.67, p < .01). This same model was repeated with
all covariates added. The association between the dimensions of distress of pain at
baseline and 10 weeks remained present (B = .20, SE = .04, t = 5.35, p < .01). Again, one
fatigue management category had a significant effect, the non-responders (B = 1.07, SE =
.26, t = 4.10, p < .01), age (B = -.03, SE = .01, t = -2.89, p < .01) and number ofco-
morbid conditions (B = .17, SE = .07, t = 2.44, p = .02). Non-responders had
significantly higher distress of pain at 10 weeks compared to fill responders. Younger
patients and patients with greater numbers of co-morbid conditions experienced higher
distress of pain at 10 weeks. See Table 12.

General linear model regression analysis was conducted with fiequency of insomnia at
10 weeks as the dependent variable, fatigue management categories, and frequency of
insomnia at baseline as covariates. This model supported an association between the
dimensions of frequency of insomnia at baseline and the frequency of insomnia at 10
weeks (B = .21, SE = .04, t = 5.59, p < .01). One fatigue management category had a
significant effect, the non-responders (B = 1.19, SE = .26, t = 4.64, p < .01). This same
model was run with all covariates added. This model continued to support an association
between the dimensions of frequency of insomnia at baseline and 10 weeks (B = .19, SE
= .04, t = 5.00, p < .01). The non-responder fatigue management category continued to

have a significant effect (B = 1.28, SE = .26, t = 4.88, p < .01). Thus, the non-responders

109

had significantly higher frequency of insomnia at 10 weeks compared to fiill-responders.
Refer to Table 13.

General linear model regression analysis was conducted with intensity of insomnia at
10 weeks as the dependent variable, fatigue management categories, and intensity of
insomnia at baseline as covariates. This model supported an association between the
variables (B = .16, SE = .04, t = 4.48, p < .01). Two fatigue management categories also
had a significant effect; the non-responders (B = 1.67, SE = .29, t = 5.72, p < .01) and the
partial responders (B = 1.27, SE = .39, t = 3.30, p = .01). This same model with all
covariates added supported the association between the dimensions of intensity of
insomnia at baseline and 10 weeks (B = .14, SE = .04, t = 3.75, p < .01). Two fatigue
management categories continued to have a significant effect; the non-responders (B =
1.79, SE = .30, t = 6.02, p < .01) and the partial responders (B = 1.31, SE = .39, t = 3.35, p
= .01). No additional covariates contributed to the model. Non-responders and partial
responders had significantly higher intensity of pain at 10 weeks compared to full
responders. Refer to Table 13.

General linear model regression analysis was conducted with the distress of insomnia
at 10 weeks as the dependent variable, fatigue management categories, and the distress of
insomnia at baseline as covariates. This model supported an association between the
dimensions of distress of insomnia at baseline and the distress of insomnia at 10 weeks (B
= .17, SE = .03, t = 5.13, p < .01). Two fatigue management categories had a significant
effect; the non-responders (B = 1.22, SE = .25, t = 4.89, p < .01) and the partial
responders (B = .91, SE = .33, t = 2.77, p = .01). This same model was run with all

covariates. This model continued to support the association between the dimensions of

110

distress of insomnia at baseline and 10 weeks (B = .16, SE = .04, t = 4.53, p < .01). The
same two fatigue management categories continued to contribute; the non-responders (B
= 1.32, SE = .25, t = 5.18, p < .01) and the partial responders (B = .90, SE = .34, t = 2.68,
p = .01). No additional covariates contributed to the model. Thus, non-responders and
partial responders had significantly higher intensity of pain at 10 weeks compared to fill]
responders. Refer to Table 13.
Power

Following data analysis post hoc power analysis was performed for each research
question. The observed effect size was determined for the essential parameters tested in
the models. The magnitude of the effect size and the sample size were considered in
calculating statistical power. In power calculations for regression the effect size of .02 is
considered small; effect sizes below .01 represent extremely small effects with no
practical interpretation. Table 14 lists the variables with eta squared values of .01 or
greater and the corresponding values of statistical power to detect the effects of these
variables on the outcomes. Power was considered adequate (3 .80) for the statistical tests

completed for this study.

111

H33 5

~93: $5343.26: m§~emw§x 3.3233 3835‘. am baasa e\~§c§=€ SE 3.83% Kagmmiai 0.98ch N a Emmi

 

 

 

 

 

 

53:33 585me 33.03
3.88533 ma. mr H w- mmfi. mg. a. w- mm”. m». H my.
mac”. <38 5.8.. <38 mndn <38
7333 2:: 58:53 m» mama—So Ba wamwosma .8 mmahco 3330363 a: 733 m" 3 96an
F883» .3 .3. Mme A .3 .3 .3 P3 A .3 .3 .8 95 A .3
3 73m
208-2050 -. H w .3 ...3 .3 -.ow .3 .. H 3 .3 .8 be .3 .3
203-33% :0 .3 93 A .3 flag be 3.3 A .3 _.- .Nm :3 A .3
$333.06 .30 as. ...: .3 3.3 .3 u be .33 .3 .3 P: .3
3:333 ooo . . . ooo . . . ooo . . .
30%. 23. 5835:. m” 35150 85 Womvoamo 8 mmmmca 385389: A? 738 m: 3 £098 86 m: Ooéimam
F883» .3 .3 Moo A .3 .3 .3 P3 A .3 .3 .3 Pmu A .3
E 73m ..3 .3 ...: .3 .3 .3 .3 b» be .3 .mm .3
225.295
205.333 f3 .3 93 A .3 3.3 .3 3.8 A .3 3.3 .3 3.; A .3
55.383 .3 .3 3.3 .8 3.3 .3 P3 A .3 .3 .m» Nam .33
mavwomu ooo . . . ooo . . . ooo . . .
>ma -.3 .3 -.3 .mm -.om .3 -3 .3 .om -.3 .3 3.3 .3
0082.23 -.oo .3 .3 _.oo .3 .om foo .mm .5 .3 #8 .3
max I Zia -.No .3 -.E .3 .3 .3 .8 .3 .5 .NN .mo .3
mania ooo . . . ooo . . . ooo . . .
mSmn I m3? .3 .3 .3 .mA .3 .3 3.3 .3 -.3 .3 -.NA .3
bag ooo . . . ooo . . . ooo . . .
0282 - 55m ..3 .3 -.E .3 -.3 .3 .33 .3 Jon .3 -.3 .mo
Zongm ooo ooo ooo

 

112

.320 3

300:0? 8 Dm~mm~ QNmma 0\ 0.951033 S Sm 303%?» 00%.: 0x $00535 am fiunmxm fi‘mcnmamk Nimamc. Am bmuaxmfi. S wambxm a“ 3 Q

 

 

 

 

 

 

 

 

 

 

 

3mm?
303508. 58:05. 33.0mm
ammo—Em 3 Ema—G 352200 3 «<0me ammo—50 3 2028

33m— Ocm. 33m— 03. 3522 00m. 0222 Ocm. 332 000. 332 00m.

$8822 m8 0022 m8 00,29. m8 0023 m8 0029. m8 0029 m8 30029.
magma mag—.00 mega magma magma magma
0
3030— «a»: van
was .8 Eco .8 3.8 .3 _.oo .: Boo .3 Eco .3 _.oo
3000— 23.. was 2:0 0: 038.283
was .3 _.oo .oe foo .3 foo .2 _.oo .3 foo .3 _.oo
>mm .8 .e3 .3 .8 .8 .e3 .8 .em .8 .e3 .8 .me
00800.23 .3 .3 .8 .3 .3 .3m .8 .8 .8 .ee .3 .33
3002 £30 58820
500820 .8 _.oo .8 foo .8 Eco .3 _.oo .8 E8 .3 Eco
3002 229 50083» 3% 0: 038.208"...

30008020 .8 .ee be foo .8 #8 .3 _.oo .3 _ foo .3 foo
>mm .8 .ee .3 up .8 .8 .3 .3 .8 .e3 .8 .me
0080—23 .3 .3 .oo .3 .oo .3 .oo .3 .3 .8 .3 .8

 

113

Discussion

Despite a recent trend to begin to examine multiple co-occurring symptoms and/or
symptom clusters in oncology nursing (Barsevick et al., 2006; Dodd et al., 2001b; Gift et
al., 2003; Kim et al., 2005; Parker et al., 2005), the majority of research to date has
focused on individual symptoms (Dodd et al., 2001a). In addition, the dimensions of
frequency, intensity and distress of symptoms are typically measured, researched and
intervened upon as they relate to individual symptoms. No randomized clinical trials
(RCT’s) containing a nursing intervention was identified that associated the dimensions
of the frequency, intensity, and distress of one symptom with these same dimensions for
additional symptoms that are co-occurring. No studies were found that examined the
effect of a change in dimension of one symptom being able to predict a change in the
dimensions of other co-occurring symptoms.

This study examined the dimensions of fiequency, intensity and distress for fatigue,
pain and insomnia. All dimensions for fatigue were associated with all of the dimensions
for pain and insomnia at baseline as well as 10 weeks into the trial. Although the
influence of a behavioral cognitive nursing intervention on the management of fatigue as
associated with the other symptoms of pain and insomnia was also of interest, the exact
effect of the intervention was not able to be determined in this study. Recall that study
group (nurse versus non-nurse intervention) and study membership (study number one or
study number 2) were adjusted for in each regression equation. In addition, all subjects,
regardless of group or study membership were included. This study did find that the
management of fatigue categories was able to differentiate non-responders from full-

responders with respect to pain and insomnia experiences. This study also found that

114

21% of the participants in this study were full-responders for fatigue management.
Although this finding supports that some patients experienced decreased severity of
fatigue at 10 weeks versus baseline, due to the regression analysis as well as study design
we can not determine that this effect was due to fatigue intervention or which
intervention or group attributed to this finding.

Pain and Fatigue

Numerous studies have been conducted that support an association between frequency
of pain and frequency of fatigue at baseline of chemotherapy (Cooley et al., 2003; Dodd
et al., 2001b; Gaston-Johansson et al., 1999; Given et al., 2001; Hickok et al., 1996). The
association between fiequency of pain and frequency of fatigue at baseline for
chemotherapy patients was validated in this current study. The addition of covariates
such as age, sex, site and stage of cancer and number of co-morbid conditions at baseline
and 10 weeks did not influence this association.

Hickok et al. (1996) noted that over time, by the sixth week of radiation therapy (RT)
treatment, less then half of their 50 lung cancer patient’s still reporting fatigue reported
pain. Eleven patients completed chemotherapy immediately prior to the start of radiation,
of these patients 54% (n=6) were fatigued throughout RT. Seven patients received
Interferon concurrently with RT, 100% (n=7) of the Interferon patients reported fatigue
during RT. Fifty-four percent (n=1 1)Likewise, Cooley et al. (2003) noted that three
months into therapy only 61% of their 117 lung cancer patients continued to report
fatigue and 33% reported pain. Despite these findings, this study found that 90.3% of all
patients (n=671) reported fatigue at baseline and 50.5% of all patients reported pain at

baseline. At 10 weeks, 75% of all patients continued to report fatigue and 40.1%

115

continued to report pain. This current study supported an association between fi'equency
of fatigue and pain that was not only present at baseline, however, remained statistically
significant at 10 weeks as well.

Age

Given et a1. (2001) found that age was not associated with the frequency of fatigue or
pain in their sample of 841 solid tumor and NHL chemotherapy patients. On the
contrary, Tishelman et al. (2005) found in their sample of 400 lung cancer patients that
age (being younger) negatively influenced the frequency of fatigue in their study. In this
study age did influence the frequency of pain at both baseline and 10 weeks. The lower
one’s age at baseline or 10 weeks the greater the frequency of pain over and above the
effect of fatigue at baseline or 10 weeks.

Sama (1993) and Given et al. (2001) both noted that age did not correlate with
symptom distress and frequency in their studies with lung cancer and solid tumor / NHL
patients. However, Bower et al. (2000), Degner & Sloan (1995), and Tishelman et al.
(2005) did find that age (being younger) negatively influenced fatigue levels throughout
their studies. This current study found that the younger one’s age at baseline and/or 10
weeks the greater the influence over and above the effect of fatigue on the frequency,
intensity and distress of pain. Thus, all dimensions (frequency, intensity and distress) of
pain were affected by age (younger) and the number of co-morbid conditions (higher
number) at baseline and 10 weeks.

To further explore this effect of age and fatigue, separate generalized linear regression
models were run with frequency, intensity, and distress of fatigue at baseline and again at

10 weeks as the dependent variable, study name, group and age as covariates. Age was

116

noted to not statistically contribute to the models for: frequency of fatigue at baseline (B =
-.01, SE = .01, t = -l.18, p = .24) and 10 weeks (B = .02, SE = .01, t = 1.62, p = .11),10
week intensity of fatigue (B = .01, SE = .01, t = .72, p = .48), and 10 week distress of
fatigue (B = .01, SE = .01, t = .16, p = .87). Only the intensity of fatigue at baseline (B = -
.02, SE = .01, t = -2.50, p = .01) and the distress of fatigue at baseline (B = -.03, SE = .01,

= -2.73, p = .01) demonstrated statistically significant negative associations in the
fatigue and age models.

Generalized linear regression models were also run with frequency, intensity and
distress of fatigue at baseline as the dependent variable and age, sex, site and stage of
cancer, number of co-morbid conditions and 10 week fi'equency, intensity or distress of
fatigue as covariates. When the dimensions of fatigue were studied with all covariates,
age was a statistically significant explanatory variable within each dimension; dependent
variable frequency of fatigue (B = -.03, SE = .01, t = -2.46, p = .01), intensity of fatigue
(B = -.04, SE = .01, t = -3.45, p < .01), and distress (B = -.04, SE = .01, t = -3.04, p < .01).
It appears that younger age does have an effect on 10 week fatigue for all dimensions
over and above all baseline fatigue for all dimensions when the covariates of sex, site and
stage of cancer, number of co-morbid conditions are considered. When fatigue is
removed from the regression model the effects of age did not change much, suggesting
the age effect persists without fatigue in the model. Thus, the effect of age noted in this
study in not simply due to the fact that fatigue and age appear together in the models.
The effect of age is unique to the dimensions of pain at baseline and 10 weeks regardless
of the association that fatigue and age have with each other. This current study supported

that younger age over and above the effect of fatigue at baseline influences frequency of

117

fatigue at baseline as well as frequency, intensity, and distress of fatigue at 10 weeks for
chemotherapy patients.

The oncology literature offers numerous studies supporting physiological and
psychological rationale for advancing age to have a negative effect on fatigue and/or pain
(Balducci & Extermann, 2000; Balducci & Yates, 2000; Lipschitz, 1995; Salive,
Comoni-Huntly, Guralnik, Phillips, Wallace, Ostfeld, et al., 1992). However, little
rationale is available to support the findings of this study and others for younger age
influencing fatigue and/or pain. It is proposed that younger patents may have increased
demands for personal and rest time versus their older counterparts such as caring for
children and/or others, employment, and household responsibilities. Discrepancies
between the types of symptoms that are reported or seen as burdensome may exist
between younger and older patients (Demaria & Cohen, 1987; Krech, Davis, Walsh, &
Curtis, 1992). Younger chemotherapy patients may receive greater doses of
chemotherapy based on the perception that the elderly may be more susceptible to
chemotherapy toxicities (Lyman, Dale, Crawford, 2003; Samet, Hunt, Key, Humble, &
Goodwin, 1986; Zelenetz, Reider, & Delgado, 2000). Thus, younger patients may
ultimately experience greater symptoms related to receiving greater doses of
chemotherapy. Chemotherapy dose was not available at the time of analysis for this
dataset to evaluate this further.

Researchers have supported that younger patients may be more likely to participate in
research then older patients (Hutchins, Unger, Crowley, Coltrnan, & Albain, 1999).

However, age was normally distributed in this sample. Refer to Figure 3. The mean age

118

was 57.6 years (SD 11.79), range 25-90, and skewness was -.009. Thus, bias is not
expected.

The effect of age was also statistically significant (with a negative association) for all
dimensions of insomnia with the exception of frequency of insomnia at 10 weeks and
distress of insomnia at 10 weeks. No covariates influenced the models for frequency of
fatigue and frequency of insomnia at 10 weeks. Thus, no covariates, including age or
number of co-morbid conditions affected frequency of insomnia at 10 weeks over and
above the effect of frequency of fatigue. These findings for the effect of age in this
study, as well as in the literature, provide support for firture research in this area. In
addition, rationale for the effect of younger age on the influence of the dimensions of
frequency, intensity, and distress for fatigue, pain and insomnia is in need of further
exploration.

Co-morbid Conditions

The number of co-morbid conditions also influenced the outcome of pain in this study.
An increased number of co-morbid conditions had an effect on frequency, intensity and
distress of pain at baseline and 10 weeks over and above the effect of frequency, intensity
and distress of baseline fatigue. This finding validates the work of numerous researchers
who have reported correlations between number of co-morbid conditions and fiequency,
intensity, and/or distress for pain and fatigue (Bower et al., 2000; Given et al., 2001;
Kurtz et al., 1993; Sama, 1993). Similar to findings in this study, Kurtz et al. (1993),
noted strong correlations between co-morbidity and age and the symptoms of pain,

fatigue, insomnia, and nausea. Kurtz et al. (1993) found that the strongest correlations

119

between number of co-morbid conditions and symptoms were with patients of younger
(age < 60) ages.

The number of co-morbid conditions was significant in the models for pain and
fatigue in this study, validating the work of Given et al. (2001). Given et al. (2001) noted
that patients who reported 3 or more co-morbid conditions were more likely then their
counterparts reporting less then 3 symptoms to experience both pain and fatigue
concurrently during chemotherapy. However, in this current study the number of co-
morbid conditions did not influence the frequency, intensity or distress of insomnia to the
extent that it did pain. The number of co-morbid conditions only produced an effect on
frequency of insomnia over and above the effect of frequency of fatigue at baseline.

The frequency of insomnia and fatigue at baseline was the only dimension / time
period for insomnia that demonstrated an effect for co-morbid conditions. A source of
support for this finding is from the work of Gift et al. (2003). Gifi et al. (2003) found in
their sample of 112 lung cancer patients that the severity score for the cluster of
symptoms that included: fatigue, weakness, weight loss, appetite loss, nausea, vomiting,
and altered taste, declined over time from diagnosis to six months later. Based on these
findings, one may propose that the fi'equency of insomnia will decrease over time. Recall
that the mean frequency for insomnia at baseline was 3.07 (SD = 2.74) and the mean
frequency for insomnia at 10 weeks was 1.62 (SD = 2.41) in this study. Over time (from
baseline to 10 weeks) as the frequency of insomnia decreases, the co-morbid conditions
may loose their influence over and above the effect of fatigue frequency. These
differences in effect of number of co-morbid conditions on fatigue and pain versus

fatigue and insomnia require further investigation in future studies.

120

Insomnia and Fatigue

Numerous studies have been conducted that support an association between insomnia
and fatigue during chemotherapy (Ancoli-Israel, Liu, Marler, Parker, Jones, Sadler et al.,
2006; Bower et al., 2000; Cooley et al., 2003; Curt et al., 2000; Degner & Sloan, 1995;
Wang et al., 2002). This study validated these findings with support that all of the
dimensions (frequency, intensity, and distress) of fatigue and insomnia were associated,
regardless of the addition of covariates such as age, sex, site and stage of cancer and
number of co-morbid conditions at baseline and 10 weeks. This finding is in contrast to
Dodd et al. (2001b) who found only a weak negative correlation between fatigue and
sleep insufficiency (r = -.013) in their solid tumor chemotherapy patients.

As noted with the variables of fatigue and pain above, the younger one’s age is at
baseline the greater the influence over and above the effect of fatigue on the frequency,
intensity, and distress of insomnia. At baseline, younger age was associated with greater
levels of frequency, intensity and distress of insomnia. However, at the 10 week point in
time age had an effect over and above fatigue only for the outcome of intensity of
insomnia. Frequency and distress of fatigue and insomnia were not influenced by age at
10 weeks. Thus, differing findings related to the variable of age between baseline and 10
weeks were noted for insomnia, these findings suggest an opportunity for future research.

Lung Cancer

Differing from the findings for pain and fatigue, the site of cancer demonstrated a
significant association with the frequency of insomnia and fiequency of fatigue at
baseline. This study found that patients that had a diagnosis of lung cancer experienced a

significantly lower frequency of insomnia at baseline compared to patients who did not

121

have lung cancer, over and above the influence of fatigue at baseline, age, and number of
co-morbid conditions. However, numerous studies claim that a diagnosis of lung cancer
results in greater frequency of symptoms versus other cancer diagnosis (Cooley, 2002;
Cooley et al., 2003; Degner & Sloan, 1995; Gifi et al., 2003; Montazeri, Gillis, &
McEwen, 1998). Sama and Brecht, 1997 and Degner and Sloan, 1995 both found that
fatigue and insomnia were the most distressing symptoms reported by lung cancer
patients. Other researchers have noted that 30-50% of patients with lung cancer
experience insomnia (Davidson, MacLean, Brundage, & Schulze, 2002; Rumble, Keefe,
Edinger, Porter & Garst, 2005; Savard & Morin, 2001; Tishelman et al., 2000). Rumble
et al. (2005) noted in their study of 32 early stage lung cancer patients that “early in their
course of therapy” lung cancer patients face emotional distress related to the diagnosis,
possible surgery and/or hospitalizations, sleep disturbing medications and lung cancer
symptoms such as pain or dyspnea that cause sleep disruption.

Due to these unique findings for lung cancer in this study the data was further
explored to determine if differences existed between lung cancer and non-lung cancer
patients. An argument may be posed that a greater number of the non-lung cancer
patients had late stage disease, thus, why they may experience greater insomnia at
baseline over the lung cancer patients. However, crosstab analysis for lung and non-lung
with early and late stage revealed that 9.29% (l 3) of the lung cancer patients in this study
had early stage disease versus 15.81% (83) of the non-lung cancer patients. In addition,
90.71% (127) experienced late stage lung cancer versus 84.19% (442) non-lung cancer
patients with late stage disease. These differences were statistically significant (12 = 3.81,

df = 1, p = .05) and do not support the argument that group differences for stage of

122

disease may have influenced the findings for insomnia; non-lung cancer patients in this
sample were more likely to have early disease then their lung cancer counterparts.

T-tests were also conducted to examine the differences between the groups (lung
versus non-lung) for means of frequency of insomnia at baseline. Lung cancer patients’
insomnia frequency mean value was 2.68 versus non-lung cancer patients’ mean of 3.17.
These mean values were not statistically significantly different (t = -1.89, df = 669, p =
.06). This non-significant difference in mean frequency of insomnia values as well as the
fact that the variable of non-lung cancer was not significantly associated with any other
regression models in this study for all dimensions of pain and insomnia at baseline and 10
weeks encourages one to further explore in future research this finding.

The Dimension of Distress and Insomnia

As noted previously, age and co-morbid conditions appear to influence fatigue, pain,
and insomnia for the dimensions of frequency and intensity at both baseline assessment
and 10 weeks. However, when examining the dimension of distress, age and co-morbid
conditions only influenced fatigue and pain at baseline and 10 weeks; age was the only
covariate that influenced distress of insomnia over and above the effect of fatigue at
baseline. The number of co-morbid conditions did not affect distress of insomnia over
and above baseline fatigue, and age and number of co-morbid conditions did not
influence distress for insomnia over and above fatigue at 10 weeks.

Distress examines how disturbing or bothersome a symptom is perceived to be. Many
researchers have suggested that the concept of distress is fundamentally different from
fiequency and/or intensity (Borj eson, Hursti, Tishelman, Peterson, & Steineck, G., 2002;

Lough, Lindsey, Shinn, & Stotts, 1987; Tishehnan, Degner, & Mueller, B., 2000;

123

Tishelman et al., 2005). Frequency and intensity reflect an incidence rate and severity
ranking, versus, distress depicts “ . . . meanings that the illness holds for an individual
and that these meanings are relative to one’s life” (Tishelman et al., 2005, p. 2014). This
view of the dimension of distress fi'om a meaning perspective differentiates distress fi'om
the dimensions of frequency and intensity that may be considered a direct reflection of
the disease (Tishelman et al., 2005). It may be possible that age and co-morbid
conditions influence one’s meaning associated with pain more so then insomnia. Recall
that age and number of co-morbid conditions influenced insomnia over and above fatigue
only at baseline and not at 10 weeks. This may be indicative that a change takes place in
what influences insomnia over time for patients receiving chemotherapy.
Summary of Findings for Research Questions 1-3

Prior to beginning a discussion of research question. 4 a summary of research
questions 1-3 will be provided. Research Question I asked: At baseline observation,
prior to entry into a clinical trial, and at 10 weeks is there an association between the
dimension of frequency of fatigue and frequency of pain, as well as frequency of fatigue
and frequency of insomnia for adults receiving chemotherapy? Is this association
influenced by; age, site and stage of cancer, sex, and co-morbid conditions? Associations
were found between frequency of fatigue and frequency of pain as well as fi'equency of
fatigue and frequency of insomnia at baseline and 10 weeks for adults receiving
chemotherapy. These associations of frequency of fatigue and frequency of pain were
influenced by age and number of co-morbid conditions at baseline and 10 weeks. The
association between frequency of fatigue and fiequency of insomnia was influenced by

age, number of co-morbid conditions, and the diagnosis of non-lung cancer at baseline.

124

However, no covariates influenced the association of frequency of fatigue and frequency
of insomnia at 10 weeks.

Research Question 2 asked: At baseline observation, prior to entry into a clinical
trial, and at 10 weeks is there an association between the dimension of intensity of fatigue
and intensity of pain, as well as intensity of fatigue and intensity of insomnia for adults
receiving chemotherapy? Is this association influenced by; age, site and stage of cancer,
sex, and co-morbid conditions? Associations were found between intensity of fatigue
and intensity of pain as well as intensity of fatigue and intensity of insomnia at baseline
and 10 weeks for adults receiving chemotherapy. These associations of intensity of
fatigue and intensity of pain were influenced by age and number of co-morbid conditions
at baseline and 10 weeks. The association between intensity of fatigue and intensity of
insomnia was influenced only by age at baseline and 10 weeks. No additional covariates,
including number of co-morbid conditions, influenced the association of intensity of
fatigue and intensity of insomnia at baseline or 10 weeks.

Research Question 3 asked: At baseline observation, prior to entry into a clinical trial,
and at 10 weeks is there an association between the dimension of distress of fatigue and
distress of pain, as well as distress of fatigue and distress of insomnia for adults receiving
chemotherapy? Is this association influenced by; age, site and stage of cancer, sex, and
comorbid conditions? Associations were found between distress of fatigue and distress
of pain as well as distress of fatigue and distress of insomnia at baseline and 10 weeks for
adults receiving chemotherapy. These associations of distress of fatigue and distress of
pain were influenced by age and number of co-morbid conditions at baseline and 10

weeks. The association between distress of fatigue and distress of insomnia was

125

influenced only by age at baseline. No additional covariates influenced the association of
distress of fatigue and distress of insomnia at baseline. No covariates, including age or
number of co-morbid conditions influenced the association of distress of fatigue and
distress of insomnia at 10 weeks.
Findings for Research Question 4

Research Question 4 asked: Using categories of response to fatigue management
(none / mild, non-response, partial response and full response) established in the
literature, determine how changes in the dimensions of frequency, intensity, and distress
of pain and insomnia are predicted by categories of response to the management of
fatigue between baseline and 10 weeks. Is this association influenced by; age, site and
stage of cancer, sex, and co-morbid conditions?
Frequency of Pain

This study found that there is an association between the dimensions of frequency of
pain at 10 weeks and management of fatigue categories after adjusting for frequency of
pain at baseline. The only fatigue management category that produced statistically
significant findings was the non-responders. Therefore, non-responders (patients who
stay within the same category [excluding the “none” or “mild” category] fiom the
baseline measure to the 10 week measure) when compared to full responders (patients
who go to the none or mild category of fatigue from moderate or severe between baseline
and 10 weeks) experience significantly increased frequency of pain at 10 weeks.

The number of co-morbid conditions was significant in the regression model. Thus, at
10 weeks the number of co-morbid conditions influenced frequency of pain over and

above the effect of fatigue management. Unlike findings for research questions 1-3, age

126

was not found to affect pain at 10 weeks when taking into account fatigue management
categories.
Intensity of Pain

An association also existed between the dimensions of intensity of pain at 10 weeks
and intensity of pain at baseline when management of fatigue categories was studied.
Two fatigue management categories produced statistically significant findings for the
dimension of intensity and pain, the non-responders and the partial responders.
Therefore, non-responders and partial responders (a patient who goes fi'om severe to the
moderate level of fatigue between baseline and 10 weeks) when compared to full
responders experience significantly increased intensity of pain at 10 weeks.

The association between intensity of pain and the effect of the fatigue management
categories was influenced by the number of co-morbid conditions and the age. Thus, at
10 weeks the number of co-morbid conditions as well as younger age influenced intensity
of pain at 10 weeks over and above the effect of fatigue management.

Distress of Pain

An association also existed between the dimensions of distress of pain at 10 weeks
and distress of pain at baseline with the consideration of management of fatigue
categories. Only one fatigue management category, the non-responders, produced a
statistically significant finding for the dimension of distress and pain. Therefore, non-
responders when compared to full responders experience significantly increased
frequency of pain at 10 weeks.

The association between distress of pain and the effect of the fatigue management

categories was also influenced by the number of co-morbid conditions and age. Thus, at

127

10 weeks the number of co-morbid conditions as well as younger age influenced distress
of pain at 10 weeks over and above the effect of fatigue management.
Frequency of Insomnia

This study also found that when fatigue categories are examined there is an
association between the dimensions of frequency of insomnia at 10 weeks and fiequency
of insomnia at baseline. Similar to the findings noted for frequency of pain, the only
fatigue management category that produced statistically significant findings was the non-
responders. Non-responders when compared to full responders experience significantly
increased frequency of insomnia at 10 weeks. Unlike pain, the association between
frequency of insomnia and the effect of the fatigue management categories was not
influenced by any of the covariates (age, sex, site and stage of cancer, and number of co-
morbid conditions). Thus, the number of co-morbid conditions or age did not influence
frequency of insomnia at 10 weeks over and above the effect of fatigue management.
Intensity of Insomnia

An association also existed between the dimensions of intensity of insomnia at 10
weeks and intensity of insomnia at baseline when management of fatigue categories was
considered. Again, similar to the findings for intensity of pain, two fatigue management
categories produced statistically significant findings for the dimension of intensity of
insomnia, the non-responders and the partial responders. Non-responders and partial
responders when compared to full responders experience significantly increased intensity
of insomnia at 10 weeks. As noted with the dimension of frequency of insomnia, the
association between intensity of insomnia and the effect of the fatigue management

categories was also not influenced by any of the covariates.

128

Distress of Insomnia

An association also existed between the dimensions of distress of insomnia at 10
weeks and distress of insomnia at baseline when management of fatigue categories was
involved. As found with intensity of insomnia, two fatigue management categories
produced statistically significant findings for the dimension of distress and insomnia; the
non-responders and the partial responders. Non-responders and partial responders when
compared to full responders experience a significantly increased distress of insomnia at
10 weeks. As noted with the dimensions of frequency and intensity of insomnia, the
association between distress of insomnia and the effect of the fatigue management
categories were not influenced by any of the covariates.

Summary of Findings for Research Question 4

To summarize, Research Question 4 asked: Using categories of response to fatigue
management (none / mild, non-response, partial response and fiill response) established
in the literature, determine how changes in the dimensions of frequency, intensity, and
distress of pain and insomnia are predicted by categories of response to the management
of fatigue between baseline and 10 weeks? Is this association influenced by; age, site and
stage of cancer, sex, and co-morbid conditions?

Frequency, intensity and distress of pain can be predicted by category of response to
fatigue management. Non-responders when compared to full responders experience
increased frequency, intensity and distress of pain at 10 weeks. In addition, partial
responders as well as non-responders when compared to full responders experience
increased intensity of pain at 10 weeks. These associations are influenced by an

increased number of co-morbid conditions. Age also influenced the associations between

129

partial and non-responders with full responders, however, only for pain intensity and
distress at 10 weeks.

Frequency, intensity and distress of insomnia can also be predicted by category of
response to fatigue management. Non-responders when compared to full responders
experience increased frequency, intensity and distress of insomnia at 10 weeks. In
addition, partial responders as well as non-responders when compared to full responders
experience increased intensity and distress of insomnia at 10 weeks. These associations
were not influenced by age, number of co-morbid conditions, sex, cancer site or stage.

In this study with no standard care control group and with study group membership
controlled for we can conclude that the fatigue management categories provide an
appropriate means of measming the broad effectiveness of the cognitive behavioral
intervention. This analysis was not designed to compare the different interventions.
However, overall the intervention components of both studies (ATSM or NASM) appear
to have demonstrated some success for fatigue management reflected by the counts and
percents for the response to fatigue management categories cited above.

One hundred and seven patients (20.08%) were considered full responders for fatigue
management, an additional 79 (14.82%) were partial responders. Mean fatigue scores fell
from baseline to 10 weeks for frequency (4.70 [SD= 2.49] to 3.68 [SD=2.85]), intensity
(4.71 [SD= 2.69] to 3.22 [SD=2.67]), and distress (4.18 [SD= 3.06] to 2.52 [SD=2.78]).
This finding supports past work by Given et al. (2001) that found in their cognitive
behavioral intervention study with a standard of care control group (n= 841) that fewer
patients in the experimental group reported pain and fatigue at 20 weeks versus the

control group. Similarly, Savard, Sinard, Ivers, & Morin, (2005) were able to

130

demonstrate that a cognitive behavioral nursing intervention positively impacted breast
cancer patient responses to insomnia.
Study Limitations

Although intervention evaluation was not proposed for this study it did examine
response to fatigue management. It is assumed that patients in the partial response and
full response fatigue management categories did benefit from one of the interventions in
the two studies that made up this dataset. A lack of ability to differentiate which
intervention was received as well as the lack of a control group in this study limits the
ability to advise practice and research on strategies that contribute to fatigue
management. Sikorskii et al. (in press) also point out that lack of a control group
provides difficulty in comparing symptom dimension changes over baseline. Reduction
in symptom frequency, intensity, and/or distress over baseline may be attributed to
response shift or regression to the mean, however, response shifi would be expected to
occur in both groups and would not bias the analyses of between group differences
(Sikorskii et al., in press).

Samples differed due to the fact that all patients experienced interviews and only some
patients experienced intervention contacts (Skorskii et al., in press). Only patients
randomized to the nursing behavioral cognitive intervention arms of both studies could
have received interventions for fatigue, pain, and/or insomnia that were beyond the scope
of the toolkit. In addition, this study did not separate patients receiving interventions for
fatigue, pain and insomnia from patients who did not receive such intervention. All

symptom assessment in this dataset was self-report, per the patient’s perception, no

131

comparisons were made with practitioner assessment via the intervention recorded data
or medical record audit.

This study was conducted as a secondary data analysis. Stage of disease was
established a priori to be a categorical variable as early (TNM 1 or 2) and late (TNM 3 or
4). Stage III and IV cancer patients have very different clinical presentation and typically
receive very different chemotherapy regimens. Stage IV patients for all diagnosis will
have metastatic disease; the extent of their disease will also vary greatly and effect
symptoms in different ways. This sample also contained only 10% non-Caucasian,
although reflective of the geography used for recruiting, this provides a limitation for
generalizability to minority populations. Some authors have reported that minorities may
be more likely then Caucasians to present with later stages of disease and also may have
difficulty accessing cancer care centers (Benjamin, Reddy, & Brawley, 2003; Gadgeel &
Kalemkerian, 2003). Practitioners will need to be aware of these limitations and future
studies will need to address the limitations cited above. Additional implications for
clinical practice and future studies will follow.

Implications for Clinical Practice

Data fiom this study provides several implications for clinical practice. It is due to the
multiple nature of symptoms that people seek health care (Cleeland et al., 2000; Rutledge
& McGuire, 2003). Three of the most common co-occurring symptoms for patients
experiencing chemotherapy are fatigue, pain and insomnia. Patients will experience
varying degrees of differing dimensions (frequency, intensity, and distress) of each
symptom over time. As noted in this current study, each dimension depicts a separate

assessment area that is uniquely associated with fatigue, pain, and/or insomnia.

132

Frequency is assessed to quantify the presence of the symptom, intensity to note the
severity, and distress to represent how bothersome the symptom is to the patient. This
study also supported the association of these dimensions over time (baseline to 10
weeks).

In the ideal world of clinical practice the nurse would have adequate time and
sophisticated computerized documentation systems that would allow for thorough
assessments and documentation of all three dimensions for all symptoms. Unfortunately,
this is not the case, the question then becomes: If you have limited time and resources
and the patient presents with fatigue, pain and insomnia, what dimension of pain and
insomnia assessment is absolutely necessary?

This study found that frequency, intensity and distress of pain when associated with
fatigue, were all influenced by younger age and number of co-morbid conditions at both
baseline and 10 weeks. Associations existed over time between each dimension of pain
and fatigue at baseline and 10 weeks. In addition, when examining fatigue response
categories, the non-responders were noted to be different from the full-responders for all
pain and fatigue dimensions at 10 weeks. Thus, similar findings were noted for all
dimensions of pain and fatigue. It may not be necessary to assess each dimension for
pain when being assessed with fatigue at baseline or 10 weeks.

Naturally, the next question is what dimension should you assess for pain if you only
have the time to assess one? As stated above all findings for pain and fatigue were
similar. However, when examining fatigue management categories, the regression model
that included the intensity dimension of pain was the only model that was able to support

non-responders as well as partial responders in demonstrating increased fatigue as

133

compared to full responders. In addition, t-tests of the differences among adjusted means
(adjusted for other variables in the model) were conducted with intensity of pain as the
dependent variable. With intensity of pain as the dependent variable it was noted that the
non-response fatigue management category is different fi'om the none / mild category and
the full responder category. The none / mild category was also noted to be different fiorn
the partial response category. Thus, the intensity of pain dimension assessment is as
likely as any other dimension to be influenced by age and co-morbid conditions,
demonstrates an association between baseline and 10 weeks, and demonstrates
differences within the fatigue management categories. Based on the above findings, the
intensity dimension is the dimension to focus the pain assessment on if time and
resources do not allow for a complete assessment of all dimensions.

The findings for insomnia were not as consistent. Associations did exist between
each dimension of insomnia and fatigue at baseline and 10 weeks. However, frequency,
intensity and distress of insomnia, when associated with fatigue, were not all influenced
by younger age and number of co-morbid conditions at either baseline or 10 weeks. For
example, age and co-morbids did influence frequency of insomnia over and above the
effect of fatigue at baseline. The lung cancer group also demonstrated a negative
association in the frequency of insomnia model at baseline; this effect was not noted for
any other dimension or for pain. No covariates influenced the frequency of insomnia at
10 weeks. No covariates influence any dimension of insomnia in the fatigue management
category models. Models that included fatigue management categories for both the
dimensions of intensity and distress demonstrated differences between non responders

and partial responders when compared to full responders. Due to these differences noted

134

in the dimensions of insomnia it is not possible to prioritize a specific dimension to
assess.

Knowing that associations exist between these dimensions and between the symptoms
of fatigue, pain and insomnia, clinical interventions must be targeted toward the
appropriate dimensions of each symptom. Interventions must also accommodate the co-
occurrence of symptoms such as fatigue and pain and/ or insomnia. Due to the
associations noted in this study between fatigue and pain and fatigue and insomnia it is
essential to treat both symptoms concurrently.

Interventions for symptoms must also occur over time. As noted in the Yates et al,
2005 and the Anderson et al, 2006 studies behavioral cognitive nursing interventions may
loose their effect over time. Both of these studies demonstrated a need for booster
intervention sessions when addressing the symptoms of fatigue or pain. Cancer
treatments are given over time, some symptoms may be most severe immediately
following chemotherapy (such as vomiting) others may increase, stay the same or
decrease as time goes by. Interventions must be provided over time and stepped to meet
the changing needs of the patient. Assessment must continue over time as well do to the
“wax and wane” effect of symptoms. Symptoms thought to be under control with a past
intervention may re-appear at a later time during or even after treatment. Some
symptoms, such as fatigue, may persist well after chemotherapy is completed.
Continuous follow up and monitoring of symptoms over time for the chemotherapy
patient is essential.

As noted in this study, the use of fatigue management categories may be an

appropriate technique to evaluate the effectiveness of interventions aimed toward single

135

or multiple symptoms. Intensity of fatigue is typically measured on a scale of 0-10. A
reduction in symptom intensity that ranges from 33% to 50% is considered clinically
significant (Guyatt et al. 2002). However, when using a 0-10 scale to measure fatigue
intensity both percent change and absolute value change values are not always clinically
meaningful. The relationship between the intensity of a symptom and distress of a
symptom is not linear across the scale (Given et al., in press). Differences in the
associations between intensity and distress appear on the 0-10 scales between 1 and 2 and
4 and 5. Therefore, the use of cut points (0 = none, 1 = mild fatigue, 2-4 moderate
fatigue, 5-10 severe fatigue) within the 0-10 intensity of fatigue scale may produce a
more clinically meaningful fatigue assessment. It is also important to note that thorough
assessment and documentation are required to be able to retrospectively calculate fatigue
management categories that can be used for quality assurance monitoring and/or research
purposes.

In order to provide appropriate assessment and intervention it is also essential to be
aware of the covariates that influence fatigue, pain, and insomnia. For all dimensions of
pain and insomnia, with the exception of 10 week fiequency of insomnia and 10 week
distress of insomnia, younger age enhanced the dimension of the outcome of pain or
insomnia over and above the effect of fatigue. An increased number of co-morbid
conditions also enhanced the fiequency, intensity and distress of pain at both baseline and
10 weeks over and above the effect of fatigue. Therefore, practitioners need to be aware
that chemotherapy patients who are younger in age and those with co-morbid conditions
may be considered at greater risk of experiencing enhanced frequency, intensity and

distress of fatigue with pain and/or insomnia.

136

Implications for Research

As cited above, this current study noted a significant affect for younger age on the
dimensions of pain and insomnia over and above the effect of fatigue with the exception
of 10 week fi'equency and distress of insomnia. The negative influence of younger age
on symptom frequency, intensity and distress is just beginning to appear in the literature
(Bower et al., 2000; Tishelman et al., 2005). Future research as to the biological,
physiological, and psychological rationale for this phenomena is required. Likewise,
various assumptions made regarding ageing and increased symptom frequency, intensity
and distress of symptoms associated with chemotherapy are now challenged. Age as a
risk factor for increased fi'equency, intensity and distress of fatigue, pain, and insomnia
when these symptoms co—occur in chemotherapy patients should be examined in future
research studies.

The number of co-morbid conditions also influenced all dimensions of pain over and
above the effect of fatigue at baseline and 10 weeks as well as fi‘equency of insomnia at
baseline over and above the effect of fatigue at baseline. Due to the fact that age and
number of co-morbid conditions influenced fatigue and insomnia over and above fatigue
only at baseline and not at 10 weeks may be reason to believe that a change takes place in
what influences the association between fatigue and insomnia over time. Time plays a
key role in symptom presentation as well as management for patients receiving multiple
cycles of chemotherapy. This unique finding that co-morbid conditions effect pain but
not insomnia and the role that time may play in this scenario requires future research.

Additional research related to specific co-morbid conditions (i.e., diabetes, high blood

137

pressure) as well as their clinical history (onset, duration, management, etc.) that are most
likely to result in increased fatigue, pain and or insomnia is also needed.

This current study found that cancer site was not significant in any of the regression
models examining fatigue with pain and insomnia with the exception of the non-lung
cancer category effecting frequency of insomnia at baseline over and above the effect of
fatigue. This unique effect for patients without a diagnosis of lung cancer as compared to
those who have lung cancer on insomnia and not pain, as well as at baseline and not 10-
weeks is contradictory to the majority of published research on lung cancer and the
symptoms of fatigue, pain, and insomnia (Cooley et al., 2003; DeMaria & Cohen, 1987;
Given et al., 2001; Hickok et al., 1996; Kurtz et al., 2000). Thus, continued research on
the differences noted with a lung cancer diagnosis (as well as stage and treatment
specifics) on pain versus insomnia when they co-occur with fatigue is needed.

Recall that this effect for non-lung cancer in this study was also only noted at baseline
for fi'equency of insomnia. Based on this finding, the effect of time on the symptom of
insomnia for the lung cancer chemotherapy patient is of interest for future research. An
appropriate research question to examine time and insomnia would include baseline and
multiple measures of various dimensions (frequency, intensity and distress) during
therapy and following the conclusion of chemotherapy. As noted from this study,
covariates such as age, number of co-morbid conditions would need to be controlled. In
addition, fi'om the literature, factors such as surgery, anxiety level, quality of life,
depression, stage, medications, and presence of pain would need to be controlled for to
examine insomnia in the lung cancer versus non-lung cancer patients (Cooley et al.,

2002; Rumble et al., 2005).

138

This study also demonstrated that frequency, intensity and distress of pain and
insomnia can be predicted by categories of response to fatigue management. This
provides an important validation of the appropriateness of the use of fatigue management
categories in clinical research for this area of study that is in its’ infancy (Given et al., in
press; Mendoza et al., 1999; Miaskowski et al., in press). Although this study was not
designed to evaluate the effectiveness of individual cognitive behavioral interventions,
significant differences were noted between the non-responders and full-responders for
each dimension (frequency, intensity and distress) of pain and insomnia at 10 weeks.
Thus, non-responders experienced significantly increased frequency, intensity and
distress for both pain and insomnia at 10 weeks when compared to the full responders.
This current study does support the appropriateness of the use of fatigue management
categories as a new research technique. Future studies should be directed toward use of
these categories to compare interventions against each other in randomized clinical trials.
The implementation of randomized control studies using stepped interventions over time
that address solo as well as multiple co-occurring symptoms will be needed to advance
research as well as clinical practice.

Conclusion

The purpose of this study was to examine the dimensions of frequency, intensity, and
distress of the co-occurring symptoms of fatigue, pain and insomnia as they occur at two
different data collection points in two randomized clinical trials of a cognitive behavioral
intervention. This study did support that at baseline as well as 10 weeks there is an
association between the dimensions of frequency, intensity and distress of fatigue and the

same dimensions for pain and insomnia.

139

In addition, categories of response to fatigue management were capable of predicting
changes in the dimensions of frequency, intensity, and distress of pain and insomnia at 10
weeks. For all dimensions of pain and insomnia, with the exception of 10 week
frequency of insomnia and 10 week distress of insomnia, younger age enhanced the
dimension of the outcome of pain or insomnia over and above the effect of fatigue. An
increased number of co-morbid conditions also enhanced the frequency, intensity and
distress of pain at both baseline and 10 weeks over and above the effect of fatigue.
However, an increased number of co-morbid conditions only influenced the frequency
dimensions of insomnia at baseline over and above the effect of fatigue at baseline.

Further studies to examine relationships between covariates such as age and number
of co-morbid conditions that influence fatigue combined with pain and/or insomnia over
time are recommended. The influence of time on the dimensions of fi'equency, intensity,
and distress for co-occurring symptoms associated with chemotherapy is strongly
encouraged. The Adapted SEM (see Figure 2) may be used in the future to guide
intervention research examining multiple co-occurring symptoms over time. The use of
fatigue management categories as a means of evaluating the effect of cognitive
behavioral interventions with co-occurring symptoms was also supported in this research.
Other researchers are encouraged to examine these variables cited above as a part of RCT
intervention studies for symptom cluster research.

An immediate next step for this research project is to continue to use the fatigue
management categories to determine if differences exist in the effectiveness of the
various cognitive behavioral interventions noted within the study. Future research with

this dataset will continue to evaluate the influence of time and all dimensions of symptom

140

presentation to continue to advance this evolving field of study of co-occurring symptoms

in oncology practice.

141

APPENDIX A

Study Schema

(See next page)

142

 

Follow up with pt. for questions/seek decision on participation.

 

 

 

V

 

l

 

 

Pt. decides to participate

 

 

 

Pt. decides NOT to participate —continues

to receive standard care per provider

 

 

 

 

 

Pt. completes 2 x weekly automated calls for 6 weeks assessing symptoms

 

 

i

l
1

 

 

Pt. meets eligibility criteria.

 

 

 

Pt. does not meet eligibility criteria —
continues to receive standard of care

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

f l
Study 1 — NASM Study 2 — ATSM
v i
Patient/caregiver completes lSt Patient completes 1St telephone
telephone survey, receives Symptom survey; receives Symptom
Management Toolkit and Medical Management Toolkit and Medical
Visit Log. Visit Log.

v t f v
Patient randomly Patient randomly Patient Patient randomly
placed in Patient placed in S_el[ randomly placed in S_el[
Assisted Management placed in Management
Intervention Intervention Patient intervention
gzoug - 6 calls up — 6 calls Assisted gzgug — 6 calls
over 8 weeks over 8 weeks Intervention over 8 weeks
with a nurse. with a research my — 6 calls with an

assistant. over 8 weeks automated
with a nurse. system.
I l L

 

 

 

Pt. completes 2"d survey & receives
evaluation survey for Symptom
Management calls/Toolkit

 

 

 

Pt. completes 2"d survey & receives
evaluation survey for Symptom
Management calls/Toolkit

 

 

 

 

 

Pt. completes 3rd survey

I

 

 

 

 

i

Pt. completes 3lrd survey

 

v

 

 

Pt.’s medical record reviewed;
medical charge data is requested

 

 

Patient’s medical record is reviewed;
medical charge data is requested

 

143

 

 

REFERENCES

Aaronson, L. 8., Tee], C.S., Cassmeyer, V., Neuberger, G.B., Pallikkathayil, L. & Pierce,
J. (1999). Defining and Measuring Fatigue. Image: Journal of Nursing
Scholarship, 31(1), 45-50.

American Cancer Society. (2006). Cancer Facts and Figures. American Cancer
Society, Atlanta: GA.

Ancoli-Israel, S., Liu, L., Marler, M.R., Parker, B.A., Jones, V., Sadler, G.R.et al.
(2006). Fatigue, sleep and circadian rhythms prior to chemotherapy for breast cancer.
Supportive Care In Cancer, 14(3), 201-209.

Anderson, K.O., Cohen, M.Z., Mendoza, T.R., Guo, H., Harle, M.T., & Cleeland, C.
(2006). Brief cognitive-behavioral audiotape interventions for cancer related pain:
Immediate but not long-term effectiveness Cancer, 10(1), 207-214.

Armstrong, T. S. (2003). Symptoms Experience: A Concept Analysis. Oncology
Nursing Forum, 30 (4), 601-606.

Armstrong, T. S., Cohen, M.Z., Eriksen, L.R., & Hickey, J .V. (2004). Symptom clusters
in oncology patients and implications for symptom research in people with primary
brain tumors. Journal of Nursing Scholarship, 36(3), 197-206.

Ashbury, F. D., Findlay, H., Reynolds, B. & McKerracher, K. (1998). A Canadian
Survey of Cancer Patients Experiences: Are their needs being met? Journal of Pain
and symptom Management, 16(5), 298-306.

Balducci, L., & Extermann, M. (2000). Management of cancer in the older person: A
practical approach. Oncologist, 5, 224-237.

Balducci, L., & Yates, J. (2000). General guidelines for the management of older
patients with cancer. Oncology, 14, 221-227.

Barsevick, A.M., Whitmer, K., Nail, L.M., Beck, S.L., & Dudley, W.N. (2006).
Symptom cluster research: Conceptual design, measurement, and analysis issues.
Journal of Pain and Symptom Management, 31(1), 85-95.

Beck, S.L., & Schwartz, A.L. (2000). Unrelieved pain contributes to fatigue and
insomnia [abstract]. Oncology Nursing Forum, 27(2), 350-351.

Bender, C.M.., Ergyn, F.S., Rosenzweig, M.Q., Cohen, S.M., & Sereika, SM. (2005).

Symptom clusters in breast cancer across 3 phases of the disease. Cancer Nursing, 28
(3), 219-225.

144

Benjamin, M., Reddy, S., & Brawley, O. (2003). Myeloma and race: A review of the
literature. Cancer and Metastasis Reviews, 22(1), 87-93.

Berger, P. J ., McCutcheon, L., Soust, M., Walker, A.M., & Wilkinson, M.H. (1991).
Electronmyographic changes in the isolated rat diaphragm during the development of
fatigue. European Journal of Applied Physiology, 62(5), 310-316.

Berger, A. M. & Walker-Nobel, S. (2001). An explanatory model of fatigue in women
receiving adjuvant breast cancer chemotherapy. Nursing Research, 50(1), 42-52.

Bemdt, E., Kallich, J ., McDermott, A., Xu, X., Lee, H., & Glaspy, J. (2005). Reductions
in anemia and fatigue are associated with improvements in productivity in cancer
patients receiving chemotherapy. Pharmacoeconomics, 23(5), 505-514.

Blesch, K.S., Paice, J .A., Wickham, R., Harte, N., Schnoor, D. Paul, S. et al. (1991).
Correlates of fatigue in people with breast or lung cancer. Oncology Nursing Forum,
18(1), 81-87.

Borjeson, S., Hursti, T.J., Tishelman, C., Peterson, C., & Steineck, G. (2002). Treatment
of nausea and emesis during cancer chemotherapy: Discrepancies between antiemetic
effect and well-being. Journal of Pain and symptom Management, 24, 345-358.

Bower, J .E. (2005). Prevalence and causes of fatigue after cancer treatment: The next
generation of research. Journal of Clinical Oncology, 23(33), 8280-8282.

Bower, J .E., Ganz, P.A., Desmond, K.A., Rowland, J .H., Meyerowitz, B.E., & Belin,
T.R (2000). Fatigue in breast cancer survivors: Occurrence, correlates, and impact
on quality of life. Journal of Clinical Oncology, 18(4), 743- 753.

Breitbart, W., Rosenfeld, B., Kaim, M. & Funesti-Esch, J. (2001). A randomized, double-
blind, placebo-controlled trial of psychostimulants for the treatment of fatigue in

ambulatory patients with Human Immunodeficiency Virus disease. Archives of
Internal Medicine, 161(3), 41 1-420.

Cella, D. (1997). The functional assessment of cancer therapy-anemia (FACT-An)
scale: A new tool for the assessment of outcomes in cancer anemia and fatigue.
Seminars in Hematology, 34 (suppl. 2), 13-19.

Cella, D. F ., & Cherin, EA. (1988). Quality of life during and after cancer treatment.
Comprehensive Therapy, 14(5), 69-75.

Cella, D., Davis, K., Breitbart, W. & Curt, G. (2001). Cancer-related fatigue:

Prevalence of proposed diagnostic criteria in a United States sample of cancer
survivors. Journal of Clinical Oncology, 19 (14), 3385-3391.

145

Cella, D., Lai, J .S., Chang,C.H., Peterman, A., & Slavin, M. (2002). Fatigue in cancer
patients compared with fatigue in the general United States population. Cancer, 94
(2), 528-538.

Cella, D.F., Tulsky, D.S., Gray, G., Sarafian, B., Linn, E., Bonomi, A., et al. (1993). The
functional assessment of cancer therapy scale: Development and validation of the
general measure. Journal of Clinical Oncology, 11 (3), 570-579.

Cleare, A. (2003). The Neuroendocrinology of Chronic Fatigue Syndrome. Endocrine
Reviews, 24(2), 236-252.

Cleeland, C. S., Bennet, G.J., Dantzer, R, Dougherty, P. M., Dun, A.J., Meyers, C.A., et
al. (2003). Are symptoms of cancer and cancer treatment due to a shared biological
mechanism? Cancer, 97(11), 2919-2925.

Cleeland, C. S., Mendoza, T.R., Wang, X.S., Chou, C., Harle, M.T., Morrissey, M., &
Engstrom, MC. (2000). Assessing symptom distress in cancer patients: The MD.
Anderson Symptom Inventory. Cancer, 89(7), 1634-1646.

Cooley, M. E. (2002). Patterns of symptom distress in adults receiving treatment for lung
cancer. Journal of Palliative Care, 18(3), 694-708.

Cooley, M., Short, T.H. & Moriarty, H.J. (2003). Symptom prevalence, distress, and
change over time in adults receiving treatment for lung cancer. Psycho-Oncology,
12(7), 694-708.

Curt, G.A., Breitbart, W., Cella, D., Groopman, J .E., Homing, S.J., Itri, L.M., et al.
(2000). Impact of cancer-related fatigue on the lives of patients: New findings fiom
the fatigue coalition. The Oncologist, 5, 353-360.

Daut, R.L., Cleeland, C.S., & Flanery, RC. (1983). Development of the Wisconsin brief
pain questionnaire to assess pain in cancer and other diseases. Pain, 1 7(2), 197-210.

Davidson, J ., MacLean, A., Brundage, M., & Schulze, K. (2002). Sleep disturbance in
cancer patients. Social Science Medicine, 54, 1309-1321.

Degner, L., & Sloan, J. (1995). Symptom distress in newly diagnosed ambulatory
cancer patients and as a predictor of survival in lung cancer. Journal of Pain and
symptom Management, 10 (1 6), 423-431.

Demaria, L.C., & Cohen, H.J. (1987). Characteristics of lung cancer in elderly patients.
Journal of Gerontology, 42(5), 540-545.

Dodd, M., Janson, S., Facione, N., Faucett, J ., Froelicher, E.S., Humphreys, J., et al.,

(2001 a). Advancing the science of symptom management. Journal of A dvanced
Nursing, 33(5), 668-676.

146

Dodd, M. J ., Miaskowski, C., & Paul, S.M. (2001b). Symptom Clusters and their effect
on the functional status of patients with cancer. Oncology Nursing Forum, 28(3), 465-
470.

Ferrell, B., Grant, G., Funk, B.M, Otis-Green, S.A., & Garcia, NJ. (1998). Quality of
life in breast cancer survivors: Implications for developing support services.
Oncology Nursing Forum, 25(5), 887-895.

Fu, J .B., Kau, T.Y., Severson, RK., & Kalemkerian, GP. (2005). Lung cancer in
women: Analysis of the National Surveillance, Epidemiology, and End Results
Database. Chest, 127(3), 768-777.

Gadgeel, S., & Kalemkerian, G. (2003). Racial differences in lung cancer. Cancer and
Metastasis Reviews, 22(1), 39-46.

Gaston-Johansson, F., Fall-Dickson, J .M., Bakos, A.B., & Kennedy, M.J. (1999).
Fatigue, pain, and depression in pre-autotransplant breast cancer patients. Cancer
Practice, 7(5), 240-247.

Gift, A.G., Stommel, M., Jablonski, A., & Given, W. (2003). A cluster of symptoms
over time in patients with lung cancer. Nursing Research, 52(6), 393-400.

Given, C., Given B., Jeon, S., & Sikorskii, A. (in press). Development of symptom
response categories to test a psycho-behavioral intervention: Incorporating clinical
interpretation into statistical tests. Journal of Pain and Symptom Management.

Given, C., Given, B., Azzouz, F., Kozacik, S. & Stommel, M. (2001). Predictors of pain
and fatigue in the year following diagnosis among elderly cancer patients. Journal of
Pain and Symptom Management, 21(6), 456-466.

Given, C. W., Given, B.A., Azzouz, F., Stommel, M., & Kozachik, S. (2000).
Comparison of changes in physical functioning of elderly patients with new diagnoses
of cancer. Medical Care, 38(5), 482- 493.

Given, B., Given, C., McCorkle, R., Kozacik, S., Cimprich, B., Rahbar, M.H., &
Woj cik,C. (2002). Pain and fatigue management: Results of a nursing randomized
clinical trial. Oncology Nursing Forum, 29(6), 949-956.

Given, B.A., & Sherwood, RR. (2005). Nursing-Sensitive patient outcomes: A white
paper. Oncology Nursing Forum, 32(4), 773-784.

Guyatt, G.H., Norman, G.R., Juniper, E.F., & Griffith, LE. (2002). A critical look at
transition ratings. Journal of Clinical Epidemiology, 55(9), 900-908.

Henly, S. J ., Kallas, K.D., Klatt, C.M., & Swenson, K.K. (2003). The notion of time in
symptom experiences. Nursing Research, 52(6), 410-417.

147

Hickok, J. T., Morrow, G.R., McDonald, 8., & Bellg, A.J. ( 1996). Fatigue in lung cancer
patients receiving radiation therapy: Implications for management. Journal of Pain
and Symptom Management, 11(6), 370-377.

Hutchins, L.F., Unger, J .M., Crowley, J .J ., Coltrnan, C.A., & Albain, KS. (1999).
Underrepresentation of patients 65 years of age or older in cancer-treatment trials.
New England Journal of Medicine, 34 I (27), 2061-2067.

Irvine, D., Vincent, L., Graydon, J ., Bubella, N., & Thompson, L. (1994). The
prevalence and correlates of fatigue in patients receiving treatment with chemotherapy
and radiation therapy: A comparison with the fatigue experienced by healthy
individuals. Cancer Nursing, 17(5), 367-378.

Jemal, A., Siegel, R., Ward, E., Murray, T., Xu, J ., & Thun, M.J. (2007). Cancer
statistics, 2007. Cancer, 57(1), 43- 66.

Keehne-Miron, J ., Given, B., Given, C., vonEye, A., & Doorenbos, A. (2005, October).
The prediction of symptom severi for breast and lung cancer chemotherapy patients.
Poster session presented at the 38 Annual Great Lakes Cancer Nursing Research
Nursing Conference, Novi, MI.

Kim, H.J., McGuire, D.B., Tulman, L., & Barsevick, A.M. (2005). Symptom clusters:
Concept analysis and clinical implications for cancer nursing. Cancer Nursing, 28(4),
270-282.

Krech, RI., Davis, J., Walsh, D., & Curtis, EB. (1992). Symptoms of lung cancer.
Palliative Medicine, 6(5), 309-315.

Kroenke, K. (2001). Studying symptoms: Sampling and measurement issues. Annals of
Internal Medicine, 134(9), 844-853.

Kurtz, M. E., Kurtz, J .C., Given, C.W., & Given, BA. (1993). Loss of physical
fimctioning among patients with cancer: A longitudinal view. Cancer Practice, 1(4),
275-281.

Kurtz, M. E., Kurtz, J .C., Stommel, M., Given, C.W., & Given, BA. (1997). Loss of
physical functioning among geriatric cancer patients: Relationships to cancer site,
treatment, comorbidity and age. European Journal of Cancer, 33(14), 23 52-2358.

Kurtz, M. E., Kurtz, J .C., Stommel, M., Given, C.W., & Given, BA. (2000).
Symptomatology and loss of physical functioning among geriatric patients with lung
cancer. Journal of Pain and Symptom Management, 19(4), 249-256.

Lee, K. A., Hicks, G., & Nino-Murcia, G. (1991). Validity and reliability of a scale to
assess fatigue. Psychiatry Research, 36, 291-298.

148

 

Lenz, E. R., Pugh, L.C., Milligan, R.A., Gift, A.G., & Suppe, F. (1997). The middle-
range theory of unpleasant symptoms: An update. Advances in Nursing Science,
19(3), 14-27.

Lenz, E. R., Suppe, F., Gift, A.G., Pugh, L.C. & Milligan, RA. (1995). Collaborative
development of middle-range nursing theories: Toward a theory of unpleasant
symptoms. Advances in Nursing Science, 1 7(3), 1-13.

Leventhal, H. & Johnson, RE. (1983). Laboratory and field experimentation
development of theory of self regulation. In P.H. Woolridge, M.H. Schmitt, & J. K.
Skipper (Eds), Behavior Science and Nursing Theory (pp. 189-262). St. Louis, MO:
Mosby.

Lipschitz, D.A. (1995). Age-related declines in hematopoietic reserve capacity.
Seminars in Oncology, 22(suppl 1), 3-5.

Lough, M.E., Lindsey, A.M., Shinn, J .A., & Stotts, NA. (1987). Impact of symptom
frequency and symptom distress on self-reported quality of life in heart transplant
recipients. Heart and Lung, 16(3), 193-200.

Lyman, G.H., Dale, D.C., & Crawford, J. (2003). Incidence and predictors of low dose-
intensity in adjuvant breast cancer chemotherapy: A nationwide study of community
practices. Journal of Clinical Oncology, 21 (24), 4524-4531.

McCorkle, R., & Young, K. (1978). Development of a symptom distress scale. Cancer
Nursing, 1(5), 373-378.

Mendoza, T.R., Wang, X.S., Cleeland, C.S., Morrissey, M., Johnson, B.A.,Wendt, J .K.,
et al. (1999). The rapid assessment of fatigue severity in cancer patients: Use of the
brief fatigue inventory. Cancer, 85(5), 1186-1196.

Miaskowski, C., Dodd, M., West, C., Paul, S.M., Schumacher, K., Tripathy, D., et al. (in
press). The use of responder analysis to identify differences in patient outcomes
following a self-care intervention to improve cancer pain management. Pain

Miaskowski, C., & Lee, K. (1999). Pain, fatigue, and sleep disturbances in oncology
outpatients receiving radiation therapy for bone metastasis: A pilot study. Journal of
Pain and Symptom Management, 1 7(5), 320-332.

Montazeri, A., Gillis, C.R., & McEwen, J. (1998). Quality of life in patients with lung
cancer. A review of the literature fi'om 1970-1995. Chest, 113(2), 467—481.

Moynihan, C. (2002). Men, women, sex and cancer. European Journal of Cancer Care,
1(3), 166-172.

149

Nail, L. & Jones, L. (1995). Fatigue as a side effect of cancer treatment: Impact on
quality of life. Quality of Life: A Nursing Challenge, 4, 8-13.

National Cancer Institute (N C1). 2007 Cancer survivorship research: Cancer control and
populations. Estimated U.S. cancer prevalence. Retrieved February 20, 2007, from
http://dccps.nci.nih.gov/ocs/prevalence/prevalence.html#survivor.

Nieboer, P., Buijs, C., Rodenhuis, S, Seynaeve, C., Beex, L.V.A.M., van der Wall, E. et
al. (2005). Fatigue and relating factors in high-risk breast cancer patients treated with
adj uvant standard or high-dose chemotherapy: A longitudinal study. Journal of
Clinical Oncology, 23(33), 8296-8304.

Parker, K.P., Kimble, L.P., Dunbar, S.B., & Clark, RC. (2005). Symptom interactions
as mechanisms underlying symptom pairs and clusters. Journal of Nursing
Scholarship, 3 7(3), 209-215.

Paul, S.M., Zelman, D.C., Smith, M. & Miaskowski, C. (2005). Categorizing the
severity of cancer pain: Further exploration of the establishment of cut points.
Journal of Pain, 113, 37—44.

Ratel, S., Lazaar, N., Williams, C.A., Bedu, M., & Duche, P. (2003). Age differences in
human skeletal muscle fatigue during high-intensity intermittent exercise. Acta
Paediatrica, 92(11), 1248-1254.

Ream, E. & Richardson, A. (1997). Fatigue in patients with cancer and chronic
obstructive airways disease: A phenomenological study. International Journal of
Nursing Studies, 34(1), 44-53.

Rhodes, V. A., & Watson, RM. (1987). Symptom distress: The concept past and
present. Seminars in Oncology Nursing, 3(4), 242-247.

Richer, M.C., & Ezer, H. (2000). Understanding beliefs and meanings in the experience
of cancer: A concept analysis. Journal of Advanced Nursing, 32, 1108-1115.

Rumble, M.E., Keefe, F .J., Edinger, J .D., Porter L.S, & Garst, J .L. (2005). A pilot study
investigating the utility of the cognitive-behavioral model of insomnia in early-stage
lung cancer patients. Journal of Pain and Symptom Management, 30(2), 160-169.

Rutledge, D.M., & McGuire, C. (2004). Chapter 1: Evidence-based symptom
management. In C.H. Yarbro, M.H. Frogge, & M. Goodman (Eds), Cancer Symptom
Management (3” ed.) (pp. 3-14). Sudbury, MA: Jones & Bartlett.

Salive, M.E., Comoni-Huntly, J ., Guralnik, J .M., Phillips, C.L., Wallace, R.B., Ostfeld,
A.M., et al. (1992). Anemia and hemoglobin levels in older persons: Relationship
with age, gender and health status. Journal of the American Geriatrics Society, 40(5),
489-496.

150

Samet, J ., Hunt, W.C., Key, C., Humble, C.G., & Goodwin, J. (1986). Choice of cancer
therapy varies with age of patient. JAMA, 255(24), 3385-3390.

Santacroce, S.J., Maccarelli, L.M., & Grey, M. (2004). Intervention fidelity. Nursing
Research, 53(1), 63-66.

Sama, L. (1993). Correlates of symptom distress in women with lung cancer. Cancer
Practice, 1(1), 21-28.

Sama, L., & Brecht, M.L. (1997). Dimensions of symptom distress in women with
advanced lung cancer: A factor analysis. Heart and Lung, 26(1), 23-30.

Savard, J ., & Morin, C. (2001). Insomnia in the context of cancer: A review of a
neglected problem. Journal of Clinical Oncology, 19(3), 895-908.

Savard, J ., Sinard, S., Ivers, H., & Morin, CM. (2005). Randomized study on the
efficacy of cognitive-behavioral therapy for insomnia secondary to breast cancer, Part
H: Immunologic effects. Journal of Clinical Oncology, 23(25), 6097-6106.

Servaes, P., Verhagen, S., & Bleijenberg, G. (2002). Determinants of chronic fatigue in
disease-free breast cancer patients: a cross sectional study. Annals of Oncology, 13(4),
589-598.

Shaver, J. L. (1999). Managing sleep and waking behaviors and the symptom of fatigue.
In A.S. I-Iinshaw, S.L. Feetham, & J .L. Shaver J .L. (Eds), Handbook of Clinical
Nursing Research (pp. 435-456). Thousand Oaks, CA: Sage Publications.

Sikorskii, A., Given, C., Given, B., Jeon, S., Decker, V. & Decker, D., et al. (in press).
Symptom management for cancer patients; A trial comparing two multimodal
interventions. Journal of Pain and Symptom Management.

Simon, A. & Zutton, R. (1999). Fatigue in cancer patients. Current Opinion in Oncology,
11(4), 244-250.

Stein, K. D., Jacobsen, P.B., Blanchard, C.M., & Thors, C. (2004). Further validation of
the multidimensional fatigue symptom inventory-short form. Journal of Pain and
Symptom Management, 27(1), 14-22.

Stone, P., Richards, M. & Hardy, J. (1998). Fatigue in patients with cancer. European
Journal of Cancer, 34 (11), 1670-1676.

Taves, DR. (1974). Minimization: A new method of assigning patients to treatment and
control groups. Clinical Pharmacological Therapy, (15), 443-453.

Theander, K. & Unosson, M. (2004). Fatigue in patients with chronic obstructive
pulmonary disease. Journal of Advanced Nursing, 45(2), 172-177.

151

Tishelman, C., Degner, L.F., & Mueller, B. (2000). Measuring symptom distress in
patients with lung carcinoma: A pilot study of experienced intensity and importance
of symptoms. Cancer Nursing, 23(2), 82-90.

Tishelman, C., Degner, L.F., Rudman, A., Bertilsson, K., Bond, R., Broberger, E. et al.
(2005). Symptoms in patients with lung carcinoma: Distinguishing distress from
intensity. Cancer, 104(9), 2013-2021.

Trendall, J. (2000.). Concept analysis: Chronic fatigue. Journal of Advanced Nursing,
32(5), 1126-1131.

Vainio, A. & Auvinen, A. (1996). Prevalence of symptoms among patients with
advanced cancer: An international collaborative study. Journal of Pain and
Symptom Management, 12(1), 3-10.

Vogelzang, N. J ., Breitbart, W., Cella, D., Curt, G.A., Groopman, J.E., Homing, S.J., et
al. (1997). The fatigue coalition: Patient, caregiver, and oncologist perceptions of
cancer-related fatigue: Results of a Tripart Assessment Survey. Seminars in
Hematology, 34(suppl 2), 4-12.

Walker, L. O. Avant, KC. (1995). Strategies for Theory Construction in Nursing.
Norwalk, CT, Appleton & Lange.

Wang, X.S., Giralt, S.A., Mendoza, T.R., Engstram, M.C., Johnson, B.A., Peterson, N., et
al. (2002). Clinical factors associated with cancer-related fatigue in patients being
treated for leukemia and non-Hodgkin’s lymphoma. Journal of Clinical Oncology,
20(5), 1319-1328.

Wilrnoth, M.C., Coleman, B.A., Smith, S.C., & Davis, C. (2004). Fatigue, weight gain,
and altered sexuality in patients with breast cancer: Exploration of a symptom cluster.
Oncology Nursing Forum, 31(6), 1069-1075.

Wyatt, G.K., Friedman, L.L., Given, C.W., Given, B.A., & Beckrow, KC. (1999).
Complementary therapy use among older cancer patients. Cancer Practice, 7(3), 136-
144.

Yancik, R. & Ries, L.A.G. (2000). Aging and cancer in America: Demographic and
epidemiologic perspectives. In L. Balducci & M. Extermann (Eds),
Hematology/Oncology Clinics of North America: Cancer in the Elderly Current
Concepts and Future Directions (pp. 1 7-24). Philadelphia: Saunders

Yates, R, Aranda, S., Hargraves, M, Mirolo, B., Clavarino, A., McLachlan, S., et al.
(2005). Randomized controlled trial of an educational intervention for managing

fatigue in women receiving adj uvant chemotherapy for early-stage breast cancer.
Journal of Clinical Oncology, 23(25), 6027-6036.

152

Yellen, S.B., Cella, D. Webster, K., Blendowski, C., & Kaplan, E. (1997). Measuring
fatigue and other anemia-related symptoms with the Functional Assessment of Cancer

Therapy (FACT) measurement system. Journal of Pain and Symptom Management,
13(2), 63-74.

Zelenetz, A.D., Reider, P.A., Delgado, DJ. (2000). Review of patterns of care among
community physicians in intermediate grade NHL (IGL) reveals significantly greater
planned and delivered dose attenuation in older patients [ASH Annual Meeting
Abstract]. Blood, 96(11), 588.

153

 

MIC

3

WT

  

  

Ill

9

 

  
  

S

If!

[tum

02

     

  

    

UNI

    

  
      

Minn"

7905

III '''''